Photo credit: © Jerry Markatos
ABOUT THE AUTHOR
Frederick P. Brooks, Jr., is Kenan Professor of Computer Science
at the University of North Carolina at Chapel Hill. He is best
known as the "father of the IBM System/360," having served as
project manager for its development and later as manager of the
Operating System/360 software project during its design phase.
For this work he, Bob Evans, and Erich Bloch were awarded the
National Medal of Technology in 1985. Earlier, he was an archi-
tect of the IBM Stretch and Harvest computers.
At Chapel Hill, Dr. Brooks founded the Department of Com-
puter Science and chaired it from 1964 through 1984. He has
served on the National Science Board and the Defense Science
Board. His current teaching and research is in computer archi-
tecture, molecular graphics, and virtual environments.
The Mythical Man-Month
Essays on Software Engineering
Anniversary Edition
Frederick P. Brooks, Jr.
University of North Carolina at Chapel Hill
ADDISON-WESLEY
Boston • San Francisco * New York « Toronto « Montreal
London « Munich * Paris e Madrid
Capetown • Sydney • Tokyo • Singapore • Mexico City
Cover drawing: C. R. Knight, Mural of the La Brea Tar Pits. Courtesy
of the George C. Page Museum of La Brea Discoveries, The Natural
History Museum of Los Angeles County. Cover designer: Diana Coe.
The essay entitled, No Silver Bullet, is from Information Processing
1986, the Proceedings of the IFIP Tenth World Computing Conference,
edited by H.-J. Kugler, 1986, pages 1069-1076. Reprinted with the kind
permission of IFIP and Elsevier Science B.V., Amsterdam, The Neth-
erlands.
Library of Congress Cataloging-in-Publication Data
Brooks, Frederick P., Jr. (Frederick Phillips)
The mythical man-month : essays on software engineering /
Frederick P. Brooks, Jr. — Anniversary ed.
p. cm.
Includes bibliographical references and index.
ISBN 0-201-83595-9
1. Software engineering. I. Title.
QA76.758.B75 1995
005.1'068—dc20 94-36653
CIP
Many of the designations used by manufacturers and sellers to distin-
guish their products are claimed as trademarks. Where those designa-
tions appear in this book, and Addison-Wesley was aware of a
trademark claim, the designations have been printed in initial caps or
all caps.
Copyright © 1995 Addison Wesley Longman, Inc.
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted in any form or by any means,
electronic, mechanical, photocopying, recording, or otherwise, without
prior written permission of the publisher and author. Printed in the
United States of America.
Text printed on recycled and acid-free paper.
ISBN 0201835959
17 1819202122 MA 05 04 03 02
17th Printing August 2002
Dedication of the 1975 edition
To two who especially enriched my IBM years:
Thomas /. Watson, Jr.,
whose deep concern for people still permeates his company,
and
Bob O. Evans,
whose bold leadership turned work into
Dedication of the 1995 edition
To Nancy,
God's gift to me.
Preface to the 20th
Anniversary Edition
To my surprise and delight, The Mythical Man-Month continues
to be popular after 20 years. Over 250,000 copies are in print.
People often ask which of the opinions and recommendations
set forth in 1975 I still hold, and which have changed, and how.
Whereas I have from time to time addressed that question in lec-
tures, I have long wanted to essay it in writing.
Peter Gordon, now a Publishing Partner at Addison-Wesley,
has been working with me patiently and helpfully since 1980.
He proposed that we prepare an Anniversary Edition. We de-
cided not to revise the original, but to reprint it untouched (ex-
cept for trivial corrections) and to augment it with more current
thoughts.
Chapter 16 reprints "No Silver Bullet: Essence and Acci-
dents of Software Engineering," a 1986 IFIPS paper that grew
out of my experience chairing a Defense Science Board study on
military software. My coauthors of that study, and our executive
secretary, Robert L. Patrick, were invaluable in bringing me
back into touch with real-world large software projects. The pa-
per was reprinted in 1987 in the IEEE Computer magazine, which
gave it wide circulation.
"No Silver Bullet" proved provocative. It predicted that a
decade would not see any programming technique that would
by itself bring an order-of-magnitude improvement in software
productivity. The decade has a year to run; my prediction seems
safe. "NSB" has stimulated more and more spirited discussion
Vll
viii Preface to the 20th Anniversary Edition
in the literature than has The Mythical Man-Month. Chapter 17,
therefore, comments on some of the published critique and up-
dates the opinions set forth in 1986.
In preparing my retrospective and update of The Mythical
Man-Month, I was struck by how few of the propositions as-
serted in it have been critiqued, proven, or disproven by on-
going software engineering research and experience. It proved
useful to me now to catalog those propositions in raw form,
stripped of supporting arguments and data. In hopes that these
bald statements will invite arguments and facts to prove, dis-
prove, update, or refine those propositions, I have included this
outline as Chapter 18.
Chapter 19 is the updating essay itself. The reader should
be warned that the new opinions are not nearly so well in-
formed by experience in the trenches as the original book was.
I have been at work in a university, not industry, and on small-
scale projects, not large ones. Since 1986, I have only taught
software engineering, not done research in it at all. My research
has rather been on virtual environments and their applications.
In preparing this retrospective, I have sought the current
views of friends who are indeed at work in software engineer-
ing. For a wonderful willingness to share views, to comment
thoughtfully on drafts, and to re-educate me, I am indebted to
Barry Boehm, Ken Brooks, Dick Case, James Coggins, Tom
DeMarco, Jim McCarthy, David Parnas, Earl Wheeler, and Ed-
ward Yourdon. Fay Ward has superbly handled the technical
production of the new chapters.
I thank Gordon Bell, Bruce Buchanan, Rick Hayes-Roth, my
colleagues on the Defense Science Board Task Force on Military
Software, and, most especially, David Parnas for their insights
and stimulating ideas for, and Rebekah Bierly for technical pro-
duction of, the paper printed here as Chapter 16. Analyzing the
software problem into the categories of essence and accident was
inspired by Nancy Greenwood Brooks, who used such analysis
in a paper on Suzuki violin pedagogy.
Preface to the 20th Anniversary Edition ix
Addison-Wesley's house custom did not permit me to ac-
knowledge in the preface to the 1975 edition the key roles
played by their staff. Two persons' contributions should be es-
pecially cited: Norman Stanton, then Executive Editor, and Her-
bert Boes, then Art Director. Boes developed the elegant style,
which one reviewer especially cited: "wide margins, [and] imag-
inative use of typeface and layout." More important, he also
made the crucial recommendation that every chapter have an
opening picture. (I had only the Tar Pit and Reims Cathedral at
the time.) Finding the pictures occasioned an extra year's work
for me, but I am eternally grateful for the counsel.
Soli Deo gloria—To God alone be glory.
Chapel Hill, N.C.. F. P. B., Jr.
March 1995
Preface to the
First Edition
In many ways, managing a large computer programming proj-
ect is like managing any other large undertaking—in more ways
than most programmers believe. But in many other ways
it is different—in more ways than most professional managers
expect.
The lore of the field is accumulating. There have been sev-
eral conferences, sessions at AFIPS conferences, some books,
and papers. But it is by no means yet in shape for any systematic
textbook treatment. It seems appropriate, however, to offer this
little book, reflecting essentially a personal view.
Although I originally grew up in the programming side of
computer science, I was involved chiefly in hardware architec-
ture during the years (1956-1963) that the autonomous control
program and the high-level language compiler were developed.
When in 1964 I became manager of Operating System/360, I
found a programming world quite changed by the progress of
the previous few years.
Managing OS/360 development was a very educational ex-
perience, albeit a very frustrating one. The team, including F. M.
Trapnell who succeeded me as manager, has much to be proud
of. The system contains many excellencies in design and exe-
cution, and it has been successful in achieving widespread use.
Certain ideas, most noticeably device-independent input-output
and external library management, were technical innovations
Preface to the First Edition, xi
now widely copied. It is now quite reliable, reasonably efficient,
and very versatile.
The effort cannot be called wholly successful, however. Any
OS/360 user is quickly aware of how much better it should be.
The flaws in design and execution pervade especially the control
program, as distinguished from the language compilers. Most of
these flaws date from the 1964-65 design period and hence must
be laid to my charge. Furthermore, the product was late, it took
more memory than planned, the costs were several times the
estimate, and it did not perform very well until several releases
after the first.
After leaving IBM in 1965 to come to Chapel Hill as origi-
nally agreed when I took over OS/360, I began to analyze the
OS/360 experience to see what management and technical les-
sons were to be learned. In particular, I wanted to explain the
quite different management experiences encountered in System/
360 hardware development and OS/360 software development.
This book is a belated answer to Tom Watson's probing ques-
tions as to why programming is hard to manage.
In this quest I have profited from long conversations with
R. P. Case, assistant manager 1964-65, and F. M, Trapnell, man-
ager 1965-68.1 have compared conclusions with other managers
of jumbo programming projects, including F. J. Corbato of
M.I.T., John Harr and V. Vyssotsky of Bell Telephone Labora-
tories, Charles Portman of International Computers Limited,
A. P. Ershov of the Computation Laboratory of the Siberian Di-
vision, U.S.S.R. Academy of Sciences, and A. M. Pietrasanta of
IBM.
My own conclusions are embodied in the essays that follow,
which are intended for professional programmers, professional
managers, and especially professional managers of program-
mers.
Although written as separable essays, there is a central ar-
gument contained especially in Chapters 2-7. Briefly, I believe
that large programming projects suffer management problems
xii Preface to the First Edition
different in kind from small ones, due to division of labor. I be-
lieve the critical need to be the preservation of the conceptual
integrity of the product itself. These chapters explore both the
difficulties of achieving this unity and methods for doing so.
The later chapters explore other aspects of software engineering
management.
The literature in this field is not abundant, but it is widely
scattered. Hence I have tried to give references that will both
illuminate particular points and guide the interested reader to
other useful works. Many friends have read the manuscript,
and some have prepared extensive helpful comments; where
these seemed valuable but did not fit the flow of the text, I have
included them in the notes.
Because this is a book of essays and not a text, all the ref-
erences and notes have been banished to the end of the volume,
and the reader is urged to ignore them on his first reading.
I am deeply indebted to Miss Sara Elizabeth Moore, Mr.
David Wagner, and Mrs. Rebecca Burns for their help in pre-
paring the manuscript, and to Professor Joseph C. Sloane for ad-
vice on illustration.
Chapel Hill, N.C. F. P. B., Jr
October 1974
Contents
Preface to the 20th Anniversary Edition vii
Preface to the First Edition .. x
Chapter 1 The Tar Pit 3
Chapter 2 The Mythical Man-Month 13
Chapter 3 The Surgical Team 29
Chapter 4 Aristocracy, Democracy, and System Design 41
Chapter 5 The Second-System Effect 53
Chapter 6 Passing the Word 61
Chapter 7 Why Did the Tower of Babel Fail? ,. 73
Chapter 8 Calling the Shot 87
Chapter 9 Ten Pounds in a Five-Pound Sack 97
Chapter 10 The Documentary Hypothesis 107
Chapter 11 Plan to Throw One Away 115
Chapter 12 Sharp Tools ,, 127
Chapter 13 The Whole and the Parts 141
Chapter 14 Hatching a Catastrophe , 153
Chapter 15 The Other Face 163
Chapter 16 No Silver Bullet—Essence and Accident 177
Chapter 17 "No Silver Bullet" Refired 205
Chapter 18 Propositions of The Mythical Man-Month:
True or False? 227
Chapter 19 The Mythical Man-Month after 20 Years - 251
Epilogue 291
Notes and References 293
Index 309
Xlll
1
The Tar Pit
1
TheTarPit
Een schip op het strand is een baken in zee.
{A ship on the beach is a lighthouse to the sea. ]
DUTCH PROVERB
C R. Knight, Mural of La Brea Tar Pits
The George C. Page Museum of La Brea Discoveries,
The Natural History Museum of Los Angeles County
3
4 The Tar Pit
No scene from prehistory is quite so vivid as that of the mortal
struggles of great beasts in the tar pits. In the mind's eye one sees
dinosaurs, mammoths, and sabertoothed tigers struggling against
the grip of the tar. The fiercer the struggle, the more entangling the
tar, and no beast is so strong or so skillful but that he ultimately
sinks.
Large-system programming has over the past decade been
such a tar pit, and many great and powerful beasts have thrashed
violently in it. Most have emerged with running systems—few
have met goals, schedules, and budgets. Large and small, massive
or wiry, team after team has become entangled in the tar. No one
thing seems to cause the difficulty—any particular paw can be
pulled away. But the accumulation of simultaneous and interact-
ing factors brings slower and slower motion. Everyone seems to
have been surprised by the stickiness of the problem, and it is hard
to discern the nature of it. But we must try to understand it if we
are to solve it.
Therefore let us begin by identifying the craft of system pro-
gramming and the joys and woes inherent in it.
The Programming Systems Product
One occasionally reads newspaper accounts of how two program-
mers in a remodeled garage have built an important program that
surpasses the best efforts of large teams. And every programmer
is prepared to believe such tales, for he knows that he could build
any program much faster than the 1000 statements/year reported
for industrial teams.
Why then have not all industrial programming teams been
replaced by dedicated garage duos? One must look at what is being
produced.
In the upper left of Fig. 1.1 is a program. It is complete in itself,
ready to be run by the author on the system on which it was
developed. That is the thing commonly produced in garages, and
The Programming Systems Product 5
Fig. 1.1 Evolution of the programming systems product
that is the object the individual programmer uses in estimating
productivity.
There are two ways a program can be converted into a more
useful, but more costly, object. These two ways are represented by
the boundaries in the diagram.
Moving down across the horizontal boundary, a program
becomes a programming product. This is a program that can be run,
6 The Tar Pit
tested, repaired, and extended by anybody. It is usable in many
operating environments, for many sets of data. To become a gener-
ally usable programming product, a program must be written in a
generalized fashion. In particular the range and form of inputs
must be generalized as much as the basic algorithm will reasonably
allow. Then the program must be thoroughly tested, so that it can
be depended upon. This means that a substantial bank of test
cases, exploring the input range and probing its boundaries, must
be prepared, run, and recorded. Finally, promotion of a program
to a programming product requires its thorough documentation, so
that anyone may use it, fix it, and extend it. As a rule of thumb,
I estimate that a programming product costs at least three times as
much as a debugged program with the same function.
Moving across the vertical boundary, a program becomes a
component in a programming system. This is a collection of interact-
ing programs, coordinated in function and disciplined in format,
so that the assemblage constitutes an entire facility for large tasks.
To become a programming system component, a program must be
written so that every input and output conforms in syntax and
semantics with precisely defined interfaces. The program must
also be designed so that it uses only a prescribed budget of re-
sources—memory space, input-output devices, computer time. Fi-
nally, the program must be tested with other system components,
in all expected combinations. This testing must be extensive, for
the number of cases grows combinatorially. It is time-consuming,
for subtle bugs arise from unexpected interactions of debugged
components. A programming system component costs at least
three times as much as a stand-alone program of the same func-
tion. The cost may be greater if the system has many components.
In the lower right-hand corner of Fig. 1.1 stands the program-
ming systems product. This differs from the simple program in all of
the above ways. It costs nine times as much. But it is the truly
useful object, the intended product of most system programming
efforts.
The Joys of the Craft 7
The Joys of the Craft
Why is programming fun? What delights may its practitioner
expect as his reward?
First is the sheer joy of making things. As the child delights
in his mud pie, so the adult enjoys building things, especially
things of his own design. I think this delight must be an image of
God's delight in making things, a delight shown in the distinctness
and newness of each leaf and each snowflake.
Second is the pleasure of making things that are useful to
other people. Deep within, we want others to use our work and
to find it helpful. In this respect the programming system is not
essentially different from the child's first clay pencil holder "for
Daddy's office."
Third is the fascination of fashioning complex puzzle-like
objects of interlocking moving parts and watching them work in
subtle cycles, playing out the consequences of principles built in
from the beginning. The programmed computer has all the fasci-
nation of the pinball machine or the jukebox mechanism, carried
to the ultimate.
Fourth is the joy of always learning, which springs from the
nonrepeating nature of the task. In one way or another the prob-
lem is ever new, and its solver learns something: sometimes practi-
cal, sometimes theoretical, and sometimes both.
Finally, there is the delight of working in such a tractable
medium. The programmer, like the poet, works only slightly re-
moved from pure thought-stuff. He builds his castles in the air,
from air, creating by exertion of the imagination. Few media of
creation are so flexible, so easy to polish and rework, so readily
capable of realizing grand conceptual structures. (As we shall see
later, this very tractability has its own problems.)
Yet the program construct, unlike the poet's words, is real in
the sense that it moves and works, producing visible outputs sepa-
rate from the construct itself. It prints results, draws pictures,
produces sounds, moves arms. The magic of myth and legend has
8 The Tar Pit
come true in our time. One types the correct incantation on a
keyboard, and a display screen comes to life, showing things that
never were nor could be.
Programming then is fun because it gratifies creative longings
built deep within us and delights sensibilities we have in common
with all men.
The Woes of the Craft
Not all is delight, however, and knowing the inherent woes makes
it easier to bear them when they appear.
First, one must perform perfectly. The computer resembles the
magic of legend in this respect, too. If one character, one pause, of
the incantation is not strictly in proper form, the magic doesn't
work. Human beings are not accustomed to being perfect, and few
areas of human activity demand it. Adjusting to the requirement
for perfection is, I think, the most difficult part of learning to
program.1
Next, other people set one's objectives, provide one's re-
sources, and furnish one's information. One rarely controls the
circumstances of his work, or even its goal. In management terms,
one's authority is not sufficient for his responsibility. It seems that
in all fields, however, the jobs where things get done never have
formal authority commensurate with responsibility. In practice,
actual (as opposed to formal) authority is acquired from the very
momentum of accomplishment.
The dependence upon others has a particular case that is espe-
cially painful for the system programmer. He depends upon other
people's programs. These are often maldesigned, poorly imple-
mented, incompletely delivered (no source code or test cases), and
poorly documented. So he must spend hours studying and fixing
things that in an ideal world would be complete, available, and
usable.
The next woe is that designing grand concepts is fun; finding
nitty little bugs is just work. With any creative activity come
The Woes of the Craft 9
dreary hours of tedious, painstaking labor, and programming is no
exception.
Next, one finds that debugging has a linear convergence, or
worse, where one somehow expects a quadratic sort of approach
to the end. So testing drags on and on, the last difficult bugs taking
more time to find than the first.
The last woe, and sometimes the last straw, is that the product
over which one has labored so long appears to be obsolete upon
(or before) completion. Already colleagues and competitors are in
hot pursuit of new and better ideas. Already the displacement of
one's thought-child is not only conceived, but scheduled.
This always seems worse than it really is. The new and better
product is generally not available when one completes his own; it
is only talked about. It, too, will require months of development.
The real tiger is never a match for the paper one, unless actual use
is wanted. Then the virtues of reality have a satisfaction all their
own.
Of course the technological base on which one builds is always
advancing. As soon as one freezes a design, it becomes obsolete in
terms of its concepts. But implementation of real products de-
mands phasing and quantizing. The obsolescence of an implemen-
tation must be measured against other existing implementations,
not against unrealized concepts. The challenge and the mission are
to find real solutions to real problems on actual schedules with
available resources.
This then is programming, both a tar pit in which many efforts
have floundered and a creative activity with joys and woes all its
own. For many, the joys far outweigh the woes, and for them the
remainder of this book will attempt to lay some boardwalks across
the tar.
2
The Mythical Man-Month
2
The Mythical Man-Month
Good cooking takes time. If you are made to wait, it is to
serve you better, and to please you.
MENU OF RESTAURANT ANTOINE. NEW ORLEANS
13
14 The Mythical Man-Month
More software projects have gone awry for lack of calendar time
than for all other causes combined. Why is this cause of disaster
so common?
First, our techniques of estimating are poorly developed. More
seriously, they reflect an unvoiced assumption which is quite un-
true, i.e., that all will go well.
Second, our estimating techniques fallaciously confuse effort
with progress, hiding the assumption that men and months are
interchangeable.
Third, because we are uncertain of our estimates, software
managers often lack the courteous stubbornness of Antoine's chef.
Fourth, schedule progress is poorly monitored. Techniques
proven and routine in other engineering disciplines are considered
radical innovations in software engineering.
Fifth, when schedule slippage is recognized, the natural (and
traditional) response is to add manpower. Like dousing a fire with
gasoline, this makes matters worse, much worse. More fire re-
quires more gasoline, and thus begins a regenerative cycle which
ends in disaster.
Schedule monitoring will be the subject of a separate essay.
Let us consider other aspects of the problem in more detail.
Optimism
All programmers are optimists. Perhaps this modern sorcery espe-
cially attracts those who believe in happy endings and fairy god-
mothers. Perhaps the hundreds of nitty frustrations drive away all
but those who habitually focus on the end goal. Perhaps it is
merely that computers are young, programmers are younger, and
the young are always optimists. But however the selection process
works, the result is indisputable: "This time it will surely run," or
"I just found the last bug."
So the first false assumption that underlies the scheduling of
systems programming is that all will go well, i.e., that each task will
hike only as long as it "ought" to take.
Optimism 15
The pervasiveness of optimism among programmers deserves
more than a flip analysis. Dorothy Sayers, in her excellent book,
The Mind of the Maker, divides creative activity into three stages:
the idea, the implementation, and the interaction. A book, then,
or a computer, or a program comes into existence first as an ideal
construct, built outside time and space, but complete in the mind
of the author. It is realized in time and space, by pen, ink, and
paper, or by wire, silicon, and ferrite. The creation is complete
when someone reads the book, uses the computer, or runs the
program, thereby interacting with the mind of the maker.
This description, which Miss Sayers uses to illuminate not
only human creative activity but also the Christian doctrine of the
Trinity, will help us in our present task. For the human makers of
things, the incompletenesses and inconsistencies of our ideas
become clear only during implementation. Thus it is that writing,
experimentation, "working out" are essential disciplines for the
theoretician.
In many creative activities the medium of execution is intract-
able. Lumber splits; paints smear; electrical circuits ring. These
physical limitations of the medium constrain the ideas that may
be expressed, and they also create unexpected difficulties in the
implementation.
Implementation, then, takes time and sweat both because of
the physical media and because of the inadequacies of the under-
lying ideas. We tend to blame the physical media for most of our
implementation difficulties; for the media are not "ours" in the
way the ideas are, and our pride colors our judgment.
Computer programming, however, creates with an exceed-
ingly tractable medium. The programmer builds from pure
thought-stuff: concepts and very flexible representations thereof.
Because the medium is tractable, we expect few difficulties in
implementation; hence our pervasive optimism. Because our ideas
are faulty, we have bugs; hence our optimism is unjustified.
In a single task, the assumption that all will go well has a
probabilistic effect on the schedule. It might indeed go as
16 The Mythical Man-Month
for there is a probability distribution for the delay that will be
encountered, and "no delay" has a finite probability. A large pro-
gramming effort, however, consists of many tasks, some chained
end-to-end. The probability that each will go well becomes van-
ishingly small.
The'Man-Month
The second fallacious thought mode is expressed in the very unit
of effort used in estimating and scheduling: the man-month. Cost
does indeed vary as the product of the number of men and the
number of months. Progress does not. Hence the man-month as a unit
for measuring the size of a job is a dangerous and deceptive myth. It
implies that men and months are interchangeable.
Men and months are interchangeable commodities only when
a task can be partitioned among many workers with no communica-
tion among them (Fig. 2.1). This is true of reaping wheat or picking
cotton; it is not even approximately true of systems programming.
Men
Fig. 2.1 Time versus number of workers—perfectly partitionable task
The Man-Month 17
When a task cannot be partitioned because of sequential con-
straints, the application of more effort has no effect on the sched-
ule (Fig. 2.2). The bearing of a child takes nine months, no matter
how many women are assigned. Many software tasks have this
characteristic because of the sequential nature of debugging.
Fig. 2.2 Time versus number of workers—unpartitionable task
In tasks that can be partitioned but which require communica-
tion among the subtasks, the effort of communication must be
added to the amount of work to be done. Therefore the best that
can be done is somewhat poorer than an even trade of men for
months (Fig. 2.3).
18 The Mythical Man-Month
Men
Fig. 2.3 Time versus number of workers—partitionable task requiring
communication
The added burden of communication is made up of two parts,
training and intercommunication. Each worker must be trained in
the technology, the goals of the effort, the overall strategy, and the
plan of work. This training cannot be partitioned, so this part of
the added effort varies linearly with the number of workers.1
Intercommunication is worse. If each part of the task must be
separately coordinated with each other part/ the effort increases as
n(n-I)/2. Three workers require three times as much pairwise
intercommunication as two; four require six times as much as two.
If, moreover, there need to be conferences among three, four, etc.,
workers to resolve things jointly, matters get worse yet. The added
effort of communicating may fully counteract the division of the
original task and bring us to the situation of Fig. 2.4.
Systems Test 19
Men
Fig. 2.4 Time versus number of workers—task with complex interrela-
tionships
Since software construction is inherently a systems effort—an
exercise in complex interrelationships—communication effort is
great, and it quickly dominates the decrease in individual task time
brought about by partitioning. Adding more men then lengthens,
not shortens, the schedule.
Systems Test
No parts of the schedule are so thoroughly affected by sequential
constraints as component debugging and system test. Further-
more, the time required depends on the number and subtlety of
the errors encountered. Theoretically this number should be zero.
Because of optimism, we usually expect the number of bugs to be
20 The Mythical Man-Month
smaller than it turns out to be. Therefore testing is usually the
most mis-scheduled part of programming.
For some years I have been successfully using the following
rule of thumb for scheduling a software task:
l
/3 planning
l
/6 coding
l
/4 component test and early system test
l
/4 system test, all components in hand.
This differs from conventional scheduling in several important
ways:
1. The fraction devoted to planning is larger than normal. Even
so, it is barely enough to produce a detailed and solid specifi-
cation, and not enough to include research or exploration of
totally new techniques.
2. The half of the schedule devoted to debugging of completed
code is much larger than normal.
3. The part that is easy to estimate, i.e., coding, is given only
one-sixth of the schedule.
In examining conventionally scheduled projects, I have found
that few allowed one-half of the projected schedule for testing,
but that most did indeed spend half of the actual schedule for that
purpose. Many of these were on schedule until and except in
system testing.2
Failure to allow enough time for system test, in particular, is
peculiarly disastrous. Since the delay comes at the end of the
schedule, no one is aware of schedule trouble until almost the
delivery date. Bad news, late and without warning, is unsettling
to customers and to managers.
Furthermore, delay at this point has unusually severe finan-
cial, as well as psychological, repercussions. The project is fully
staffed, and cost-per-day is maximum. More seriously, the soft-
ware is to support other business effort (shipping of computers,
operation of new facilities, etc.) and the secondary costs of delay-
ing these are very high, for it is almost time for software shipment.
Regenerative Schedule Disaster 21
Indeed, these secondary costs may far outweigh all others. It is
therefore very important to allow enough system test time in the
original schedule.
Gutless Estimating
Observe that for the programmer, as for the chef, the urgency of
the patron may govern the scheduled completion of the task, but
it cannot govern the actual completion. An omelette, promised in
two minutes, may appear to be progressing nicely. But when it has
not set in two minutes, the customer has two choices—wait or eat
it raw. Software customers have had the same choices.
The cook has another choice; he can turn up the heat. The
result is often an omelette nothing can save—burned in one part,
raw in another.
Now I do not think software managers have less inherent
courage and firmness than chefs, nor than other engineering man-
agers. But false scheduling to match the patron's desired date is
much more common in our discipline than elsewhere in engineer-
ing. It is very difficult to make a vigorous, plausible, and job-
risking defense of an estimate that is derived by no quantitative
method, supported by little data, and certified chiefly by the
hunches of the managers.
Clearly two solutions are needed. We need to develop and
publicize productivity figures, bug-incidence figures, estimating
rules, and so on. The whole prof ession can only profit from sharing
such data.
Until estimating is on a sounder basis, individual managers
will need to stiffen their backbones and defend their estimates
with the assurance that their poor hunches are better than wish-
derived estimates.
Regenerative Schedule Disaster
What does one do when an essential software project is behind
schedule? Add manpower, naturally. As Figs. 2.1 through 2.4 sug-
gest, this may or may not help.
22 The Mythical Man-Month
Let us consider an example.3 Suppose a task is estimated at 12
man-months and assigned to three men for four months, and that
there are measurable mileposts A, B, C, D, which are scheduled to
fall at the end of each month (Fig. 2.5).
Now suppose the first milepost is not reached until two
months have elapsed (Fig. 2.6). What are the alternatives facing
the manager?
1. Assume that the task must be done on time. Assume that only
the first part of the task was misestimated, so Fig. 2.6 tells the
story accurately. Then 9 man-months of effort remain, and
two months, so 4V£ men will be needed. Add 2 men to the 3
assigned.
2. Assume that the task must be done on time. Assume that the
whole estimate was uniformly low, so that Fig. 2.7 really
describes the situation. Then 18 man-months of effort remain,
and two months, so 9 men will be needed. Add 6 men to the
3 assigned.
Figure 2.5
Regenerative Schedule Disaster 23
Figure 2,6
Figure 2.7
24 The Mythical Man-Month
3. Reschedule. I like the advice given by P. Fagg, an experienced
hardware engineer, "Take no small slips." That is, allow
enough time in the new schedule to ensure that the work can
be carefully and thoroughly done, and that rescheduling will
not have to be done again.
4. Trim the task. In practice this tends to happen anyway, once
the team observes schedule slippage. Where the secondary
costs of delay are very high, this is the only feasible action.
The manager's only alternatives are to trim it formally and
carefully, to reschedule, or to watch the task get silently
trimmed by hasty design and incomplete testing.
In the first two cases, insisting that the unaltered task be
completed in four months is disastrous. Consider the regenerative
effects, for example, for the first alternative (Fig. 2.8). The two new
men, however competent and however quickly recruited, will re-
quire training in the task by one of the experienced men. If this
takes a month, 3 man-months will have been devoted to work not in the
original estimate. Furthermore, the task, originally partitioned three
ways, must be repartitioned into five parts; hence some work
already done will be lost, and system testing must be lengthened.
So at the end of the third month, substantially more than 7 man-
months of effort remain, and 5 trained people and one month are
available. As Fig. 2.8 suggests, the product is just as late as if no
one had been added (Fig. 2.6).
To hope to get done in four months, considering only training
time and not repartitioning and extra systems test, would require
adding 4 men, not 2, at the end of the second month. To cover
repartitioning and system test effects, one would have to add still
other men. Now, however, one has at least a 7-man team, not a
3-man one; thus such aspects as team organization and task divi-
sion are different in kind, not merely in degree.
Notice that by the end of the third month things look very
black. The March 1 milestone has not been reached in spite of all
Regenerative Schedule Disaster 25
Figure 2.8
the managerial effort. The temptation is very strong to repeat the
cycle, adding yet more manpower. Therein lies madness.
The foregoing assumed that only the first milestone was
misestimated. If on March I one makes the conservative assump-
tion that the whole schedule was optimistic, as Fig. 2.7 depicts, one
wants to add 6 men just to the original task. Calculation of the
training, repartitioning, system testing effects is left as an exercise
for the reader. Without a doubt, the regenerative disaster will
yield a poorer product, later, than would rescheduling with the
original three men, unaugmented.
Oversimplifying outrageously, we state Brooks's Law:
Adding manpower to a late software project makes it later.
This then is the demythologizing of the man-month. The
number of months of a project depends upon its sequential con-
26 The Mythical Man-Month
straints. The maximum number of men depends upon the number
of independent subtasks. From these two quantities one can derive
schedules using fewer men and more months. (The only risk is
product obsolescence.) One cannot, however, get workable sched-
ules using more men and fewer months. More software projects
have gone awry for lack of calendar time than for all other causes
combined.
3
The Surgical Team
3
The Surgical Team
These studies revealed large individual differences between
high and low performers, often by an order of magnitude.
SACKMAN. ERIKSON. AND GRANT1
UPI Photo/The Bettman Archive
29
30 The Surgical Team
At computer society meetings one continually hears young pro-
gramming managers assert that they favor a small, sharp team of
first-class people, rather than a project with hundreds of program-
mers, and those by implication mediocre. So do we all.
But this naive statement of the alternatives avoids the hard
problem—how does one build large systems on a meaningful
schedule? Let us look at each side of this question in more detail.
The Problem
Programming managers have long recognized wide productivity
variations between good programmers and poor ones. But the
actual measured magnitudes have astounded all of us. In one of
their studies, Sackman, Erikson, and Grant were measuring perfor-
mances of a group of experienced programmers. Within just this
group the ratios between best and worst performances averaged
about 10:1 on productivity measurements and an amazing 5:1 on
program speed and space measurements! In short the $20,000/year
programmer may well be 10 times as productive as the
$10,000/year one. The converse may be true, too. The data
showed no correlation whatsoever between experience and per-
formance. (I doubt if that is universally true.)
I have earlier argued that the sheer number of minds to be
coordinated affects the cost of the effort, for a major part of the
cost is communication and correcting the ill effects of miscom-
munication (system debugging). This, too, suggests that one wants
the system to be built by as few minds as possible. Indeed, most
experience with large programming systems shows that the brute-
force approach is costly, slow, inefficient, and produces systems
that are not conceptually integrated. OS/360, Exec 8, Scope 6600,
Multics, TSS, SAGE, etc.—the list goes on and on.
The conclusion is simple: if a 200-man project has 25 manag-
ers who are the most competent and experienced programmers,
fire the 175 troops and put the managers back to programming.
The Problem 31
Now let's examine this solution. On the one hand, it fails to
approach the ideal of the small sharp team, which by common
consensus shouldn't exceed 10 people. It is so large that it will need
to have at least two levels of management, or about five managers.
It will additionally need support in finance, personnel, space, sec-
retaries, and machine operators.
On the other hand, the original 200-man team was not large
enough to build the really large systems by brute-force methods.
Consider OS/360, for example. At the peak over 1000 people were
working on it—programmers, writers, machine operators, clerks,
secretaries, managers, support groups, and so on. From 1963
through 1966 probably 5000 man-years went into its design, con-
struction, and documentation. Our postulated 200-man team
would have taken 25 years to have brought the product to its
present stage, if men and months traded evenly!
This then is the problem with the small, sharp team concept:
it is too slow for really big systems. Consider the OS/360 job as it
might be tackled with a small, sharp team. Postulate a 10-man
team. As a bound, let them be seven times as productive as medi-
ocre programmers in both programming and documentation, be-
cause they are sharp. Assume OS/360 was built only by mediocre
programmers (which is far from the truth). As a bound, assume
that another productivity improvement factor of seven comes
from reduced communication on the part of the smaller team.
Assume the same team stays on the entire job. Well, 5000/(10 X
7 X 7 ) = 10; they can do the 5000 man-year job in 10 years. Will
the product be interesting 10 years after its initial design? Or will
it have been made obsolete by the rapidly developing software
technology?
The dilemma is a cruel one. For efficiency and conceptual
integrity, one prefers a few good minds doing design and construc-
tion. Yet for large systems one wants a way to bring considerable
manpower to bear, so that the product can make a timely appear-
ance. How can these two needs be reconciled?
32 The Surgical Team
Mills's Proposal
A proposal by Harlan Mills offers a fresh and creative solu-
tion.2'3 Mills proposes that each segment of a large job be tackled
by a team, but that the team be organized like a surgical team
rather than a hog-butchering team. That is, instead of each mem-
ber cutting away on the problem, one does the cutting and the
others give him every support that will enhance his effectiveness
and productivity.
A little thought shows that this concept meets the desiderata,
if it can be made to work. Few minds are involved in design and
construction, yet many hands are brought to bear. Can it work?
Who are the anesthesiologists and nurses on a programming team,
and how is the work divided? Let me freely mix metaphors to
suggest how such a team might work if enlarged to include all
conceivable support.
The surgeon. Mills calls him a chief programmer. He personally
defines the functional and performance specifications, designs the
program, codes it, tests it, and writes its documentation. He writes
in a structured programming language such as PL/I, and has effec-
tive access to a computing system which not only runs his tests but
also stores the various versions of his programs, allows easy file
updating, and provides text editing for his documentation. He
needs great talent, ten years experience, and considerable systems
and application knowledge, whether in applied mathematics,
business data handling, or whatever.
The copilot. He is the alter ego of the surgeon, able to do any
part of the job, but is less experienced. His main function is to
share in the design as a thinkerrdiscussant, and evaluator. The
surgeon tries ideas on him, but is not bound by his advice. The
copilot often represents his team in discussions of function and
interface with other teams. He knows all the code intimately. He
researches alternative design strategies. He obviously serves as
insurance against disaster to the surgeon. He may even write code,
but he is not responsible for any part of the code.
Mills's Proposal 33
The administrator. The surgeon is boss, and he must have the
last word on personnel, raises, space, and so on, but he must spend
almost none of his time on these matters. Thus he needs a profes-
sional administrator who handles money, people, space, and ma-
chines, and who interfaces with the administrative machinery of
the rest of the organization. Baker suggests that the administrator
has a full-time job only if the project has substantial legal, con-
tractual, reporting, or financial requirements because of the user-
producer relationship. Otherwise, one administrator can serve two
teams.
The editor. The surgeon is responsible for generating the docu-
mentation—for maximum clarity he must write it. This is true of
both external and internal descriptions. The editor, however, takes
the draft or dictated manuscript produced by the surgeon and
criticizes it, reworks it, provides it with references and bibliogra-
phy, nurses it through several versions, and oversees the mechan-
ics of production.
Two secretaries. The administrator and the editor will each need
a secretary; the administrator's secretary will handle project corre-
spondence and non-product files.
The program clerk. He is responsible for maintaining all the
technical records of the team in a programming-product library.
The clerk is trained as a secretary and has responsibility for both
machine-readable and human-readable files.
All computer input goes to the clerk, who logs and keys it if
required. The output listings go back to him to be filed and in-
dexed. The most recent runs of any model are kept in a status
notebook; all previous ones are filed in a chronological archive.
Absolutely vital to Mills's concept is the transformation of
programming "from private art to public practice" by making all
the computer runs visible to all team members and identifying all
programs and data as team property, not private property.
The specialized function of the program clerk relieves pro-
grammers of clerical chores, systematizes and ensures proper per-
34 The Surgical Team
formance of those oft-neglected chores, and enhances the team's
most valuable asset—its work-product. Clearly the concept as set
forth above assumes batch runs. When interactive terminals are
used, particularly those with no hard-copy output, the program
clerk's functions do not diminish, but they change. Now he logs
all updates of team program copies from private working copies,
still handles all batch runs, and uses his own interactive facility to
control the integrity and availability of the growing product.
The toolsmith. File-editing, text-editing, and interactive debug-
ging services are now readily available, so that a team will rarely
need its own machine and machine-operating crew. But these
services must be available with unquestionably satisfactory re-
sponse and reliability; and the surgeon must be sole judge of the
adequacy of the service available to him. He needs a toolsmith,
responsible for ensuring this adequacy of the basic service and for
constructing, maintaining, and upgrading special tools—mostly
interactive computer services—needed by his team. Each team will
need its own toolsmith, regardless of the excellence and reliability
of any centrally provided service, for his job is to see to the tools
needed or wanted by his surgeon, without regard to any other
team's needs. The tool-builder will often construct specialized
utilities, catalogued procedures, macro libraries.
The tester. The surgeon will need a bank of suitable test cases
for testing pieces of his work as he writes it, and then for testing
the whole thing. The tester is therefore both an adversary who
devises system test cases from the functional specs, and an assis-
tant who devises test data for the day-by-day debugging. He
would also plan testing sequences and set up the scaffolding re-
quired for component tests.
The language lawyer. By the time Algol came along, people
began to recognize that most computer installations have one or
two people who delight in mastery of the intricacies of a program-
ming language. And these experts turn out to be very useful and
very widely consulted. The talent here is rather different from that
of the surgeon, who is primarily a system designer and who thinks
How It Works 35
representations. The language lawyer can 6nd a neat and efficient
way to use the language to do difficult, obscure, or tricky things.
Often he will need to do small studies (two or three days) on good
technique. One language lawyer can service two or three surgeons.
This, then, is how 10 people might contribute in well-
differentiated and specialized roles on a programming team built
on the surgical model.
How It Works
The team just defined meets the desiderata in several ways. Ten
people, seven of them professionals, are at work on the problem,
but the system is the product of one mind—or at most two, acting
uno animo.
Notice in particular the differences between a team of two
programmers conventionally organized and the surgeon-copilot
team. First, in the conventional team the partners divide the work,
and each is responsible for design and implementation of part of
the work. In the surgical team, the surgeon and copilot are each
cognizant of all of the design and all of the code. This saves the
labor of allocating space, disk accesses, etc. It also ensures the
conceptual integrity of the work.
Second, in the conventional team the partners are equal, and
the inevitable differences of judgment must be talked out or com-
promised. Since the work and resources are divided, the differ-
ences in judgment are confined to overall strategy and interfacing,
but they are compounded by differences of interest—e.g., whose
space will be used for a buffer. In the surgical team, there are no
differences of interest, and differences of judgment are settled by
the surgeon unilaterally. These two differences—lack of division
of the problem and the superior-subordinate relationship—make
it possible for the surgical team to act uno animo.
Yet the specialization of function of the remainder of the team
is the key to its efficiency, for it permits a radically simpler com-
munication pattern among the members, as Fig. 3.1 shows.
4
Aristocracy, Democracy,
and System Design
4
Aristocracy, Democracy,
and System Design
This great church is an incomparable work of art. There is
neither aridity nor confusion in the tenets it sets forth. . , ,
It is the zenith of a style, the work of artists who had
understood and assimilated all their predecessors' successes,
in complete possession of the techniques of their times, but
using them without indiscreet display nor gratuitous feats
of skill.
It was Jean d 'Orbais who undoubtedly conceived the
general plan of the building, a plan which was respected,
at least in its essential elements, by his successors. This is
one of the reasons for the extreme coherence and unity of
the edifice.
REIMS CATHEDRAL GUIDEBOOK1
Photographies Emmanuel Boudot-Lamotte
41
42 Aristocracy, Democracy, and System Design
Conceptual Integrity
Most European cathedrals show differences in plan or architec-
tural style between parts built in different generations by different
builders. The later builders were tempted to "improve" upon the
designs of the earlier ones, to reflect both changes in fashion and
differences in individual taste. So the peaceful Norman transept
abuts and contradicts the soaring Gothic nave, and the result pro-
claims the pridefulness of the builders as much as the glory of
God.
Against these, the architectural unity of Reims stands in glori-
ous contrast. The joy that stirs the beholder comes as much from
the integrity of the design as from any particular excellences. As
the guidebook tells, this integrity was achieved by the self-abne-
gation of eight generations of builders, each of whom sacrificed
some of his ideas so that the whole might be of pure design. The
result proclaims not only the glory of God, but also His power to
salvage fallen men from their pride.
Even though they have not taken centuries to build, most
programming systems reflect conceptual disunity far worse than
that of cathedrals. Usually this arises not from a serial succession
of master designers, but from the separation of design into many
tasks done by many men.
I will contend that conceptual integrity is the most important
consideration in system design. It is better to have a system omit
certain anomalous features and improvements, but to reflect one
set of design ideas, than to have one that contains many good but
independent and uncoordinated ideas. In this chapter and the next
two, we will examine the consequences of this theme for program-
ming system design:
• How is conceptual integrity to be achieved?
• Does not this argument imply an elite, or aristocracy of archi-
tects, and a horde of plebeian implementers whose creative
talents and ideas are suppressed?
Achieving Conceptual Integrity 43
• How does one keep the architects from drifting off into the
blue with unimplementable or costly specifications?
• How does one ensure that every trifling detail of an architec-
tural specification gets communicated to the implementer,
properly understood by him, and accurately incorporated into
the product?
Achieving Conceptual Integrity
The purpose of a programming system is to make a computer easy
to use. To do this, it furnishes languages and various facilities that
are in fact programs invoked and controlled by language features.
But these facilities are bought at a price: the external description
of a programming system is ten to twenty times as large as the
external description of the computer system itself. The user finds
it far easier to specify any particular function, but there are far
more to choose from, and far more options and formats to remem-
ber.
Ease of use is enhanced only if the time gained in functional
specification exceeds the time lost in learning, remembering, and
searching manuals. With modern programming systems this gain
does exceed the cost, but in recent years the ratio of gain to cost
seems to have fallen as more and more complex functions have
been added. I am haunted by the memory of the ease of use of the
IBM 650, even without an assembler or any other software at all.
Because ease of use is the purpose, this ratio of function to
conceptual complexity is the ultimate test of system design. Nei-
ther function alone nor simplicity alone defines a good design.
This point is widely misunderstood. Operating System/360 is
hailed by its builders as the finest ever built, because it indisputa-
bly has the most function. Function, and not simplicity, has al-
ways been the measure of excellence for its designers. On the
other hand, the Time-Sharing System for the PDP-10 is hailed by
its builders as the finest, because of its simplicity and the spareness
44 Aristocracy, Democracy, and System Design
of its concepts. By any measure, however, its function is not even
in the same class as that of OS/360. As soon as ease of use is held
up as the criterion, each of these is seen to be unbalanced, reaching
for only half of the true goal.
For a given level of function, however, that system is best in
which one can specify things with the most simplicity aiid
straightforwardness. Simplicity is not enough. Mooers's TRAC
language and Algol 68 achieve simplicity as measured by the num-
ber of distinct elementary concepts. They are not, however,
straightforward. The expression of the things one wants to do often
requires involuted and unexpected combinations of the basic facil-
ities. It is not enough to learn the elements and rules of combina-
tion; one must also learn the idiomatic usage, a whole lore of how
the elements are combined in practice. Simplicity and straightfor-
wardness proceed from conceptual integrity. Every part must re-
flect the same philosophies and the same balancing of desiderata.
Every part must even use the same techniques in syntax and
analogous notions in semantics. Ease of use, then, dictates unity
of design, conceptual integrity.
Aristocracy and Democracy
Conceptual integrity in turn dictates that the design must proceed
from one mind, or from a very small number of agreeing resonant
minds.
Schedule pressures, however, dictate that system building
needs many hands. Two techniques are available for resolving this
dilemma. The first is a careful division of labor between architec-
ture and implementation. The second is the new way of structur-
ing programming implementation teams discussed in the previous
chapter.
The separation of architectural effort from implementation is
a very powerful way of getting conceptual integrity on very large
projects. I myself have seen it used with great success on IBM's
Stretch computer and on the System/360 computer product line.
Aristocracy and Democracy 45
I have seen it fail through lack of application on Operating Sys-
tem/360.
By the architecture of a system, I mean the complete and de-
tailed specification of the user interface. For a computer this is the
programming manual. For a compiler it is the language manual. For
a control program it is the manuals for the language or languages
used to invoke its functions. For the entire system it is the union
of the manuals the user must consult to do his entire job.
The architect of a system, like the architect of a building, is
the user's agent. It is his job to bring professional and technical
knowledge to bear in the unalloyed interest of the user, as opposed
to the interests of the salesman, the fabricator, etc.8
Architecture must be carefully distinguished from implemen-
tation. As Blaauw has said, "Where architecture tells what hap-
pens, implementation tells how it is made to happen."3 He gives
as a simple example a clock, whose architecture consists of the
face, the hands, and the winding knob. When a child has learned
this architecture, he can tell time as easily from a wristwatch as
from a church tower. The implementation, however, and its real-
ization, describe what goes on inside the case—powering by any
of many mechanisms and accuracy control by any of many.
In System/360, for example, a single computer architecture is
implemented quite differently in each of some nine models. Con-
versely, a single implementation, the Model 30 data flow, memory,
and microcode, serves at different times for four different architec-
tures: a System/360 computer, a multiplex channel with up to 224
logically independent subchannels, a selector channel, and a 1401
computer.4
The same distinction is equally applicable to programming
systems. There is a U.S. standard Fortran IV. This is the architec-
ture for many compilers. Within this architecture many imple-
mentations are possible: text-in-core or compiler-in-core,
fast-compile or optimizing, syntax-directed or ad-hoc. Likewise
any assembler language or job-control language admits of many
implementations of the assembler or scheduler.
46 Aristocracy, Democracy, and System Design
Now we can deal with the deeply emotional question of aris-
tocracy versus democracy. Are not the architects a new aristocracy,
an intellectual elite, set up to tell the poor dumb implementers
what to do? Has not all the creative work been sequestered for this
elite, leaving the implementers as cogs in the machine? Won't one
get a better product by getting the good ideas from all the team,
following a democratic philosophy, rather than by restricting the
development of specifications to a few?
As to the last question, it is the easiest. I will certainly not
contend that only the architects will have good architectural ideas.
Often the fresh concept does come from an implementer or from
a user. However, all my own experience convinces me, and I have
tried to show, that the conceptual integrity of a system determines
its ease of use. Good features and ideas that do not integrate with
a system's basic concepts are best left out. If there appear many
such important but incompatible ideas, one scraps the whole sys-
tem and starts again on an integrated system with different basic
concepts.
As to the aristocracy charge, the answer must be yes and no.
Yes, in the sense that there must be few architects, their product
must endure longer than that of an implementer, and the architect
sits at the focus of forces which he must ultimately resolve in the
user's interest. If a system is to have conceptual integrity, someone
must control the concepts. That is an aristocracy that needs no
apology.
No, because the setting of external specifications is not more
creative work than the designing of implementations. It is just
different creative work. The design of an implementation, given an
architecture, requires and allows as much design creativity, as
many new ideas, and as much technical brilliance as the design of
the external specifications. Indeed, the cost-performance ratio of
the product will depend most heavily on the implementer, just as
ease of use depends most heavily on the architect.
There are many examples from other arts and crafts that lead
one to believe that discipline is good for art. Indeed, an artist's
What Does the Implementer Do While Waiting? 47
aphorism asserts, "Form is liberating." The worst buildings are
those whose budget was too great for the purposes to be served.
Bach's creative output hardly seems to have been squelched by the
necessity of producing a limited-form cantata each week. I am sure
that the Stretch computer would have had a better architecture
had it been more tightly constrained; the constraints imposed by
the System/360 Model 30's budget were in my opinion entirely
beneficial for the Model 75's architecture.
Similarly, I observe that the external provision of an architec-
ture enhances, not cramps, the creative style of an implementing
group. They focus at once on the part of the problem no one has
addressed, and inventions begin to flow. In an unconstrained im-
plementing group, most thought and debate goes into architectural
decisions, and implementation proper gets short shrift.5
This effect, which I have seen many times, is confirmed by
R. W. Conway, whose group at Cornell built the PL/C compiler
for the PL/I language. He says, "We finally decided to implement
the language unchanged and unimproved, for the debates about
language would have taken all our effort."6
What Does the Implementer Do While Waiting?
It is a very humbling experience to make a multimillion-dollar
mistake, but it is also very memorable. I vividly recall the night
we decided how to organize the actual writing of external specifi-
cations for OS/360. The manager of architecture, the manager of
control program implementation, and I were threshing out the
plan, schedule, and division of responsibilities.
The architecture manager had 10 good men. He asserted that
they could write the specifications and do it right. It would take
ten months, three more than the schedule allowed.
The control program manager had 150 men. He asserted that
they could prepare the specifications, with the architecture team
coordinating; it would be well-done and practical, and he could do
it on schedule. Furthermore, if the architecture team did it, his 150
men would sit twiddling their thumbs for ten months.
48 Aristocracy, Democracy, and System Design
To this the architecture manager responded that if I gave the
control program team the responsibility, the result would not in
fact be on time, but would also be three months late, and of much
lower quality. I did, and it was. He was right on both counts.
Moreover, the lack of conceptual integrity made the system far
more costly to build and change, and I would estimate that it
added a year to debugging time.
Many factors, of course, entered into that mistaken decision;
but the overwhelming one was schedule time and the appeal of
putting all those 150 implementers to work. It is this siren song
whose deadly hazards I would now make visible.
When it is proposed that a small architecture team in fact
write all the external specifications for a computer or a program-
ming system, the implementers raise three objections:
• The specifications will be too rich in function and will not
reflect practical cost considerations.
• The architects will get all the creative fun and shut out the
inventiveness of the implementers.
« The many implementers will have to sit idly by while the
specifications come through the narrow funnel that is the
architecture team.
The first of these is a real danger, and it will be treated in the
next chapter. The other two are illusions, pure and simple. As we
have seen above, implementation is also a creative activity of the
first order. The opportunity to be creative and inventive in imple-
mentation is not significantly diminished by working within a
given external specification, and the order of creativity may even
be enhanced by that discipline. The total product will surely be.
The last objection is one of timing and phasing. A quick an-
swer is to refrain from hiring implementers until the specifications
are complete. This is what is done when a building is constructed.
In the computer systems business, however, the pace is
quicker, and one wants to compress the schedule as much as
possible. How much can specification and building be overlapped?
What Does the Implementer Do While Waiting? 49
As Blaauw points out, the total creative effort involves three
distinct phases: architecture, implementation, and realization. It
turns out that these can in fact be begun in parallel and proceed
simultaneously.
In computer design, for example, the implementer can start as
soon as he has relatively vague assumptions about the manual,
somewhat clearer ideas about the technology, and well-defined
cost and performance objectives. He can begin designing data
flows, control sequences, gross packaging concepts, and so on. He
devises or adapts the tools he will need, especially the record-
keeping system, including the design automation system.
Meanwhile, at the realization level, circuits, cards, cables,
frames, power supplies, and memories must each be designed,
refined, and documented. This work proceeds in parallel with
architecture and implementation.
The same thing is true in programming system design. Long
before the external specifications are complete, the implementer
has plenty to do. Given some rough approximations as to the
function of the system that will be ultimately embodied in the
external specifications, he can proceed. He must have well-defined
space and time objectives. He must know the system configuration
on which his product must run. Then he can begin designing
module boundaries, table structures, pass or phase breakdowns,
algorithms, and all kinds of tools. Some time, too, must be spent
in communicating with the architect.
Meanwhile, on the realization level there is much to be done
also. Programming has a technology, too. If the machine is a new
one, much work must be done on subroutine conventions, super-
visory techniques, searching and sorting algorithms.7
Conceptual integrity does require that a system reflect a single
philosophy and that the specification as seen by the user flow from
a few minds. Because of the real division of labor into architecture,
implementation, and realization, however, this does not imply that
a system so designed will take longer to build. Experience shows
the opposite, that the integral system goes together faster and
50 Aristocracy, Democracy, and System Design
takes less time to test. In effect, a widespread horizontal division
of labor has been sharply reduced by a vertical division of labor,
and the result is radically simplified communications and im-
proved conceptual integrity.
5
The Second-System Effect
5
The Second-System Effect
Adde parvum parvo magnus acervus erit.
{Add little to little and there will be a Ug pile. ]
OVID
Turning house for air traffic. Lithograph, Paris, 1882
From Le Vingtiime Stecle by A. Robida
53
54 The Second-System Effect
If one separates responsibility for functional specification from
responsibility for building a fast, cheap product, what discipline
bounds the architect's inventive enthusiasm?
The fundamental answer is thoroughgoing, careful, and sym-
pathetic communication between architect and builder. Neverthe-
less there are finer-grained answers that deserve attention.
Interactive Discipline for the Architect
The architect of a building works against a budget, using estimat-
ing techniques that are later confirmed or corrected by the con-
tractors' bids. It often happens that all the bids exceed the budget.
The architect then revises his estimating technique upward and his
design downward for another iteration. He may perhaps suggest
to the contractors ways to implement his design more cheaply
than they had devised.
An analogous process governs the architect of a computer
system or a programming system. He has, however, the advantage
of getting bids from the contractor at many early points in his
design, almost any time he asks for them. He usually has the
disadvantage of working with only one contractor, who can raise
or lower his estimates to reflect his pleasure with the design. In
practice, early and continuous communication can give the archi-
tect good cost readings and the builder confidence in the design
without blurring the clear division of responsibilities.
The architect has two possible answers when confronted with
an estimate that is too high: cut the design or challenge the esti-
mate by suggesting cheaper implementations. This latter is inher-
ently an emotion-generating activity. The architect is now
challenging the builder's way of doing the builder's job. For it to
be successful, the architect must
• remember that the builder has the inventive and creative re-
sponsibility for the implementation; so the architect suggests,
not dictates;
Self-Discipline—The Second-System 55
• always be prepared to suggest a way of implementing any-
thing he specifies, and be prepared to accept any other way
that meets the objectives as well;
• deal quietly and privately in such suggestions;
• be ready to forego credit for suggested improvements.
Normally the builder will counter by suggesting changes to
the architecture. Often he is right—some minor feature may have
unexpectedly large costs when the implementation is worked out.
Self-Discipline—The Second-System Effect
An architect's first work is apt to be spare and clean. He knows he
doesn't know what he's doing, so he does it carefully and with
great restraint.
As he designs the first work, frill after frill and embellishment
after embellishment occur to him. These get stored away to he
used "next time." Sooner or later the first system is finished, and
the architect, with firm, confidence and a demonstrated mastery of
that class of systems, is ready to build a second system.
This second is the most dangerous system a man ever designs.
When he does his third and later ones, his prior experiences will
confirm each other as to the general characteristics of such sys-
tems, and their differences will identify those parts of his experi-
ence that are particular and not generalizable.
The general tendency is to over-design the second system,
using all the ideas and frills that were cautiously sidetracked on
the first one. The result, as Ovid says, is a "big pile." For example,
consider the IBM 709 architecture, later embodied in the 7090.
This is an upgrade, a second system for the very successful and
clean 704. The operation set is so rich and profuse that only about
half of it was regularly used.
Consider as a stronger case the architecture, implementation,
and even the realization of the Stretch computer, an outlet for the
56 The Second-System Effect
pent-up inventive desires of many people, and a second system for
most of them. As Strachey says in a review:
I get the impression that Stretch is in some way the end of one line
of development. Like some early computer programs it is immensely
ingenious, immensely complicated, and extremely effective, but some-
how at the same time crude, wasteful, and inelegant, and one feels
that there must be a better way of doing things. *
Operating System/360 was the second system for most of its
designers. Groups of its designers came from building the 1410-
7010 disk operating system, the Stretch operating system, the
Project Mercury real-time system, and IBSYS for the 7090. Hardly
anyone had experience with two previous operating systems.8 So
OS/360 is a prime example of the second-system effect, a Stretch
of the software art to which both the commendations and the
reproaches of Strachey's critique apply unchanged.
For example, OS/360 devotes 26 bytes of the permanently
resident date-turnover routine to the proper handling of Decem-
ber 31 on leap years (when it is Day 366). That might have been
left to the operator.
The second-system effect has another manifestation some-
what different from pure functional embellishment. That is a ten-
dency to refine techniques whose very existence has been made
obsolete by changes in basic system assumptions. OS/360 has
many examples of this.
Consider the linkage editor, designed to load separately-com-
piled programs and resolve their cross-references. Beyond this
basic function it also handles program overlays. It is one of the
finest overlay facilities ever built. It allows overlay structuring to
be done externally, at linkage time, without being designed into
the source code. It allows the overlay structure to be changed from
run to run without recompilation. It furnishes a rich variety of
useful options and facilities. In a sense it is the culmination of
years of development of static overlay technique.
Self-Discipline—The Second-System Effect. 57
Yet it is also the last and finest of the dinosaurs, for it belongs
to a system in which multiprogramming is the normal mode and
dynamic core allocation the basic assumption. This is in direct
conflict with the notion of using static overlays. How much better
the system would work if the efforts devoted to overlay manage-
ment had been spent on making the dynamic core allocation and
the dynamic cross-referencing facilities really fast!
Furthermore, the linkage editor requires so much space and
itself contains many overlays that even when it is used just for
linkage without overlay management, it is slower than most of the
system compilers. The irony of this is that the purpose of the
linker is to avoid recompilation. Like a skater whose stomach gets
ahead of his feet, refinement proceeded until the system assump-
tions had been quite outrun.
The TESTRAN debugging facility is another example of this
tendency. It is the culmination of batch debugging facilities, fur-
nishing truly elegant snapshot and core dump capabilities. It uses
the control section concept and an ingenious generator technique
to allow selective tracing and snapshotting without interpretive
overhead or recompilation. The imaginative concepts of the Share
Operating System3 for the 709 have been brought to full bloom.
Meanwhile, the whole notion of batch debugging without
reeompilation was becoming obsolete. Interactive computing sys-
tems, using language interpreters or incremental compilers have
provided the most fundamental challenge. But even in batch sys-
tems, the appearance of fast-compile/slow-execute compilers has
made source-level debugging and snapshotting the preferred tech-
nique. How much better the system would have been if the TES-
TRAN effort had been devoted instead to building the interactive
and fast-compile facilities earlier and better!
Yet another example is the scheduler, which provides truly
excellent facilities for managing a fixed-batch job stream. In a real
sense, this scheduler is the refined, improved, and embellished
second system succeeding the 1410-7010 Disk Operating System,
§8 The Second-System Effect
a batch system unmultiprogranuned except for input-output and
intended chiefly for business applications. As such, the OS/360
scheduler is good. But it is almost totally uninfluenced by the
OS/360 needs of remote job entry, multiprogramming, and per-
manently resident interactive subsystems. Indeed, the scheduler's
design makes these hard.
How does the architect avoid the second-system effect? Well,
obviously he can't skip his second system. But he can be conscious
of the peculiar hazards of that system, and exert extra self-disci-
pline to avoid functional ornamentation and to avoid extrapola-
tion of functions that are obviated by changes in assumptions and
purposes.
A discipline that will open an architect's eyes is to assign each
little function a value: capability x is worth not more than m bytes
of memory and n microseconds per invocation. These values will
guide initial decisions and serve during implementation as a guide
and warning to all.
How does the project manager avoid the second-system
effect? By insisting on a senior architect who has at least two
systems under his belt. Too, by staying aware of the special temp-
tations, he can ask the right questions to ensure that the philo-
sophical concepts and objectives are fully reflected in the detailed
design.
6
Passing the Word
6
Passing the Word
Hell sit here and he'll say, "Do this! Do that!" And
nothing will happen.
HARRYS. TRUMAN. ON PRESIDENTIAL POWER1
"The Seven Trumpets" from The Wells Apocalypse, 14th century
The Bettman Archive
61
62 Passing the Word
Assuming that he has the disciplined, experienced architects and
that there are many implementers, how shall the manager ensure
that everyone hears, understands, and implements the architects'
decisions? How can a group of 10 architects maintain the concep-
tual integrity of a system which 1000 men are building? A whole
technology for doing this was worked out for the System/360
hardware design effort, and it is equally applicable to software
projects.
Written Specifications—the Manual
The manual, or written specification, is a necessary tool, though
not a sufficient one. The manual is the external specification of the
product. It describes and prescribes every detail of what the user
sees. As such, it is the chief product of the architect.
Round and round goes its preparation cycle, as feedback from
users and implementers shows where the design is awkward to use
or build. For the sake of implementers it is important that the
changes be quantized—that there be dated versions appearing on
a schedule.
The manual must not only describe everything the user does
see, including all interfaces; it must also refrain from describing
what the user does not see. That is the implementer's business, and
there his design freedom must be unconstrained. The architect
must always be prepared to show an implementation for any
feature he describes, but he must not attempt to dictate the imple-
mentation.
The style must be precise, full, and accurately detailed. A user
will often refer to a single definition, so each one must repeat all
the essentials and yet all must agree. This tends to make manuals
dull reading, but precision is more important than liveliness.
The unity of System/360's Principles of Operation springs from
the fact that only two pens wrote it: Gerry Blaauw's and Andris
Padegs'. The ideas are those of about ten men, but the casting of
those decisions into prose specifications must be done by only one
Formal Definitions 63
or two, if the consistency of prose and product is to be maintained.
For the writing of a definition will necessitate a host of mini-
decisions which are not of full-debate importance. An example in
System/360 is the detail of how the Condition Code is set after
each operation. Not trivial, however, is the principle that such
mini-decisions be made consistently throughout.
I think the finest piece of manual writing I have ever seen is
Blaauw's Appendix to System/360 Principles of Operation. This de-
scribes with care and precision the limits of System/360 compati-
bility. It defines compatibility, prescribes what is to be achieved,
and enumerates those areas of external appearance where the ar-
chitecture is intentionally silent and where results from one model
may differ from those of another, where one copy of a given model
may differ from another copy, or where a copy may differ even
from itself after an engineering change. This is the level of preci-
sion to which manual writers aspire, and they must define what
is not prescribed as carefully as what is.
Formal Definitions
English, or any other human language, is not naturally a precision
instrument for such definitions. Therefore the manual writer must
strain himself and his language to achieve the precision needed.
An attractive alternative is to use a formal notation for such defini-
tions. After all, precision is the stock in trade, the raison d'etre of
formal notations.
Let us examine the merits and weaknesses of formal defini-
tions. As noted, formal definitions are precise. They tend to be
complete; gaps show more conspicuously, so they are filled sooner.
What they lack is comprehensibility. With English prose one can
show structural principles, delineate structure in stages or levels,
and give examples. One can readily mark exceptions and empha-
size contrasts. Most important, one can explain why. The formal
definitions put forward so far have inspired wonder at their ele-
gance and confidence in their precision. But they have demanded
64 Passing the Word
prose explanations to make their content easy to learn and teach.
For these reasons, I think we will see future specifications to con-
sist of both a formal definition and a prose definition.
An ancient adage warns, "Never go to sea with two chronom-
eters; take one or three." The same thing clearly applies to prose
and formal definitions. If one has both, one must be the standard,
and the other must be a derivative description, clearly labeled as
such. Either can be the primary standard. Algol 68 has a formal
definition as standard and a prose definition as descriptive. PL/I
has the prose as standard and the formal description as derivative.
System/360 also has prose as standard with a derived formal de-
scription.
Many tools are available for formal definition. The Backus-
Naur Form is familiar for language definition, and it is amply
discussed in the literature.2 The formal description of PL/I uses
new notions of abstract syntax, and it is adequately described.3
Iverson's APL has been used to describe machines, most notably
the IBM 70904 and System/360.5
Bell and Newell have proposed new notations for describing
both configurations and machine architectures, and they have il-
lustrated these with several machines, including the DEC PDP-8,6
the 7090,6 and System/360.7
Almost all formal definitions turn out to embody or describe
an implementation of the hardware or software system whose
externals they are prescribing. Syntax can be described without
this, but semantics are usually defined by giving a program that
carries out the defined operation. This is of course an implementa-
tion, and as such it over-prescribes the architecture. So one must
take care to indicate that the formal definition applies only to
externals, and one must say what these are.
Not only is a formal definition an implementation, an imple-
mentation can serve as a formal definition. When the first compat-
ible computers were built, this was exactly the technique used.
The new machine was to match an existing machine. The manual
was vague on some points? "Ask the machine!" A test program
Formal Definitions 65
would be devised to determine the behavior, and the new machine
would be built to match.
A programmed simulator of a hardware or software system
can serve in precisely the same way. It is an implementation; it
runs. So all questions of definition can be resolved by testing it.
Using an implementation as a definition has some advantages.
All questions can be settled unambiguously by experiment. De-
bate is never needed, so answers are quick. Answers are always as
precise as one wants, and they are always correct, by definition.
Opposed to these one has a formidable set of disadvantages. The
implementation may over-prescribe even the externals. Invalid
syntax always produces some result; in a policed system that result
is an invalidity indication and nothing more. In an unpoliced system
all kinds of side effects may appear, and these may have been used
by programmers. When we undertook to emulate the IBM 1401
on System/360, for example, it developed that there were 30
different "curios"—side effects of supposedly invalid operations—
that had come into widespread use and had to be considered as
part of the definition. The implementation as a definition overpre-
scribed; it not only said what the machine must do, it also said a
great deal about how it had to do it.
Then, too, the implementation will sometimes give unex-
pected and unplanned answers when sharp questions are asked,
and the de facto definition will often be found to be inelegant in
these particulars precisely because they have never received any
thought. This inelegance will often turn out to be slow or costly
to duplicate in another implementation. For example, some ma-
chines leave trash in the multiplicand register after a multiplica-
tion. The precise nature of this trash turns out to be part of the
de facto definition, yet duplicating it may preclude the use of a
faster multiplication algorithm.
Finally, the use of an implementation as a formal definition is
peculiarly susceptible to confusion as to whether the prose de-
scription or the formal description is in fact the standard. This is
especially true of programmed simulations. One must also refrain
66 Passing the Word
from modifications to the implementation while it is serving as a
standard.
Direct Incorporation
A lovely technique for disseminating and enforcing definitions is
available for the software system architect. It is especially useful
for establishing the syntax, if not the semantics, of intermodule
interfaces. This technique is to design the declaration of the passed
parameters or shared storage, and to require the implementations
to include that declaration via a compile-time operation (a macro
or a % INCLUDE in PL/I). If, in addition, the whole interface is
referenced only by symbolic names, the declaration can be
changed by adding or inserting new variables with only recompi-
lation, not alteration, of the using program.
Conferences and Courts
Needless to say, meetings are necessary. The hundreds of man-to-
man consultations must be supplemented by larger and more for-
mal gatherings. We found two levels of these to be useful. The first
is a weekly half-day conference of all the architects, plus official
representatives of the hardware and software implementers, and
the market planners. The chief system architect presides.
Anyone can propose problems or changes, but proposals are
usually distributed in writing before the meeting. A new problem
is usually discussed a while. The emphasis is on creativity, rather
than merely decision. The group attempts to invent many solu-
tions to problems, then a few solutions are passed to one or more
of the architects for detailing into precisely worded manual change
proposals.
Detailed change proposals then come up for decisions. These
have been circulated and carefully considered by implementers
and users, and the pros and cons are well delineated. If a consensus
emerges, well and good. If not, the chief architect decides. Minutes
Conferences and Courts §7
are kept and decisions are formally, promptly, and widely dis-
seminated.
Decisions from the weekly conferences give quick results and
allow work to proceed. If anyone is too unhappy, instant appeals
to the project manager are possible, but this happens very rarely.
The fruitfulness of these meetings springs from several
sources:
1. The same group—architects, users, and implementers—meets
weekly for months. No time is needed for bringing people up
to date.
2. The group is bright, resourceful, well versed in the issues, and
deeply involved in the outcome. No one has an "advisory"
role. Everyone is authorized to make binding commitments.
3. When problems are raised, solutions are sought both within
and outside the obvious boundaries.
4. The formality of written proposals focuses attention, forces
decision, and avoids committee-drafted inconsistencies.
5. The clear vesting of decision-making power in the chief archi-
tect avoids compromise and delay.
As time goes by, some decisions don't wear well. Some minor
matters have never been wholeheartedly accepted by one or an-
other of the participants. Other decisions have developed unfore-
seen problems, and sometimes the weekly meeting didn't agree to
reconsider these. So there builds up a backlog of minor appeals,
open issues, or disgruntlements. To settle these we held annual
supreme court sessions, lasting typically two weeks. (I would hold
them every six months if I were doing it again.)
These sessions were held just before major freeze dates for the
manual. Those present included not only the architecture group
and the programmers' and implementers' architectural representa-
tives, but also the managers of programming, marketing, and im-
plementation efforts. The System/360 project manager presided.
The agenda typically consisted of about 200 items, mostly minor,
which were enumerated in charts placarded around the room. All
68 Passing the Word
sides were heard and decisions made. By the miracle of computer-
ized text editing (and lots of fine staff work), each participant
found an updated manual, embodying yesterday's decisions, at his
seat every morning.
These "fall festivals" were useful not only for resolving deci-
sions, but also for getting them accepted. Everyone was heard,
everyone participated, everyone understood better the intricate
constraints and interrelationships among decisions.
Multiple Implementations
System/360 architects had two almost unprecedented advantages:
enough time to work carefully, and political clout equal to that of
the implementers. The provision of enough time came from the
schedule of the new technology; the political equality came from
the simultaneous construction of multiple implementations. The
necessity for strict compatibility among these served as the best
possible enforcing agent for the specifications.
In most computer projects there comes a day when it is discov-
ered that the machine and the manual don't agree. When the
confrontation follows, the manual usually loses, for it can be
changed far more quickly and cheaply than the machine. Not so,
however, when there are multiple implementations. Then the de-
lays and costs associated with fixing the errant machine can be
overmatched by delays and costs in revising the machines that
followed the manual faithfully.
This notion can be fruitfully applied whenever a programming
language is being defined. One can be certain that several inter-
preters or compilers will sooner or later have to be built to meet
various objectives. The definition will be cleaner and the discipline
tighter if at least two implementations are built initially.
The Telephone Log
As implementation proceeds, countless questions of architectural
interpretation arise, no matter how precise the specification. Obvi-
Product Test 69'
ously many such questions require amplifications and clarifica-
tions in the text. Others merely reflect misunderstandings.
It is essential, however, to encourage the puzzled implementer
to telephone the responsible architect and ask his question, rather
than to guess and proceed. It is just as vital to recognize that the
answers to such questions are ex cathedra architectural pronounce-
ments that must be told to everyone.
One useful mechanism is a telephone log kept by the architect.
In it he records every question and every answer. Each week the
logs of the several architects are concatenated, reproduced, and
distributed to the users and implementers. While this mechanism
is quite informal, it is both quick and comprehensive.
Product Test
The project manager's best friend is his daily adversary, the inde-
pendent product-testing organization. This group checks ma-
chines and programs against specifications and serves as a devil's
advocate, pinpointing every conceivable defect and discrepancy.
Every development organization needs such an independent tech-
nical auditing group to keep it honest.
In the last analysis the customer is the independent auditor.
In the merciless light of real use, every flaw will show. The prod-
uct-testing group then is the surrogate customer, specialized for
finding flaws. Time after time, the careful product tester will find
places where the word didn't get passed, where the design deci-
sions were not properly understood or accurately implemented.
For this reason such a testing group is a necessary link in the chain
by which the design word is passed, a link that needs to operate
early and simultaneously with design.
7
Why Did the Tower
of Babel Fail?
7
Why Did theTower
of Babel Fail?
Now the whole earth used only one language, with few
words. On the occasion of a migration from the east, men
discovered a plain in the land of Shinar, and settled there.
Then they said to one another, "Come, let us make bricks,
burning them well. " So they used bricks for stone, and
bitumen for mortar. Then they said, "Come, let us build
ourselves a city with a tower whose top shall reach the
heavens (thus making a name for ourselves), so that we
may not be scattered all over the earth. " Then the Lord
came down to look at the city and tower which human
beings had built. The Lord said, "They are just one people,
and they all have the same language. If this is what they
can do as a beginning, then nothing that they resolve to do
will be impossible for them. Come, let us go down, and
there make such a babble of their language that they wilt
not understand one another's speech. " Thus the Lord
dispersed them from there all over the earth, so that they
had to stop building the city.
GENESIS 11:1-8
P. Breughel, the Elder, "Turmbau zu Babel," 1563
Kunsthistorisches Museum, Vienna
73
74 Why Did The Tower of Babel Fail?
A Management Audit of the Babel Project
According to the Genesis account, the tower of Babel was man's
second major engineering undertaking, after Noah's ark. Babel
was the first engineering fiasco.
The story is deep and instructive on several levels. Let us,
however, examine it purely as an engineering project, and see what
management lessons can be learned. How well was their project
equipped with the prerequisites for success? Did they have:
1. A clear mission? Yes, although naively impossible. The project
failed long before it ran into this fundamental limitation.
2. Manpower? Plenty of it.
3. Materials? Clay and asphalt are abundant in Mesopotamia.
4. Enough time? Yes, there is no hint of any time constraint.
5. Adequate technology? Yes, the pyramidal or conical structure
is inherently stable and spreads the compressive load well.
Clearly masonry was well understood. The project failed be-
fore it hit technological limitations.
Well, if they had all of these things, why did the project fail?
Where did they lack? In two respects—communication, and its con-
sequent, organization. They were unable to talk with each other;
hence they could not coordinate. When coordination failed, work
ground to a halt. Reading between the lines we gather that lack
of communication led to disputes, bad feelings, and group jeal-
ousies. Shortly the clans began to move apart, preferring isolation
to wrangling.
Communication in the Large Programming Project
So it is today. Schedule disaster, functional misfits, and system
bugs all arise because the left hand doesn't know what the right
hand is doing. As work proceeds, the several teams slowly change
the functions, sizes, and speeds of their own programs, and they
explicitly or implicitly change their assumptions about the inputs
available and the uses to be made of the outputs.
The Project Workbook 75
For example, the implementer of a program-overlaying func-
tion may run into problems and reduce speedr relying on statistics
that show how rarely this function will arise in application pro-
grams. Meanwhile, back at the ranch, his neighbor may be design-
ing a major part of the supervisor so that it critically depends upon
the speed of this function. This change in speed itself becomes a
major specification change, and it needs to be proclaimed abroad
and weighed from a system point of view.
How, then, shall teams communicate with one another? In as
many ways as possible.
• Informally. Good telephone service and a clear definition of
intergroup dependencies will encourage the hundreds of calls
upon which common interpretation of written documents de-
pends.
• Meetings. Regular project meetings, with one team after an-
other giving technical briefings, are invaluable. Hundreds of
minor misunderstandings get smoked out this way.
• Workbook. A formal project workbook must be started at the
beginning. This deserves a section by itself.
The Project Workbook
What. The project workbook is not so much a separate docu-
ment as it is a structure imposed on the documents that the project
will be producing anyway.
All the documents of the project need to be part of this struc-
ture. This includes objectives, external specifications, interface
specifications, technical standards, internal specifications, and ad-
ministrative memoranda.
Why. Technical prose is almost immortal. If one examines the
genealogy of a customer manual for a piece of hardware or soft-
ware, one can trace not only the ideas, but also many of the very
sentences and paragraphs back to the first memoranda proposing
the product or explaining the first design. For the technical writer,
the paste-pot is as mighty as the pen.
76 Why Did The Tower of Babel Fail?
Since this is so, and since tomorrow's product-quality manuals
will grow from today's memos, it is very important to get the
structure of the documentation right. The early design of the
project workbook ensures that the documentation structure itself
is crafted, not haphazard. Moreover, the establishment of a struc-
ture molds later writing into segments that fit into that structure.
The second reason for the project workbook is control of the
distribution of information. The problem is not to restrict infor-
mation, but to ensure that relevant information gets to all the
people who need it.
The first step is to number all memoranda, so that ordered lists
of titles are available and each worker can see if he has what he
wants. The organization of the workbook goes well beyond this
to establish a tree-structure of memoranda. The tree-structure
allows distribution lists to be maintained by subtree, if that is
desirable.
Mechanics. As with so many programming management prob-
lems, the technical memorandum problem gets worse nonlinearly
as size increases. With 10 people, documents can simply be num-
bered. With 100 people, several linear sequences will often suffice.
With 1000, scattered inevitably over several physical locations, the
need for a structured workbook increases and the size of the work-
book increases. How then shall the mechanics be handled?
I think this was well done on the OS/360 project. The need
for a well-structured workbook was strongly urged by O. S.
Locken, who had seen its effectiveness on his previous project, the
1410-7010 operating system.
We quickly decided that each programmer should see all the
material, i.e., should have a copy of the workbook in his own
office.
Of critical importance is timely updating. The workbook must
be current. This is very difficult to do if whole documents must be
retyped for changes. In a looseleaf book, however, only pages need
to be changed. We had available a computer-driven text-editing
system, and this proved invaluable for timely maintenance. Offset
The Project Workbook 77
masters were prepared directly on the computer printer, and
turnaround time was less than a day. The recipient of all these
updated pages has an assimilation problem, however. When he
first receives a changed page, he wants to know, "What has been
changed?" When he later consults it, he wants to know, "What is
the definition today?"
The latter need is met by the continually maintained docu-
ment. Highlighting of changes requires other steps. First, one must
mark changed text on the page, e.g., by a vertical bar in the margin
alongside every altered line. Second, one needs to distribute with
the new pages a short, separately written change summary that
lists the changes and remarks on their significance.
Our project had not been under way six months before we hit
another problem. The workbook was about five feet thick! If we
had stacked up the 100 copies serving programmers in our offices
in Manhattan's Time-Life Building, they would have towered
above the building itself. Furthermore, the daily change distribu-
tion averaged two inches, some 150 pages to be interfiled in the
whole. Maintenance of the workbook began to take a significant
time from each workday.
At this point we switched to microfiche, a change that saved
a million dollars, even allowing for the cost of a microfiche reader
for each office. We were able to arrange excellent turnaround on
microfiche production; the workbook shrank from three cubic feet
to one-sixth of a cubic foot and, most significantly, updates ap-
peared in hundred-page chunks, reducing the interfiling problem
a hundredfold.
Microfiche has its drawbacks. From the manager's point of
view the awkward interfiling of paper pages ensured that the
changes were read, which was the purpose of the workbook. Mi-
crofiche would make workbook maintenance too easy, unless the
update fiche are distributed with a paper document enumerating
the changes.
Also, a microfiche cannot readily be highlighted, marked, and
commented by the reader. Documents with which the reader has
78 Why Did The Tower of Babel Fail?
interacted are more effective for the author and more useful for the
reader.
On balance I think the microfiche was a very happy mecha-
nism, and I would recommend it over a paper workbook for very
large projects.
How would one do it today? With today's system technology
available, I think the technique of choice is to keep the workbook
on the direct-access file, marked with change bars and revision
dates. Each user would consult it from a display terminal (type-
writers are too slow). A change summary, prepared daily, would
be stored in LIFO fashion at a fixed access point. The programmer
would probably read that daily, but if he missed a day he would
need only read longer the next day. As he read the change sum-
mary, he could interrupt to consult the changed text itself.
Notice that the workbook itself is not changed. It is still the
assemblage of all project documentation, structured according to
a careful design. The only change is in the mechanics of distribu-
tion and consultation. D. C. Engelbart and his colleagues at the
Stanford Research Institute have built such a system and are using
it to build and maintain documentation for the ARPA network.
D. L. Parnas of Carnegie-Mellon University has proposed a
still more radical solution.1 His thesis is that the programmer is
most effective if shielded from, rather than exposed to the details
of construction of system parts other than his own. This presup-
poses that all interfaces are completely and precisely defined.
While that is definitely sound design, dependence upon its perfect
accomplishment is a recipe for disaster. A good information sys-
tem both exposes interface errors and stimulates their correction.
Organization in the Large Programming Project
If there are n workers on a project, there are (n2-n)/2 interfaces
across which there may be communication, and there are poten-
tially almost 2" teams within which coordination must occur. The
purpose of organization is to reduce the amount of communication
Organization in the Large Programming Project 79
and coordination necessary; hence organization is a radical attack
on the communication problems treated above.
The means by which communication is obviated are division of
labor and specialization of function. The tree-like structure of orga-
nizations reflects the diminishing need for detailed communica-
tion when division and specialization of labor are applied.
In fact, a tree organization really arises as a structure of au-
thority and responsibility. The principle that no man can serve
two masters dictates that the authority structure be tree-like. But
the communication structure is not so restricted, and the tree is a
barely passable approximation to the communication structure,
which is a network. The inadequacies of the tree approximation
give rise to staff groups, task forces, committees, and even the
matrix-type organization used in many engineering laboratories.
Let us consider a tree-like programming organization, and
examine the essentials which any subtree must have in order to be
effective. They are:
1. a mission
2. a producer
3. a technical director or architect
4. a schedule
5. a division of labor
6. interface definitions among the parts
All of this is obvious and conventional except the distinction
between the producer and the technical director. Let us first con-
sider the two roles, then their relationship.
What is the role of the producer? He assembles the team,
divides the work, and establishes the schedule. He acquires and
keeps on acquiring the necessary resources. This means that a
major part of his role is communication outside the team, upwards
and sideways. He establishes the pattern of communication and
reporting within the team. Finally, he ensures that the schedule is
met, shifting resources and organization to respond to changing
circumstances.
80 Why Did The Tower of Babel Fail?
How about the technical director? He conceives of the design
to be built, identifies its subparts, specifies how it will look from
outside, and sketches its internal structure. He provides unity and
conceptual integrity to the whole design; thus he serves as a limit
on system complexity. As individual technical problems arise, he
invents solutions for them or shifts the system design as required.
He is, in Al Capp's lovely phrase, "inside-man at the skunk
works." His communications are chiefly within the team. His
work is almost completely technical.
Now it is clear that the talents required for these two roles are
quite different. Talents come in many different combinations; and
the particular combination embodied in the producer and the di-
rector must govern the relationship between them. Organizations
must be designed around the people available; not people fitted
into pure-theory organizations.
Three relationships are possible, and all three are found in
successful practice.
The producer and the technical director may be the same man.
This is readily workable on very small teams, perhaps three to six
programmers. On larger projects it is very rarely workable, for two
reasons. First, the man with strong management talent and strong
technical talent is rarely found. Thinkers are rare; doers are rarer;
and thinker-doers are rarest.
Second, on the larger project each of the roles is necessarily a
full-time job, or more. It is hard for the producer to delegate
enough of his duties to give him any technical time. It is impossi-
ble for the director to delegate his without compromising the
conceptual integrity of the design.
The producer may be boss, the director his right-hand man.
The difficulty here is to establish the director's authority to make
technical decisions without impacting his time as would putting
him in the management chain-of-command.
Obviously the producer must proclaim the director's technical
authority, and he must back it in an extremely high proportion of
Organization in the Large Programming Project 81
the test cases that will arise. For this to be possible, the producer
and the director must see alike on fundamental technical philoso-
phy; they must talk out the main technical issues privately, before
they really become timely; and the producer must have a high
respect for the director's technical prowess.
Less obviously, the producer can do all sorts of subtle things
with the symbols of status (office size, carpet, furnishing, carbon
copies, etc.) to proclaim that the director, although outside the
management line, is a source of decision power.
This can be made to work very effectively. Unfortunately it
is rarely tried. The job done least well by project managers is to
utilize the technical genius who is not strong on management
talent.
The director may be boss, and the producer his right-hand man.
Robert Heinlein, in The Man Who Sold the Moon, describes such
an arrangement in a graphic for-instance:
Coster buried his face in his hands, then looked up. "I know it. I know
what needs to be done—but every time I try to tackle a technical
problem some bloody fool wants me to make a decision about trucks
—or telephones—or some damn thing. I'm sorry, Mr. Harriman.
I thought I could do it. "
Harriman said very gently, "Don't let it throw you. Bob. You
haven 7 had much sleep lately, have you? Tell you what—we 11 put
over a fast one on Ferguson. I'll take that desk you 're at for a few
days and build you a set-up to protect you against such things. I want
that brain of yours thinking about reaction vectors and fuel efficiencies
and design stresses, not about contracts for trucks. "Harriman stepped
to the door, looked around the outer office and spotted a, man who might
or might not be the office's chief clerk. "Hey you! C'mere. "
The man looked startled, got up, came to the door and said, "Yes?"
"I want that desk in the corner and all the stuff that's on it moved
to an empty office on this floor, right away. "
82 Why Did The Tower of Babel FaU?
He supervised getting Coster and his other desk moved into another
office, saw to it that the phone in the new office was disconnected, and,
as an afterthought, had a couch moved in there, too. "We'll install
a projector, and a drafting machine and bookcases and other junk like
that tonight, " he told Coster. "Just make a list of anything you need
—to work on engineering." He went back to the nominal chief-
engineer's office and got happily to work trying to figure where the
organization stood and what was wrong with it.
Some four hours later he took Berkeley in to meet Coster. The chief
engineer was asleep at his desk, head cradled on his arms. Harriman
started to back out, but Coster roused. "Oh! Sorry, " he said, blush-
ing, "I must have dozed off. "
"That's why I brought you the couch, " said Harriman. "It's more
restful. Bob, meet Jock Berkeley. He's your new slave. You remain
chief engineer and top, undisputed boss. Jock is Lord High Everything
Else. From now on you 've got absolutely nothing to worry about—
except for the little detail of building a Moon ship. "
They shook hands. "Just one thing I ask, Mr. Coster, "Berkeley said
seriously, "bypass me all you want to—you'll have to run the
technical show—but for God's sake record it so I'll know what's going
on. I'm going to have a switch placed on your desk that will operate
a sealed recorder at my desk. "
"Fine!" Coster was looking, Harriman thought, younger already.
"And if you want something that is not technical, don't do it yourself.
Just flip a switch and whistle; it'll get done!" Berkeley glanced at
Harriman. "The Boss says he wants to talk with you about the real
job. I'll leave you and get busy. "He left.
Harriman sat down; Coster followed suit and said, "Whew!"
"Feel better?"
"I like the looks of that fellow Berkeley. "
Organization in the Large Programming Project 83
"That's good; he's your twin brother from now on. Stop worrying;
I've used him before. You 'II think you 're living in a well-run hospi-
tal. "*
This account hardly needs any analytic commentary. This
arrangement, too, can be made to work effectively.
I suspect that the last arrangement is best for small teams, as
discussed in Chapter 3, "The Surgical Team." I think the producer
as boss is a more suitable arrangement for the larger subtrees of
a really big project.
The Tower of Babel was perhaps the first engineering fiasco,
but it was not the last. Communication and its consequent, orga-
nization, are critical for success. The techniques of communication
and organization demand from the manager much thought and as
much experienced competence as the software technology itself.
8
Calling the Shot
8
Calling the Shot
Practice is the best of all instructors.
PUBUUUS
Experience is a dear teacher, but fools will learn at no
other.
POOR RICHARD'S ALMANAC
Douglass Crockwell, "Ruth calls his shot," World Series, 1932
Reproduced by permission of Esquire Magazine and Douglass Crockwell, © 1945
(renewed 1973) by Esquire, Inc., and courtesy of the National Baseball Museum.
87
88 Calling the Shot
How long will a system programming job take? How much effort
will be required? How does one estimate?
I have earlier suggested ratios that seem to apply to planning
time, coding, component test, and system test. First, one must say
that one does not estimate the entire task by estimating the coding
portion only and then applying the ratios. The coding is only
one-sixth or so of the problem, and errors in its estimate or in the
ratios could lead to ridiculous'results.
Second, one must say that data for building isolated small
programs are not applicable to programming systems products. For
a program averaging about 3200 words, for example, Sackman,
Erikson, and Grant report an average code-plus-debug time of
about 178 hours for a single programmer, a figure which would
extrapolate to give an annual productivity of 35,800 statements
per year. A program half that size took less than one-fourth as
long, and extrapolated productivity is almost 80,000 statements
per year.1 Planning, documentation, testing, system integration,
and training times must be added. The linear extrapolation of such
sprint figures is meaningless. Extrapolation of times for the hun-
dred-yard dash shows that a man can run a mile in under three
minutes.
Before dismissing them, however, let us note that these num-
bers, although not for strictly comparable problems, suggest that
effort goes as a power of size even when no communication is
involved except that of a man with his memories.
Figure 8.1 tells the sad story. It illustrates results reported from
a study done by Nanus and Farr2 at System Development Corpo-
ration. This shows an exponent of 1.5; that is,
effort = (constant) X (number of instructions)15.
Another SDC study reported by Weinwurm3 also shows an expo-
nent near 1.5.
A few studies on programmer productivity have been made,
and several estimating techniques have been proposed. Morin has
prepared a survey of the published data.4 Here I shall give only a
few items that seem especially illuminating.
Portman's Data 89
Thousands of machine instructions
Fig. 8.1 Programming effort as a function of program size
Portman's Data
Charles Portman, manager of ICL's Software Division, Computer
Equipment Organization (Northwest) at Manchester, offers an-
other useful personal insight.5
He found his programming teams missing schedules by about
one-half—each job was taking approximately twice as long as
estimated. The estimates were very careful, done by experienced
teams estimating man-hours for several hundred subtasks on a
PERT chart. When the slippage pattern appeared, he asked them
to keep careful daily logs of time usage. These showed that the
estimating error could be entirely accounted for by the fact that
his teams were only realizing 50 percent of the working week as
actual programming and debugging time. Machine downtime,
higher-priority short unrelated jobs, meetings, paperwork, com-
90 Calling the Shot
pany business, sickness, personal time, etc. accounted for the rest.
In short, the estimates made an unrealistic assumption about the
number of technical work hours per man-year. My own experi-
ence quite confirms his conclusion.6
Aron's Data
Joel Aron, manager of Systems Technology at IBM in Gaithers-
burg, Maryland, has studied programmer productivity when
working on nine large systems (briefly, large means more than 25
programmers and 30,000 deliverable instructions).7 He divides
such systems according to interactions among programmers (and
system parts) and finds productivities as follows:
Very few interactions 10,000 instructions per man-year
Some interactions 5,000
Many interactions 1,500
The man-years do not include support and system test activi-
ties, only design and programming. When these figures are diluted
by a factor of two to cover system test, they closely match Harr's
data.
Harr's Data
John Hair, manager of programming for the Bell Telephone Labo-
ratories' Electronic Switching System, reported his and others'
experience in a paper at the 1969 Spring Joint Computer Confer-
ence.8 These data are shown in Figs. 8.2, 8.3, and 8.4.
Of these, Fig. 8.2 is the most detailed and the most useful. The
first two jobs are basically control programs; the second two are
basically language translators. Productivity is stated in terms of
debugged words per man-year. This includes programming, com-
ponent test, and system test. It is not clear how much of the
planning effort, or effort in machine support, writing, and the like,
is included.
Han's Data 91
Number of Maa- Program Words/'
units programmers Years years words man-yf
Operational 50 S3 4 101 52,OQG 515
Maintenance 36 60 4 81 51,000 630
Compiler 13 9 •2'A 1? 38,000
Translator 15 13 2% 11 25,000 2270
(Data assembler)
Fig. 8.2 Summary of four No. 1 ESS program jobs
The productivities likewise fall into two classifications; those
for control programs are about 600 words per man-year; those for
translators are about 2200 words per man-year. Note that all four
programs are of similar size—the variation is in size of the work
groups, length of time, and number of modules. Which is cause
and which is effect? Did the control programs require more people
because they were more complicated? Or did they require more
modules and more man-months because they were assigned more
people? Did they take longer because of the greater complexity,
or because more people were assigned? One can't be sure. The
control programs were surely more complex. These uncertainties
aside, the numbers describe the real productivities achieved on a
large system, using present-day programming techniques. As such
they are a real contribution.
Figures 8.3 and 8.4 show some interesting data on program-
ming and debugging rates as-compared to predicted rates.
92 Calling the Shot
Fig. 8.3 ESS predicted and actual programming rates
Fig. 8.4 ESS predicted and actual debugging rates
Corbatd's Data 93
OS/360 Data
IBM OS/360 experience, while not available in the detail of Hair's
data, confirms it. Productivities in range of 600-800 debugged
instructions per man-year were experienced by control program
groups. Productivities in the 2000-3000 debugged instructions per
man-year were achieved by language translator groups. These
include planning done by the group, coding component test, sys-
tem test, and some support activities. They are comparable to
Han's data, so far as I can tell.
Aron's data, Hair's data, and the OS/360 data all confirm
striking differences in productivity related to the complexity and
difficulty of the task itself. My guideline in the morass of estimat-
ing complexity is that compilers are three times as bad as normal
batch application programs, and operating systems are three times
as bad as compilers.9
Corbato's Data
Both Hair's data and OS/360 data are for assembly language pro-
gramming. Little data seem to have been published on system
programming productivity using higher-level languages. Corbato
of MIT's Project MAC reports, however, a mean productivity of
1200 lines of debugged PL/I statements per man-year on the
MULTICS system (between 1 and 2 million words).10
This number is very exciting. Like the other projects, MUL-
TICS includes control programs and language translators. Like the
others, it is producing a system programming product, tested and
documented. The data seem to be comparable in terms of kind of
effort included. And the productivity number is a good average
between the control program and translator productivities of other
projects.
But Corbato's number is lines per man-year, not wordsl Each
statement in his system corresponds to about three to five words
of handwritten code! This suggests two important conclusions.
94 Calling the Shot
• Productivity seems constant in tenns of elementary state-
ments, a conclusion that is reasonable in terms of the thought
a statement requires and the errors it may include."
• Programming productivity may be increased as much as five
times when a suitable high-level language is used.18
9
Ten Pounds
in a Five-Pound Sack
9
Ten Pounds
in a Five-Pound Sack
The author should gaze at Noah, and . . . learn, as they
did in the Ark, to crowd a great deal of matter into a very
small compass.
SYDNEY SMITH. EDINBURGH REVIEW
Engraved_ from a painting by Heywood Hardy
The Bettman Archive
97
98 Ten Pounds in a Five-Pound Sack
Program Space as Cost
How big is it? Aside from running time, the space occupied by a
program is a principal cost. This is true even for proprietary pro-
grams, where the user pays the author a fee that is essentially a
share of the development cost. Consider the IBM APL interactive
software system. It rents for $400 per month and, when used, takes
at least 160 K bytes of memory. On a Model 165, memory rents
for about $12 per kilobyte per month. If the program is available
full-time, one pays $400 software rent and $1920 memory rent for
using the program. If one uses the APL system only four hours a
day, the costs are $400 software rent and $320 memory rent per
month.
One frequently hears horror expressed that a 2 M byte ma-
chine may have 400 K devoted to its operating system. This is as
foolish as criticizing a Boeing 747 because it costs $27 million. One
must also ask, "What does it do?" What does one get in ease-of-
use and in performance (via efficient system utilization) for the
dollars so spent? Could the $4800 per month thus invested in
memory rental have been more fruitfully spent for other hard-
ware, for programmers, for application programs?
The system designer puts part of his total hardware resource
into resident-program memory when he thinks it will do more for
the user in that form than as adders, disks, etc. To do otherwise
would be grossly irresponsible. And the result must be judged as
a whole. No one can criticize a programming system for size per se
and at the same time consistently advocate closer integration of
hardware and software design.
Since size is such a large part of the user cost of a programming
system product, the builder must set size targets, control size, and
devise size-reduction techniques, just as the hardware builder sets
component-count targets, controls component count, and devises
count-reduction techniques. Like any cost, size itself is not bad,
but unnecessary size is.
Size Control 99
Size Control
For the project manager, size control is partly a technical job and
partly a managerial one. One has to study users and their applica-
tions to set the sizes of the systems to be offered. Then these
systems have to be subdivided, and each component given a size
target. Since size-speed trade-offs come in rather big quantum
jumps, setting size targets is a tricky business, requiring knowl-
edge of the available trade-offs within each piece. The wise man-
ager also saves himself a kitty, to be allocated as work proceeds.
In OS/360, even though all of this was done very carefully,
still other lessons had to be painfully learned.
First, setting size targets for core is not enough; one has to
budget all aspects of size. In most previous operating systems,
systems residence had been on tape, and the long search times of
tape meant that one was not tempted to use it casually to bring
in program segments. OS/360 was disk-resident, like its immedi-
ate predecessors, the Stretch Operating System and the 1410-7010
Disk Operating System. Its builders rejoiced in the freedom of
cheap disk accesses. The initial result was disastrous to perfor-
mance.
In setting core sizes for each component, we had not simulta-
neously set access budgets. As anyone with 20-20 hindsight
would expect, a programmer who found his program slopping
over his core target broke it into overlays. This process in itself
added to the total size and slowed execution down. Most seri-
ously, our management control system neither measured nor
caught this. Each man reported as to how much core he was using,
and since he was within target, no one worried.
Fortunately, there came a day early in the effort when the
OS/360 performance simulator began to work. The first result
indicated deep trouble. Fortran H, on a Model 65 with drums,
simulated compiling at five statements per minute! Digging-in
showed that the control program modules were each making
100 Ten Pounds in a Five-Pound Sack
many, many disk accesses. Even high-frequency supervisor
modules were making many trips to the well, and the result was
quite analogous to page thrashing.
The first moral is clear: Set total size budgets as well as resi-
dent-space budgets; set budgets on backing-store accesses as well
as on sizes.
The next lesson was very similar. The space budgets were set
before precise functional allocations were made to each module.
As a result, any programmer in size trouble examined his code to
see what he could throw over the fence into a neighbor's space.
So buffers managed by the control program became part of the
user's space. More seriously, so did all kinds of control blocks, and
the effect was utterly compromising to the security and protection
of the system.
So the second moral is also clear: Define exactly what a
module must do when you specify how big it must be.
A third and deeper lesson shows through these experiences.
The project was large enough and management communication
poor enough to prompt many members of the team to see them-
selves as contestants making brownie points, rather than as build-
ers making programming products. Each suboptimized his piece to
meet his targets; few stopped to think about the total effect on the
customer. This breakdown in orientation and communication is a
major hazard for large projects. All during implementation, the
system architects must maintain continual vigilance to ensure con-
tinued system integrity. Beyond this policing mechanism, how-
ever, lies the matter of attitude of the implementers themselves.
Fostering a total-system, user-oriented attitude may well be the
most important function of the programming manager.
Space Techniques
No amount of space budgeting and control can make a program
small. That requires invention and craftsmanship.
Space Techniques 101
Obviously, more function means more space, speed being held
constant. So the first area of craftsmanship is in trading function
for size. Here there comes an early and deep policy question. How
much of that choice shall be reserved for the user? One can design
a program with many optional features, each of which takes a little
space. One can design a generator that will take an option Jist and
tailor a program to it. But for any particular set of options, a more
monolithic program would take less space. It's rather like a car; if
the map light, cigarette lighter, and clock are priced together as a
single option, the package will cost less than if one can choose each
separately. So the designer must decide how fine-grained the user
choice of options will be.
In designing a system for a range of memory sizes, another
basic question arises. A limiting effect keeps the range of suitabil-
ity from being made arbitrarily wide, even with fine-grained
modularity of function. In the smallest system, most modules will
be overlaid. A substantial part of the smallest system's resident
space must be set aside as a transient or paging area into which
other parts are fetched. The size of this determines the size of all
modules. And breaking functions into small modules costs both
performance and space. So a large system, which can afford a
transient area twenty times as large, only saves accesses thereby.
It still suffers in both speed and space because the module size is
so small. This effect limits the maximum efficient system that can
be generated from the modules of a small system.
The second area of craftsmanship is space-time trade-offs. For
a given function, the more space, the faster. This is true over an
amazingly large range. It is this fact that makes it feasible to set
space budgets.
The manager can do two things to help his team make good
space-time trade-offs. One is to ensure that they are trained in
programming technique, not just left to rely on native wit and
previous experience. For a new language or machine this is espe-
cially important. The peculiarities of its skillful use need to be
102 Ten Pounds in a Five-Pound Sack
learned quickly and shared widely, perhaps with special prizes or
praises for new techniques.
The second is to recognize that programming has a technol-
ogy, and components need to be fabricated. Every project needs
a notebook full of good subroutines or macros for queuing, search-
ing, hashing, and sorting. For each such function the notebook
should have at least two programs, the quick and the squeezed.
The development of such technology is an important realization
task that can be done in parallel with system architecture.
Representation Is the Essence of Programming
Beyond craftsmanship lies invention, and it is here that lean,
spare, fast programs are born. Almost always these are the result
of stategic breakthrough rather than tactical cleverness. Some-
times the strategic breakthrough will be a new algorithm, such as
the Cooley-Tukey Fast Fourier Transform or the substitution of
an n log n sort for an n* set of comparisons.
Much more often, strategic breakthrough will come from
redoing the representation of the data or tables. This is where the
heart of a program lies. Show me your flowcharts and conceal your
tables, and I shall continue to be mystified. Show me your tables,
and I won't usually need your flowcharts; they'll be obvious.
It is easy to multiply examples of the power of representa-
tions. I recall a young man undertaking to build an elaborate
console interpreter for an IBM 650. He ended up packing it onto
an incredibly small amount of space by building an interpreter for
the interpreter, recognizing that human interactions are slow and
infrequent, but space was dear. Digitek's elegant little Fortran
compiler uses a very dense, specialized representation for the com-
piler code itself, so that external storage is not needed. That time
lost in decoding this representation is gained back tenfold by
avoiding input-output. (The exercises at the end of Chapter 6 in
Brooks and Iverson, Automatic Data Processing1 include a collection
of such examples, as do many of Knuth's exercises.2)
Representation Is the Essence of Programming 103
The programmer at wit's end for lack of space can often do
best by disentangling himself from his code, rearing back, and
contemplating his data. Representation is the essence of program-
ming.
10
The Documentary
Hypothesis
The hypothesis:
Amid a wash of paper, a small number of documents
become the critical pivots around which every project's
management revolves. These are the manager's chief
personal tools.
W. Bengough, "Scene in the old Congressional Library," 1897
The Bettman Archive
107
108 The Documentary Hypothesis
The technology, the surrounding organization, and the traditions
of the craft conspire to define certain items of paperwork that a
project must prepare. To the new manager, fresh from operating
as a craftsman himself, these seem an unmitigated nuisance, an
unnecessary distraction, and a white tide that threatens to engulf
him. And indeed, most of them are exactly that.
Bit by bit, however, he comes to realize that a certain small set
of these documents embodies and expresses much of his manage-
rial work. The preparation of each one serves as a major occasion
for focusing thought and crystallizing discussions that otherwise
would wander endlessly. Its maintenance becomes his surveillance
and warning mechanism. The document itself serves as a check
list, a status control, and a data base for his reporting.
In order to see how this should work for a software project,
let us examine the specific documents useful in other contexts and
see if a generalization emerges.
Documents for a Computer Product
Suppose one is building a machine. What are the critical docu-
ments?
Objectives. This defines the need to be met and the goals,
desiderata, constraints, and priorities.
Specifications. This is a computer manual plus performance
specifications. It is one of the first documents generated in propos-
ing a new product, and the last document finished.
Schedule
Budget. Not merely a constraint, the budget is one of the mana-
ger's most useful documents. Existence of the budget forces tech-
nical decisions that otherwise would be avoided; and, more
important, it forces and clarifies policy decisions.
Organization chart
Space allocations
Documents for a University Department 109
Estimate, forecast, prices. These three have cyclic Interlocking,
which determines the success or failure of the project:
To generate a market forecast, one needs performance specifi-
cations and postulated prices. The quantities from the forecast
combine with component counts from the design to determine the
manufacturing cost estimate, and they determine the per-unit
share of development and fixed costs. These costs in turn deter-
mine prices.
If the prices are below those postulated, a joyous success spiral
begins. Forecasts rise, unit costs drop, and prices drop yet further.
If the prices are above those postulated, a disastrous spiral
begins, and all hands must struggle to break it. Performance must
be squeezed up and new applications developed to support larger
forecasts. Costs must be squeezed down to yield lower estimates.
The stress of this cycle is a discipline that often evokes the best
work of marketer and engineer.
It can also bring about ridiculous vacillation. I recall a machine
whose instruction counter popped in or out of memory every six
months during a three-year development cycle. At one phase a
little more performance would be wanted, so the instruction coun-
ter was implemented in transistors. At the next phase cost reduc-
tion was the theme, so the counter would be implemented as a
memory location. On another project the best engineering man-
ager I ever saw served often as a giant flywheel, his inertia damp-
ing the fluctuations that came from market and management
people.
Documents for a University Department
In spite of the immense differences in purpose and activity, a
similar number of similar documents form the critical set for the
110 The Documentary Hypothesis
chairman of a university department. Almost every decision by
dean, faculty meeting, or chairman is a specification or change of
these documents:
Objectives
Course descriptions
Degree requirements
Research proposals (hence plans, when funded)
Class schedule and teaching assignments
Budget
Space allocation
Assignment of staff and graduate students
Notice that the components are very like those of the com-
puter project: objectives, product specifications, time allocations,
money allocations, space allocations, and people allocations. Only
the pricing documents are missing; here the legislature does that
task. The similarities are not accidental—the concerns of any man-
agement task are what, when, how much, where, and who.
Documents for a Software Project
In many software projects, people begin by holding meetings to
debate structure; then they start writing programs. No matter how
small the project, however, the manager is wise to begin immedi-
ately to formalize at least mini-documents to serve as his data
base. And he turns out to need documents much like those of other
managers.
What: objectives. This defines the need to be met and the goals,
desiderata, constraints, and priorities.
What: product specifications. This begins as a proposal and
ends up as the manual and internal documentation. Speed and
space specifications are a critical part.
Why Have Formal Documents? 111
When: schedule
How much: budget
Where: space allocation
Who: organization chart. This becomes intertwined with the
interface specification, as Conway's Law predicts: "Organizations
which design systems are constrained to produce systems which
are copies of the communication structures of these organiza-
tions."1 Conway goes on to point out that the organization chart
will initially reflect the first system design, which is almost surely
not the right one. If the system design is to be free to change, the
organization must be prepared to change.
Why Have Formal Documents?
First, writing the decisions down is essential. Only when one
writes do the gaps appear and the inconsistencies protrude. The act
of writing turns out to require hundreds of mini-decisions, and it
is the existence of these that distinguishes clear, exact policies
from fuzzy
Second, the documents will communicate the decisions to oth-
ers. The manager will be continually amazed that policies he took
for common knowledge are totally unknown by some member of
his team. Since his fundamental job is to keep everybody going in
the same direction, his chief daily task will be communication, not
decision-making, and his documents will immensely lighten this
load.
Finally, a manager's documents give him a data base and
checklist. By reviewing them periodically he sees where he is, and
he sees what changes of emphasis or shifts in direction are needed.
I do not share the salesman-projected vision of the "manage-
ment total-information system," wherein the executive strokes an
inquiry into a computer, and a display screen flashes his answer.
There are many fundamental reasons why this will never happen.
112 The Documentary Hypothesis
One reason is that only a small part—perhaps 20 percent—of the
executive's time is spent on tasks where he needs information
from outside his head. The rest is communication: hearing, report-
ing, teaching, exhorting, counseling, encouraging. But for the frac-
tion that is data-based, the handful of critical documents are vital,
and they will meet almost all needs.
The task of the manager is to develop a plan and then to realize
it. But only the written plan is precise and communicable. Such a
plan consists of documents on what, when, how much, where, and
who. This small set of critical documents encapsulates much of the
manager's work. If their comprehensive and critical nature is rec-
ognized in the beginning, the manager can approach them as
friendly tools rather than annoying busywork. He will set his
direction much more crisply and quickly by doing so.
11
Plan toThrow
One Away
11
Plan toThrow
OneAway
There is nothing in this world constant but inconstancy.
SWIFT
It is common sense to take a method and try it. If it fails,
admit it frankly and try another. But above all try
FRANKLIN D. ROOSEVELT
Collapse of the aerodynamically misdesigned Tacoma Narrows Bridge,
1940
UPI Photo/The Bettman Archive
115
116 Plan to Throw One Away
Pilot Plants and Scaling Up
Chemical engineers learned long ago that a process that works in
the laboratory cannot be implemented in a factory in only one
step. An intermediate step called the pilot plant is necessary to give
experience in scaling quantities up and in operating in nonprotec-
tive environments. For example, a laboratory process for desalting
water will be tested in a pilot plant of 10,000 gallon/day capacity
before being used for a 2,000,000 gallon/day community water
system.
Programming system builders have also been exposed to this
lesson, but it seems to have not yet been learned. Project after
project designs a set of algorithms and then plunges into construc-
tion of customer-deliverable software on a schedule that demands
delivery of the first thing built.
In most projects, the first system built is barely usable. It may
be too slow, too big, awkward to use, or all three. There is no
alternative but to start again, smarting but smarter, and build a
redesigned version in which these problems are solved. The dis-
card and redesign may be done in one lump, or it may be done
piece-by-piece. But all large-system experience shows that it will
be done.2 Where a new system concept or new technology is used,
one has to build a system to throw away, for even the best plan-
ning is not so omniscient as to get it right the first time.
The management question, therefore, is not whether to build
a pilot system and throw it away. You will do that. The only
question is whether to plan in advance to build a throwaway, or
to promise to deliver the throwaway to customers. Seen this way,
the answer is much clearer. Delivering that throwaway to custom-
ers buys time, but it does so only at the cost of agony for the user,
distraction for the builders while they do the redesign, and a bad
reputation for the product that the best redesign will find hard to
live down.
Hence plan to throw one away; you will, anyhow.
Plan the System for Change 117
The Only Constancy Is Change Itself
Once one recognizes that a pilot system must be built and dis-
carded, and that a redesign with changed ideas is inevitable, it
becomes useful to face the whole phenomenon of change. The first
step is to accept the fact of change as a way of life, rather than an
untoward and annoying exception. Cosgrove has perceptively
pointed out that the programmer delivers satisfaction of a user
need rather than any tangible product. And both the actual need
and the user's perception of that need will change as programs are
built, tested, and used.3
Of course this is also true of the needs met by hardware
products, whether new cars or new computers. But the very exis-
tence of a tangible object serves to contain and quantize user
demand for changes. Both the tractability and the invisibility of
the software product expose its builders to perpetual changes in
requirements.
Far be it from me to suggest that all changes in customer
objectives and requirements must, can, or should be incorporated
in the design. Clearly a threshold has to be established, and it must
get higher and higher as development proceeds, or no product ever
appears.
Nevertheless, some changes in objectives are inevitable, and it
is better to be prepared for them than to assume that they won't
come. Not only are changes in objective inevitable, changes in
development strategy and technique are also inevitable. The
throw-one-away concept is itself just an acceptance of the fact
that as one learns, he changes the design.4
Plan the System for Change
The ways of designing a system for such change are well known
and widely discussed in the literature—perhaps more widely dis-
118 Plan to Throw One Away
cussed than practiced. They include careful modularization, ex-
tensive subroutining, precise and complete definition of
intermodule interfaces, and complete documentation of these.
Less obviously one wants standard calling sequences and table-
driven techniques used wherever possible.
Most important is the use of a high-level language and self-
documenting techniques so as to reduce errors induced by
changes. Using compile-time operations to incorporate standard
declarations helps powerfully in making changes.
Quantization of change is an essential technique. Every prod-
uct should have numbered versions, and each version must have
its own schedule and a freeze date, after which changes go into the
next version.
Plan the Organization for Change
Cosgrove advocates treating all plans, milestones, and schedules as
tentative, so as to facilitate change. This goes much too far—the
common failing of programming groups today is too little manage-
ment control, not too much.
Nevertheless, he offers a great insight. He observes that the
reluctance to document designs is not due merely to laziness or
time pressure. Instead it comes from the designer's reluctance to
commit himself to the defense of decisions which he knows to be
tentative. "By documenting a design, the designer exposes himself
to the criticisms of everyone, and he must be able to defend
everything he writes. If the organizational structure is threatening
in any way, nothing is going to be documented until it is com-
pletely defensible."
Structuring an organization for change is much harder than
designing a system for change. Each man must be assigned to jobs
that broaden him, so that the whole force is technically flexible.
On a large project the manager needs to keep two or three top
programmers as a technical cavalry that can gallop to the rescue
wherever the battle is thickest.
Plan the Organization for Change 119
Management structures also need to be changed as the system
changes. This means that the boss must give a great deal of atten-
tion to keeping his managers and his technical people as inter-
changeable as their talents allow.
The barriers are sociological, and they must be fought with
constant vigilance. First, managers themselves often think of se-
nior people as "too valuable" to use for actual programming. Next,
management jobs carry higher prestige. To overcome this problem
some laboratories, such as Bell Labs, abolish all job titles. Each
professional employee is a "member of the technical staff." Oth-
ers, like IBM, maintain a dual ladder of advancement, as Fig. 11.1
shows. The corresponding rungs are in theory equivalent.
Senior Associate Programmer
Fig. 11.1 IBM dual ladder of advancement
It is easy to establish corresponding salary scales for rungs. It
is much harder to give them corresponding prestige. Offices have
to be of equal size and appointment. Secretarial and other support
services must correspond. A reassignment from the technical lad-
der to a corresponding level on the managerial one should never
be accompanied by a raise, and it should be announced always as
122 Plan to Throw One Away
The fundamental problem with program maintenance is that
fixing a defect has a substantial (20-50 percent) chance of intro-
ducing another. So the whole process is two steps forward and one
step back.
Why aren't defects fixed more cleanly? First, even a subtle
defect shows itself as a local failure of some kind. In fact it often
has system-wide ramifications, usually nonobvious. Any attempt
to fix it with minimum effort will repair the local and obvious, but
unless the structure is pure or the documentation very fine, the
far-reaching effects of the repair will be overlooked. Second, the
repairer is usually not the man who wrote the code, and often he
is a junior programmer or trainee.
As a consequence of the introduction of new bugs, program
maintenance requires far more system testing per statement writ-
ten than any other programming. Theoretically, after each fix one
must run the entire bank of test cases previously run against the
system, to ensure that it has not been damaged in an obscure way.
In practice such regression testing must indeed approximate this
theoretical ideal, and it is very costly.
Clearly, methods of designing programs so as to eliminate or
at least illuminate side effects can have an immense payoff in
maintenance costs. So can methods of implementing designs with
fewer people, fewer interfaces, and hence fewer bugs.
One Step Forward and One Step Back
Lehman and Belady have studied the history of successive releases
in a large operating system.8 They find that the total number of
modules increases linearly with release number, but that the num-
ber of modules affected increases exponentially with release num-
ber. All repairs tend to destroy the structure, to increase the
entropy and disorder of the system. Less and less effort is spent
on fixing original design flaws; more and more is spent on fixing
flaws introduced by earlier fixes. As time passes, the system
becomes less and less well-ordered. Sooner or later the fixing
One Step Forward and One Step Back 123
ceases to gain any ground. Each forward step is matched by a
backward one. Although in principle usable forever, the system
has worn out as a base for progress. Furthermore, machines
change, configurations change, and user requirements change, so
the system is not in fact usable forever. A brand-new, from-the-
ground-up redesign is necessary.
And so from a statistical mechanical model, Belady and Leh-
man arrive for programming-systems at a more general conclusion
supported by the experience of all the earth. "Things are always
at their best in the beginning," said Pascal. C. S. Lewis has stated
it more perceptively:
That is the key to history. Terrific energy is expended—civilizations
are built up—excellent institutions devised; but each time something
goes wrong. Some fatal flaw always brings the selfish and cruel people
to the top, and then it all slides back into misery and ruin. In fact,
the machine conks. It seems to start up all right and runs a few yards,
and then it breaks down."1
Systems program building is an entropy-decreasing process,
hence inherently metastable. Program maintenance is an entropy-
increasing process, and even its most skillful execution only delays
the subsidence of the system into unfixable obsolescence.
12
Sharp Tools
12
Sharp Tools
A good workman is known by his tools.
PROVERB
A. Pisano, "Lo Scultore," from the Campanile di Santa Maria del
Fiore, Florence, c. 1335
Scala/Art Resource, NY
127
128 Sharp Tools
Even at this late date, many programming projects are still oper-
ated like machine shops so far as tools are concerned. Each master
mechanic has his own personal set, collected over a lifetime and
carefully locked and guarded—the visible evidences of personal
skills. Just so, the programmer keeps little editors, sorts, binary
dumps, disk space utilities, etc., stashed away in his file.
Such an approach, however, is foolish for a programming
project. First, the essential problem is communication, and indi-
vidualized tools hamper rather than aid communication. Second,
the technology changes when one changes machines or working
language, so tool lifetime is short. Finally, it is obviously much
more efficient to have common development and maintenance of
the general-purpose programming tools.
General-purpose tools are not enough, however. Both special-
ized needs and personal preferences dictate the need for special-
ized tools as well; so in discussing programming teams I have
postulated one toolmaker per team. This man masters all the com-
mon tools and is able to instruct his client-boss in their use. He
also builds the specialized tools his boss needs.
The manager of a project, then, needs to establish a philoso-
phy and set aside resources for the building of common tools. At
the same time he must recognize the need for specialized tools, and
not begrudge his working teams their own tool-building. This
temptation is insidious. One feels that if all those scattered tool
builders were gathered in to augment the common tool team,
greater efficiency would result. But it is not so.
What are the tools about which the manager must philoso-
phize, plan, and organize? First, a computer facility. This requires
machines, and a scheduling philosophy must be adopted. It re-
quires an operating system, and service philosophies must be estab-
lished. It requires language, and a language policy must be laid
down. Then there are utilities, -debugging aids, test-case generators,
and a text-processing system to handle documentation. Let us look
at these one by one.1
Target Machines 129
Target Machines
Machine support is usefully divided into the target machine and the
vehicle machines. The target machine is the one for which software
is being written, and on which it must ultimately be tested. The
vehicle machines are those that provide the services used in build-
ing the system. If one is building a new operating system for an
old machine, it may serve not only as the target, but as the vehicle
as well.
What kind of target facility? Teams building new supervisors or
other system-heart software will of course need machines of their
own. Such systems will need operators and a system programmer
or two who keeps the standard support on the machine current
and serviceable.
If a separate machine is needed, it is a rather peculiar thing—
it need not be fast, but it needs at least a million bytes of main
storage, a hundred million bytes of on-line disk, and terminals.
Only alphanumeric terminals are needed, but they must go much
faster than the 15 characters per second that characterizes type-
writers. A large memory adds greatly to productivity by allowing
overlaying and size trimming to be done after functional testing.
The debugging machine, or its software, also needs to be in-
strumented, so that counts and measurements of all kinds of pro-
gram parameters can be automatically made during debugging.
Memory-use patterns, for instance, are powerful diagnostics of
the causes of weird logical behavior or unexpectedly slow perfor-
mance.
Scheduling. When the target machine is rtew, as when its first
operating system is being built, machine time is scarce, and sched-
uling it is a major problem. The requirement for target machine
time has a peculiar growth curve. In OS/360 development we had
good System/360 simulators and other vehicles. From previous
experience we projected how many hours of S/360 time we would
need, and began to acquire early machines from factory produc-
130 Sharp Tools
tion. But they sat idle, month after month. Then all at once all 16
systems were fully loaded, and rationing was the problem. The
utilization looked something like Fig. 12.1. Everyone began to
debug his first components at the same time, and thereafter most
of the team was constantly debugging something.
Model 40 hours
per month
Jan '65 '66
Fig. 12.1 Growth in use of target machines
We centralized all our machines and tape library and set up a
professional and experienced machine-room team to run them. To
maximize scarce S/360 time, we ran all debugging runs in batch
on whichever system was free and appropriate. We tried for four
shots per day (two-and-one-half-hour turnaround) and de-
manded four-hour turnaround. An auxiliary 1401 with terminals
was used to schedule runs, to keep track of the thousands of jobs,
and to monitor turnaround time.
But all that organization was quite overdone. After a few
months of slow turnaround, mutual recriminations, and other
agony, we went to allocating machine time in substantial blocks.
Vehicle Machines and Data Services 131
The whole fifteen-man sort team, for example, would be given a
system for a four-to-six-hour block. It was up to them to schedule
themselves on it. If it sat idle, no outsider could use it.
That, it develops, was a better way to allocate and schedule.
Although machine utilization may have been a little lower (and
often it wasn't), productivity was way up. For each man on such
a team, ten shots in a six-hour block are far more productive than
ten shots spaced three hours apart, because sustained concentra-
tion reduces thinking time. After such a sprint, a team usually
needed a day or two to catch up on the paperwork before asking
for another block. Often as few as three programmers can fruit-
fully share and subschedule a block of time. This seems to be the
best way to use a target machine when debugging a new operating
system.
It has always been so in practice, though never in theory.
System debugging has always been a graveyard-shift occupation,
like astronomy. Twenty years ago, on the 701,1 was initiated into
the productive informality of the predawn hours, when all the
machine-room bosses are fast asleep at home, and the operators
are disinclined to be sticklers for rules. Three machine generations
have passed; technologies have changed totally; operating systems
have arisen; and yet this preferred method of working hasn't
changed. It endures because it is most productive. The time has
come to recognize its productivity and embrace the fruitful prac-
tice openly.
Vehicle Machines and Data Services
Simulators. If the target computer is new, one needs a logical
simulator for it. This gives a debugging vehicle long before the real
target exists. Equally important, it gives access to a dependable
debugging vehicle even after one has a target machine available.
Dependable is not the same as accurate. The simulator will
surely fail in some respect to be a faithful and accurate implemen-
132 Sharp Tools
tation of the new machine's architecture. But it will be the same
implementation from one day to the next, and the new hardware
will not.
We are accustomed nowadays to having computer hardware
work correctly almost all the time. Unless an application program-
mer sees a system behaving inconsistently from run to identical
run, he is well advised to look for bugs in his code rather than in
his engine.
This experience, however, is bad training for the programming
of support for a new machine. Lab-built, preproduction, or early
hardware does not work as defined, does not work reliably, and
does not stay the same from day to day. As bugs are found,
engineering changes are made in all machine copies, including
those of the programming group. This shifting base is bad enough.
Hardware failures, usually intermittent, are worse. The uncer-
tainty is worst of all, for it robs one of incentive to dig diligently
in his code for a bug—it may not be there at all. So a dependable
simulator on a well-aged vehicle retains its usefulness far longer
than one would expect.
Compiler and assembler vehicles. For the same reasons, one
wants compilers and assemblers that run on dependable vehicles
but compile object code for the target system. This can then start
being debugged on the simulator.
With high-level language programming, one can do much of
the debugging by compiling for and testing object code on the
vehicle machine before beginning to test target-machine code at
all. This gives the efficiency of direct execution, rather than that
of simulation, combined with the dependability of the stable ma-
chine.
Program libraries and accounting. A very successful and im-
portant use of a vehicle machine in the OS/360 development effort
was for the maintenance of program libraries. A system developed
under the leadership of W. R. Crowley had two 7010's connected,
sharing a large disk data bank. The 7010's also provided an S/360
Vehicle Machines and Data Services 133
assembler. All the code tested or under test was kept in this li-
brary, both source code and assembled load modules. The library
was in fact divided into sublibraries with different access rules.
First, each group or programmer had an area where he kept
copies of his programs, his test cases, and the scaffolding he
needed for component testing. In this playpen area there were no
restrictions on what a man could do with his own programs; they
were his.
When a man had his component ready for integration into a
larger piece, he passed a copy over to the manager of that larger
system, who put this copy into a system integration Now
the original programmer could not change it, except by permission
of the integration manager. As the system came together, the latter
would proceed with all sorts of system tests, identifying bugs and
getting fixes.
From time to time a system version would be ready for wider
use. Then it would be promoted to the current version sublibrary.
This copy was sacrosanct, touched only to fix crippling bugs. It
was available for use in integration and testing of all new module
versions. A program directory on the 7010 kept track of each
version of each module, its status, its whereabouts, and its
changes.
Two notions are important here. The first is control, the idea
of program copies belonging to managers who alone can authorize
their change. The"second is that of formal separation and progression
from the playpen, to integration, to release.
In my opinion this was one of the best-done things in the
OS/360 effort. It is a piece of management technology that seems
to have been independently developed on several massive pro-
gramming projects including those at Bell Labs, ICL, and Cam-
bridge University.8 It is applicable to documentation as well as to
programs. It is an indispensable technology.
Program tools. As new debugging techniques appear, the old
ones diminish but do not vanish. Thus one needs dumps, source-
file editors, snapshot dumps, even traces.
134 Sharp Tools
Likewise one needs a full set of utilities for putting decks on
disks, making tape copies, printing files, changing catalogs. If one
commissions a project toolmaker early in the process, these can be
done once and can be ready by time they are needed.
Documentation system. Among all tools, the one that saves the
most labor may well be a computerized text-editing system, oper-
ating on a dependable vehicle. We had a very handy one, devised
by J. W. Franklin. Without it I expect OS/360 manuals would have
been far later and more cryptic. There are those who would argue
that the OS/360 six-foot shelf of manuals represents verbal diarr-
hea, that the very voluminosity introduces a new kind of incom-
prehensibility. And there is some truth in that.
But I respond in two ways. First, the OS/360 documentation
is overwhelming in bulk, but the reading plan is carefully laid out;
if one uses it selectively, he can ignore most of the bulk most of
the time. One must consider the OS/360 documentation as a li-
brary or an encyclopedia, not a set of mandatory texts.
Second, this is far preferable to the severe underdocumenta-
tion that characterizes most programming systems. I will quickly
agree, however, that the writing could be vastly improved in some
places, and that the result of better writing would be reduced bulk.
Some parts (e.g., Concepts and Facilities) are very well-written now.
Performance simulator. Better have one. Build it outside-in, as
we will discuss in the next chapter. Use the same top-down design
for the performance simulator, the logical simulator, and the prod-
uct. Start it very early. Listen to it when it speaks.
High-Level Language and Interactive Programming
The most important two tools for system programming today are
two that were not used in OS/360 development almost a decade
ago. They are still not widely used, but all evidence points to their
power and applicability. They are (1) high-level language and
(2) interactive programming. I am convinced that only inertia and
High-Level Language and Interactive Programming 135
sloth prevent the universal adoption of these tools; the technical
difficulties are no longer valid excuses.
High-level language. The chief reasons for using a high-level
language are productivity and debugging speed. We have dis-
cussed productivity earlier (Chapter 8). There is not a lot of nu-
merical evidence, but what there is suggests improvement by
integral factors, not just incremental percentages.
The debugging improvement comes from the fact that there
are fewer bugs, and they are easier to find. There are fewer because
one avoids an entire level of exposure to error, a level on which
one makes not only syntactic errors but semantic ones, such as
misusing registers. The bugs are easier to find because the compiler
diagnostics help find them and, more important, because it is very
easy to insert debugging snapshots.
For me, these productivity and debugging reasons are over-
whelming. I cannot easily conceive of a programming system I
would build in assembly language.
Well, what about the classical objections to such a tool? There
are three: It doesn't let me do what I want. The object code is too
big. The object code is too slow.
As to function, I believe the objection is no longer valid. All
testimony indicates that one can do what he needs to do, but that
it takes work to find out how, and one may occasionally need
unlovely artifices.3'4
As to space, the new optimizing compilers are beginning to be
very satisfactory, and this improvement will continue.
As to speed, optimizing compilers now produce some code
that is faster than most programmer's handwritten code, Further-
more, one can usually solve speed problems by replacing from one
to five percent of a compiler-generated program by handwritten
substitute after the former is fully debugged.5
What high-level language should one use for system program-
ming? The only reasonable candidate today is PL/I.8 It has a very
136 Sharp Tools
full set of functions; it is matched to operating system environ-
ments; and a variety of compilers are available, some interactive,
some fast, some very diagnostic, and some producing highly opti-
mized code. I myself find it faster to work out algorithms in APL;
then I translate these to PL/I for matching to the system environ-
ment.
Interactive programming. One of the justifications for MIT's
Multics project was its usefulness for building programming sys-
tems. Multics (and following it, IBM's TSS) differs in concept from
other interactive computing systems in exactly those respects nec-
essary for systems programming: many levels of sharing and pro-
tection for data and programs, extensive library management, and
facilities for cooperative work among terminal users. I am con-
vinced that interactive systems will never displace batch systems
for many applications. But I think the Multics team has made its
most convincing case in the system-programming application.
There is not yet much evidence available on the true fruitful-
ness of such apparently powerful tools. There is a widespread
recognition that debugging is the hard and slow part of system
programming, and slow turnaround is the bane of debugging. So
the logic of interactive programming seems inexorable.7
Fig. 12.2 Comparative productivity under batch and conversational pro-
gramming
High-Level Language and Interactive Programming 137
Further, we hear good testimonies from many who have built
little systems or parts of systems in this way. The only numbers
I have seen for effects on programming of large systems were
reported by John Harr of Bell Labs. They are shown in Fig. 12.2.
These numbers are for writing, assembling, and debugging pro-
grams. The first program is mostly control program; the other three
are language translators, editors, and such. Hair's data suggest that
an interactive facility at least doubles productivity in system pro-
gramming.8
The effective use of most interactive tools requires that the
work be done in a high-level language, for teletype and typewriter
terminals cannot be used to debug by dumping memory. With a
high-level language, source can be easily edited and selective
printouts easily done. Together they make a pair of sharp tools
indeed.
13
The Whole and the Parts
13
The Whole and the Parts
I can call spirits from the vasty deep.
Why so can /, or so can any man; but will they come
when you do call for them?
SHAKESPEARE. KING HENRY W, PARTI
© The Walt Disney Company
141
142 The Whole and the Parts
The modern magic, like the old, has its boastful practitioners: "I
can write programs that control air traffic, intercept ballistic mis-
siles, reconcile bank accounts, control production lines." To which
the answer comes, "So can I, and so can any man, but do they work
when you do write them?"
How does one build a program to work? How does one test
a program? And how does one integrate a tested set of component
programs into a tested and dependable system? We have touched
upon the techniques here and there; let us now consider them
somewhat more systematically.
Designing the Bugs Out
Bug-proofing the definition. The most pernicious and subtle
bugs are system bugs arising from mismatched assumptions made
by the authors of various components. The approach to conceptual
integrity discussed above in Chapters 4, 5, and 6 addresses these
problems directly. In short, conceptual integrity of the product not
only makes it easier to use, it also makes it easier to build and less
subject to bugs.
So does the detailed, painstaking architectural effort implied
by that approach. V. A. Vyssotsky, of Bell Telephone Laborato-
ries' Safeguard Project, says, "The crucial task is to get the product
defined. Many, many failures concern exactly those aspects that
were never quite specified."1 Careful function definition, careful
specification, and the disciplined exorcism of frills of function and
flights of technique all reduce the number of system bugs that
have to be found.
Testing the specification. Long before any code exists, the spec-
ification must be handed to an outside testing group to be scruti-
nized for completeness and clarity. As Vyssotsky says, the
developers themselves cannot do this: "They won't tell you they
don't understand it; they will happily invent their way through
the gaps and obscurities."
Designing the Bugs Out 143
Top-down design. In a very clear 1971 paper, Niklaus Wirth
formalized a design procedure which had been used for years by
the best programmers.2 Furthermore, his notions, although stated
for program design, apply completely to the design of complex
systems of programs. The division of system building into archi-
tecture, implementation, and realization is an embodiment of
these notions; furthermore, each of the architecture, implementa-
tion, and realization can be best done by top-down methods.
Briefly, Wirth's procedure is to identify design as a sequence
of refinement steps. One sketches a rough task definition and a
rough solution method that achieves the principal result. Then one
examines the definition more closely to see how the result differs
from what is wanted, and one takes the large steps of the solution
and breaks them down into smaller steps. Each refinement in the
definition of the task becomes a refinement in the algorithm for
solution, and each may be accompanied by a refinement in the
data representation.
From this process one identifies modules of solution or of data
whose further refinement can proceed independently of other
work. The degree of this modularity determines the adaptability
and changeability of the program.
Wirth advocates using as high-level a notation as is possible
at each step, exposing the concepts and concealing the details until
further refinement becomes necessary.
A good top-down design avoids bugs in several ways. First,
the clarity of structure and representation makes the precise state-
ment of requirements and functions of the modules easier. Second,
the partitioning and independence of modules avoids system bugs.
Third, the suppression of detail makes flaws in the structure more
apparent. Fourth, the design can be tested at each of its refinement
steps, so testing can start earlier and focus on the proper level of
detail at each step.
The process of step-wise refinement does not mean that one
never has to go back, scrap the top level, and start the whole thing
144 The Whole and the Parts
again as he encounters some unexpectedly knotty detail. Indeed,
that happens often. But it is much easier to see exactly when and
why one should throw away a gross design and start over. Many
poor systems come from an attempt to salvage a bad basic design
and patch it with all kinds of cosmetic relief. Top-down design
reduces the temptation.
I am persuaded that top-down design is the most important
new programming formalization of the decade.
Structured programming. Another important set of new ideas
for designing the bugs out of programs derives largely from
Dijkstra,3 and is built on a theoretical structure by Bohm and
Jacopini.4
Basically the approach is to design programs whose control
structures consist only of loops defined by a statement such as DO
WHILE, and conditional portions delineated into groups of state-
ments marked with brackets and conditioned by an IF ... THEN
. . . ELSE. Bohm and Jacopini show these structures to be theoreti-
cally sufficient; Dijkstra argues that the alternative, unrestrained
branching via GO TO, produces structures that lend themselves
to logical errors.
The basic notion is surely sound. Many criticisms have been
made, and additional control structures, such as an n-way branch
(the so-called CASE statement) for distinguishing among many
contingencies, and a disaster bail-out (GO TO ABNORMAL
END) are very convenient. Further, some have become very doc-
trinaire about avoiding all GO TO's, and that seems excessive.
The important point, and the one vital to constructing bug-
free programs, is that one wants to think about the control struc-
tures of a system as control structures, not as individual branch
statements. This way of thinking is a major step forward.
Component Debugging
The procedures for debugging programs have been through a great
cycle in the past twenty years, and in some ways they are back
Component Debugging 145
where they started. The cycle has gone through four steps, and it
is fun to trace them and see the motivation for each.
On-machine debugging. Early machines had relatively poor in-
put-output equipment, and long input-output delays. Typically,
the machine read and wrote paper tape or magnetic tape and
off-line facilities were used for tape preparation and printing. This
made tape input-output intolerably awkward for debugging, so
the console was used instead. Thus debugging was designed to
allow as many trials as possible per machine session.
The programmer carefully designed his debugging procedure
—planning where to stop, what memory locations to examine,
what to find there, and what to do if he didn't. This meticulous
programming of himself as a debugging machine might well take
half as long as writing the computer program to be debugged.
The cardinal sin was to push START boldly without having
segmented the program into test sections with planned stops.
Memory dumps. On-machine debugging was very effective. In
a two-hour session, one could get perhaps a dozen shots. But
computers were very scarce, and very costly, and the. thought of
all that machine time going to waste was horrifying.
So when high-speed printers were attached on-line, the tech-
nique changed. One ran a program until a check failed, and then
dumped the whole memory. Then began the laborious desk work,
accounting for each memory location's contents. The desk time
was not much different than that for on-machine debugging; but
it occurred after the test run, in deciphering, rather than before,
in planning. Debugging for any particular user took much longer,
because test shots depended upon batch turnaround time. The
whole procedure, however, was designed to minimize computer
time use, and to serve as many programmers as possible.
Snapshots. The machines on which memory dumping was de-
veloped had 2000-4000 words, or 8K to 16K bytes of memory. But
memory sizes grew by leaps and bounds, and total memory dump-
ing became impractical. So people developed techniques for selec-
146 The Whole and the Parts
tive dumping, selective tracing, and for inserting snapshots into
programs. The OS/360 TESTRAN is an end-of-the-line in this
direction, allowing one to insert snapshots into a program without
reassembly or recompilation.
Interactive debugging. In 1959 Codd and his coworkers5 and
Strachey8 each reported work aimed at time-shared debugging, a
way of achieving both the instant turnaround of on-machine
debugging and the efficient machine use of batch debugging. The
computer would have multiple programs in memory, ready for
execution. A terminal, controlled only by program, would be asso-
ciated with each program being debugged. Debugging would be
under control of a supervisory program. When the programmer at
a terminal stopped his program to examine progress or to make
changes, the supervisor would run another program, thus keeping
the machines busy.
Codd's multiprogramming system was developed, but the em-
phasis was on throughput enhancement by efficient input-output
utilization, and interactive debugging was not implemented. Stra-
chey's ideas were improved and implemented in 1963 in an experi-
mental system for the 7090 by Corbato and colleagues at MIT.7
This development led to the MULTICS, TSS, and other time-
sharing systems of today.
The chief user-perceived differences between on-machine
debugging as first practiced and the interactive debugging of today
are the facilities made possible by the presence of the supervisory
program and its associated language interpreters. One can program
and debug in a high-level language. Efficient editing facilities
make changes and snapshots easy.
Return to the instant-turnaround capability of on-machine
debugging has not yet brought a return to the preplanning of
debugging sessions. In a sense such preplanning is not so necessary
as before, since machine time doesn't waste away while one sits
and thinks.
Nevertheless, Gold's interesting experimental results show
that three times as much progress in interactive debugging is made
on the first interaction of each session as on subsequent interac-
System Debugging 147
tions.8 This strongly suggests that we are not realizing the poten-
tial of interaction due to lack of session planning. The time has
come to dust off the old on-machine techniques.
I find that proper use of a good terminal system requires two
hours at the desk for each two-hour session on the terminal. Half
of this time is spent in sweeping up after the last session: updating
my debugging log, filing updated program listings in my system
notebook, explaining strange phenomena. The other half is spent
in preparation: planning changes and improvements and designing
detailed tests for next time. Without such planning, it is hard to
stay productive for as much as two hours. Without the post-
session sweep-up, it is hard to keep the succession of terminal
sessions systematic and forward-moving.
Test cases. As for the design of actual debugging procedures and
test cases, Gruenberger has an especially good treatment,9 and
there are shorter treatments in other standard texts. 10'n
System Debugging
The unexpectedly hard part of building a programming system is
system test. I have already discussed some of the reasons for both
the difficulty and its unexpectedness. From all of that, one should
be convinced of two things: system debugging will take longer
than one expects, and its difficulty justifies a thoroughly system-
atic and planned approach. Let us now see what such an approach
involves.12
Use debugged components. Common sense, if not common
practice, dictates that one should begin system debugging only
after the pieces seem to work.
Common practice departs from this in two ways. First is the
bolt-it-together-and-try approach. This seems to be based on the
notion that there will be system (i.e., interface) bugs in addition
to the component bugs. The sooner one puts the pieces together,
the sooner the system bugs will emerge. Somewhat less sophis-
ticated is the notion that by using the pieces to test each other, one
148 The Whole and the Parts
avoids a lot of test scaffolding. Both of these are obviously true,
but experience shows that they are not the whole truth—the use
of clean, debugged components saves much more time in system
testing than that spent on scaffolding and thorough component
test.
A little more subtle is the "documented bug" approach. This
says that a component is ready to enter system test when all the
flaws are found, well before the time when all are fixed. Then in
system testing, so the theory goes, one knows the expected effects
of these bugs and can ignore those effects, concentrating on the
new phenomena.
All this is just wishful thinking, invented to rationalize away
the pain of slipped schedules. One does not know all the expected
effects of known bugs. If things were straightforward, system
testing wouldn't be hard. Furthermore, the fixing of the docu-
mented component bugs will surely inject unknown bugs, and
then system test is confused.
Build plenty of scaffolding. By scaffolding I mean all programs
and data built for debugging purposes but never intended to be in
the final product. It is not unreasonable for there to be half as
much code in scaffolding as there is in product.
One form of scaffolding is the dummy component, which con-
sists only of interfaces and perhaps some faked data or some small
test cases. For example, a system may include a sort program
which isn't finished yet. Its neighbors can be tested by using a
dummy program that merely reads and tests the format of input
data, and spews out a set of well-formatted meaningless but or-
dered data.
Another form is the miniature file. A very common form of
system bug is misunderstanding of formats for tape and disk files.
So it is worthwhile to build some little files that have only a few
typical records, but all the descriptions, pointers, etc.
The limiting case of miniature file is the dummy file, which
really isn't there at all. OS/360's Job Control Language provides
such facility, and it is extremely useful for component debugging.
System Debugging 149
Yet another form of scaffolding are auxiliary programs. Gener-
ators for test data, special analysis printouts, cross-reference table
analyzers, are all examples of the special-purpose jigs and fixtures
one may want to build.13
Control changes. Tight control during test is one of the impres-
sive techniques of hardware debugging, and it applies as well to
software systems.
First, somebody must be in charge. He and he alone must
authorize component changes or substitution of one version for
another.
Then, as discussed above, there must be controlled copies of
the system: one locked-up copy of the latest versions, used for
component testing; one copy under test, with fixes being installed;
playpen copies where each man can work away on his component,
doing both fixes and extensions.
In System/360 engineering models, one saw occasional
strands of purple wire among the routine yellow wires. When a
bug was found, two things were done. A quick fix was devised and
installed on the system, so testing could proceed. This change was
put on in purple wire, so it stuck out like a sore thumb. It was
entered in the log. Meanwhile, an official change document was
prepared and started into the design automation mill. Eventually
this resulted in updated drawings and wire lists, and a new back
panel in which the change was implemented in printed circuitry
or yellow wire. Now the physical model and the paper were to-
gether again, and the purple wire was gone.
Programming needs a purple-wire technique, and it badly
needs tight control and deep respect for the paper that ultimately
is the product. The vital ingredients of such technique are the
logging of all changes in a journal and the distinction, carried
conspicuously in source code, between quick patches and
thought-through, tested, documented fixes.
Add one component at a time. This precept, too, is obvious, but
optimism and laziness tempt us to violate it. To do it requires
150 The Whole and the Parts
dummies and other scaffolding, and that takes work. And after all,
perhaps all that work won't be needed? Perhaps there are no bugs?
No! Resist the temptation! That is what systematic system
testing is all about. One must assume that there will be lots of
bugs, and plan an orderly procedure for snaking them out.
Note that one must have thorough test cases, testing the par-
tial systems after each new piece is added. And the old ones, run
successfully on the last partial sum, must be rerun on the new one
to test for system regression.
Quantize updates. As the system comes up, the component
builders will from time to time appear, bearing hot new versions
of their pieces—faster,smaller, more complete, or putatively less
buggy. The replacement of a working component by a new version
requires the same systematic testing procedure that adding a new
component does, although it should require less time, for more
complete and efficient test cases will usually be available.
Each team building another component has been using the
most recent tested version of the integrated system as a test bed
for debugging its piece. Their work will be set back by having that
test bed change under them. Of course it must. But the changes
need to be quantized. Then each user has periods of productive
stability, interrupted by bursts of test-bed change. This seems to
be much less disruptive than a constant rippling and trembling.
Lehman and Belady offer evidence that quanta should be very
large and widely spaced or else very small and frequent.14 The
latter strategy is more subject to instability, according to their
model. My experience confirms it: I would never risk that strategy
in practice.
Quantized changes neatly accommodate a purple-wire tech-
nique. The quick patch holds until the next regular release of the
component, which should incorporate the fix in tested and docu-
mented form.
14
Hatching a Catastrophe
14
Hatching a Catastrophe
None love the bearer of bad news.
SOPHOCLES
How does a project get to be a year late?
. . , One day at a time.
A. Canova, "Ercole e lica," 1802. Hercules hurls to his death the
messenger Lycas, who innocently brought the death-garment.
Scala/Art Resource, NY
153
154 Hatching a Catastrophe
When one hears of disastrous schedule slippage in a project, he
imagines that a series of major calamities must have befallen it.
Usually, however, the disaster is due to termites, not tornadoes;
and the schedule has slipped imperceptibly but inexorably. In-
deed, major calamities are easier to handle; one responds with
major force, radical reorganization, the invention of new ap-
proaches. The whole team rises to the occasion.
But the day-by-day slippage is harder to recognize, harder to
prevent, harder to make up. Yesterday a key man was sick, and a
meeting couldn't be held. Today the machines are all down, be-
cause lightning struck the building's power transformer. Tomor-
row the disk routines won't start testing, because the first disk is
a week late from the factory. Snow, jury duty, family problems,
emergency meetings with customers, executive audits—the list
goes on and on. Each one only postpones some activity by a
half-day or a day. And the schedule slips, one day at a time.
Milestones or Millstones?
How does one control a big project on a tight schedule? The first
step is to have a schedule. Each of a list of events, called milestones,
has a date. Picking the dates is an estimating problem, discussed
already and crucially dependent on experience.
For picking the milestones there is only one relevant rule.
Milestones must be concrete, specific, measurable events, defined
with knife-edge sharpness. Coding, for a counterexample, is "90
percent finished" for half of the total coding time. Debugging is
"99 percent complete" most of the time. "Planning complete" is
an event one can proclaim almost at will.1
Concrete milestones, on the other hand, are 100-percent
events. "Specifications signed by architects and implementers,"
"source coding 100 percent complete, keypunched, entered into
disk library," "debugged version passes all test cases." These con-
crete milestones demark the vague phases of planning, coding,
debugging.
"The Other Piece Is Late, Anyway" 155
It is more important that milestones be sharp-edged and un-
ambiguous than that they be easily verifiable by the boss. Rarely
will a man lie about milestone progress, if the milestone is so sharp
that he can't deceive himself. But if the milestone is fuzzy, the boss
often understands a different report from that which the man
gives. To supplement Sophocles, no one enjoys bearing bad news,
either, so it gets softened without any real intent to deceive.
Two interesting studies of estimating behavior by government
contractors on large-scale development projects show that:
1. Estimates of the length of an activity, made and revised care-
fully every two weeks before the activity starts, do not signifi-
cantly change as the start time draws near, no matter how
wrong they ultimately turn out to be.
2. During the activity, overestimates of duration come steadily
down as the activity proceeds.
3. Underestimates do not change significantly during the activity
until about three weeks before the scheduled completion.2
Sharp milestones are in fact a service to the team, and one they
can properly expect from a manager. The fuzzy milestone is the
harder burden to live with. It is in fact a millstone that grinds
down morale, for it deceives one about lost time until it is ir-
remediable. And chronic schedule slippage is a morale-killer.
"The Other Piece Is Late, Anyway"
A schedule slips a day; so what? Who gets excited about a one-day
slip? We can make it up later. And the other piece into which ours
fits is late, anyway.
A baseball manager recognizes a nonphysical talent, hustle, as
an essential gift of great players and great teams. It is the charac-
teristic of running faster than necessary, moving sooner than nec-
essary, trying harder than necessary. It is essential for great
programming teams, too. Hustle provides the cushion, the reserve
capacity, that enables a team to cope with routine mishaps, to
156 Hatching a Catastrophe
anticipate and forfend minor calamities. The calculated response,
the measured effort, are the wet blankets that dampen hustle. As
we have seen, one must get excited about a one-day slip. Such are
the elements of catastrophe.
But not all one-day slips are equally disastrous. So some calcu-
lation of response is necessary, though hustle be dampened. How
does one tell which slips matter? There is no substitute for a PERT
chart or a critical-path schedule. Such a network shows who waits
for what. It shows who is on the critical path, where any slip
moves the end date. It also shows how much an activity can slip
before it moves into the critical path.
The PERT technique, strictly speaking, is an elaboration of
critical-path scheduling in which one estimates three times for
every event, times corresponding to different probabilities of
meeting the estimated dates. I do not find this refinement to be
worth the extra effort, but for brevity I will call any critical path
network a PERT chart.
The preparation of a PERT chart is the most valuable part of
its use. Laying out the network, identifying the dependencies, and
estimating the legs all force a great deal of very specific planning
very early in a project. The first chart is always terrible, and one
invents and invents in making the second one.
As the project proceeds, the PERT chart provides the answer
to the demoralizing excuse, "The other piece is late anyhow." It
shows how hustle is needed to keep one's own part off the critical
path, and it suggests ways to make up the lost time in the other
part.
Under the Rug
When a first-line manager sees his small team slipping behind, he
is rarely inclined to run to the boss with this woe. The team might
be able to make it up, or he should be able to invent or reorganize
to solve the problem. Then why worry the boss with it? So far, so
Under the Rwg' 157
good. Solving such problems is exactly what the first-line manager
is there for. And the boss does have enough real worries demand-
ing his action that he doesn't seek others. So all the dirt gets swept
under the rug.
But every boss needs two kinds of information, exceptions to
plan that require action and a status picture for education.3 For
that purpose he needs to know the status of all his teams. Getting
a true picture of that status is hard.
The first-line manager's interests and those of the boss have
an inherent conflict here. The first-line manager fears that if he
reports his problem, the boss will act on it. Then his action will
preempt the manager's function, diminish his authority, foul up
his other plans. So as long as the manager thinks he can solve it
alone, he doesn't tell the boss.
Two rug-lifting techniques are open to the boss. Both must be
used. The first is to reduce the role conflict and inspire sharing of
status. The other is to yank the rug back.
Reducing the role conflict. The boss must first distinguish be-
tween action information and status information. He must disci-
pline himself not to act on problems his managers can solve, and
never to act on problems when he is explicitly reviewing status. I
once knew a boss who invariably picked up the phone to give
orders before the end of the first paragraph in a status report. That
response is guaranteed to squelch full disclosure.
Conversely, when the manager knows his boss will accept
status reports without panic or preemption, he comes to give hon-
est appraisals.
This whole process is helped if the boss labels meetings, re-
views, conferences, as status-review meetings versus problem-action
meetings, and controls himself accordingly. Obviously one may
call a problem-action meeting as a consequence of a status meet-
ing, if he believes a problem is out of hand. But at least everybody
knows what the score is, and the boss thinks twice before grabbing
the ball.
158 Hatching a Catastrophe
Yanking the rug off. Nevertheless, it is necessary to have review
techniques by which the true status is made known, whether
cooperatively or not. The PERT chart with its frequent sharp
milestones is the basis for such review. On a large project one may
want to review some part of it each week, making the rounds once
a month or so.
A report showing milestones and actual completions is the key
document. Figure 14.1 shows an excerpt from such a report. This
report shows some troubles. Specifications approval is overdue on
several components. Manual (SLR) approval is overdue on an-
other, and one is late getting out of the first state (Alpha) of the
independently conducted product test. So such a report serves as
an agenda for the meeting of 1 February. Everyone knows the
questions, and the component manager should be prepared to
explain why it's late, when it will be finished, what steps he's
taking, and what help, if any, he needs from the boss or collateral
groups.
V. Vyssotsky of Bell Telephone Laboratories adds the follow-
ing observation:
/ have found it handy to carry both "scheduled" and "estimated"
dates in the milestone report. The scheduled dates are the property of
the project manager and represent a consistent work plan for the
project as a whole, and one which is a priori a reasonable plan. The
estimated dates are the property of the lowest level manager who has
cognizance over the piece of work in question, and represents his best
judgment as to when it will actually happen, given the resources he
has available and when he received (or has commitments for delivery
of) his prerequisite inputs. The project manager has to keep his fingers
off the estimated dates, and put the emphasis on getting accurate,
unbiased estimates rather than palatable optimistic estimates or self-
protective conservative ones. Once this is clearly established in every-
one 's mind, the project manager can see quite a ways into the future
where he is going to be in trouble if he doesn 't do something. *
Under the Rug 159
160 Hatching a Catastrophe
The preparation of the PERT chart is a function of the boss
and the managers reporting to him. Its updating, revision, and
reporting requires the attention of a small (one to three man) staff
group which serves as an extension of the boss. Such a Plans and
Controls team is invaluable for a large project. It has no authority
except to ask all the line managers when they will have set or
changed milestones, and whether milestones have been met. Since
the Plans and Controls group handles all the paperwork, the bur-
den on the line managers is reduced to the essentials—making the
decisions.
We had a skilled, enthusiastic, and diplomatic Plans and Con-
trols group, run by A. M. Pietrasanta, who devoted considerable
inventive talent to devising effective but unobtrusive control
methods. As a result, I found his group to be widely respected and
more than tolerated. For a group whose role is inherently that of
an irritant, this is quite an accomplishment.
The investment of a modest amount of skilled effort in a Plans
and Controls function is very rewarding. It makes far more differ-
ence in project accomplishment than if these people worked di-
rectly on building the product programs. For the Plans and
Controls group is the watchdog who renders the imperceptible
delays visible and who points up the critical elements. It is the
early warning system against losing a year, one day at a time.
15
The Other Face
15
The Other Face
What we do not understand we do not possess.
GOETHE
O give me commentators plain,
Who with no deep researches vex the brain.
CRABBE
A reconstruction of Stonehenge, the world's largest undocumented
computer.
The Bettman Archive
163
164 The Other Face
A computer program is a message from a man to a machine. The
rigidly marshaled syntax and the scrupulous definitions all exist
to make intention clear to the dumb engine.
But a written program has another face, that which tells its
story to the human user. For even the most private of programs,
some such communication is necessary; memory will fail the au-
thor-user, and he will require refreshing on the details of his
handiwork.
How much more vital is the documentation for a public pro-
gram, whose user is remote from the author in both time and
space! For the program product, the other face to the user is fully
as important as the face to the machine.
Most of us have quietly excoriated the remote and anonymous
author of some skimpily documented program. And many of us
have therefore tried to instill in new programmers an attitude
about documentation that would inspire for a lifetime, overcom-
ing sloth and schedule pressure. By and large we have failed. I
think we have used wrong methods.
Thomas J. Watson, Sr. told the story of his first experience as
a cash register salesman in upstate New York. Charged with en-
thusiasm, he sallied out with his wagon loaded with cash registers.
He worked his territory diligently but without selling a one.
Downcast, he reported to his boss. The sales manager listened a
while, then said, "Help me load some registers into the wagon,
harness the horse, and let's go again." They did, and the two called
on customer after customer, with the older man showing how to sell
cash registers. All evidence indicates that the lesson took.
For several years I diligently lectured my software engineering
class on the necessity and propriety of good documentation, ex-
horting them ever more fervently and eloquently. It didn't work.
I assumed they had learned how to document properly and were
failing from lack of zeal. Then I tried loading some cash registers
into the wagon; i.e., showing them how the job is done. This has
been much more successful. So the remainder of this essay will
downplay exhortation and concentrate on the "how" of good
documentation.
What Documentation Is Required? 165
What Documentation Is Required?
Different levels of documentation are required for the casual user
of a program, for the user who must depend upon a program, and
for the user who must adapt a program for changes in circum-
stance or purpose.
To use a program. Every user needs a prose description of the
program. Most documentation fails in giving too little overview.
The trees are described, the bark and leaves are commented, but
there is no map of the forest. To write a useful prose description,
stand way back and come in slowly:
1. Purpose. What is the main function, the reason for the pro-
gram?
2. Environment. On what machines, hardware configurations,
and operating system configurations will it run?
3. Domain and range. What domain of input is valid? What range
of output can legitimately appear?
4. Functions realized and algorithms used. Precisely what does it do?
5. Input-output formats,, precise and complete.
6. Operating instructions, including normal and abnormal ending
behavior, as seen at the console and on the outputs.
7. Options. What choices does the user have about functions?
Exactly how are those choices specified?
8. Running time. How long does it take to do a problem of speci-
fied size on a specified configuration?
9. Accuracy and checking. How precise are the answers expected
to be? What means of checking accuracy are incorporated?
Often all this information can be set forth in three or four
pages. That requires close attention to conciseness and precision.
Most of this document needs to be drafted before the program is
written, for it embodies basic planning decisions.
To believe a program. The description of how it is used must be
supplemented with some description of how one knows it is work-
ing. This means test cases.
166 The Other Face
Every copy of a program shipped should include some small
test cases that can be routinely used to reassure the user that he
has a faithful copy, accurately loaded into the machine.
Then one needs more thorough test cases, which are normally
run only after a program is modified. These fall into three parts of
the input data domain:
1. Mainline cases that test the program's chief functions for com-
monly encountered data.
2. Barely legitimate cases that probe the edge of the input data
domain, ensuring that largest possible values, smallest possi-
ble values, and all kinds of valid exceptions work.
3. Barely illegitimate cases that probe the domain boundary from
the other side, ensuring that invalid inputs raise proper diag-
nostic messages.
To modify a program. Adapting a program or fixing it requires
considerably more information. Of course the full detail is re-
quired, and that is contained in a well-commented listing. For the
modifier, as well as the more casual user, the crying need is for a
clear, sharp overview, this time of the internal structure. What are
the components of such an overview?
1. A flow chart or subprogram structure graph. More on this
later.
2. Complete descriptions of the algorithms used, or else refer-
ences to such descriptions in the literature.
3. An explanation of the layout of all files used.
4. An overview of the pass structure—the sequence in which
data or programs are brought from tape or disk—and what is
accomplished on each pass.
5. A discussion of modifications contemplated in the original
design, the nature and location of hooks and exits, and discur-
sive discussion of the ideas of the original author about what
modifications might be desirable and how one might proceed.
His observations on hidden pitfalls are also useful.
The How-Chart Cone 167
The Flow-Chart Curse
The flow chart is a most thoroughly oversold piece of program
documentation. Many programs don't need flow charts at all; few
programs need more than a one-page flow chart.
Flow charts show the decision structure of a program, which
is only one aspect of its structure. They show decision structure
rather elegantly when the flow chart is on one page, but the over-
Fig. 15.1 A program structure graph. (Courtesy of W. V. Wright)
168 The Other Face
view breaks down badly when one has multiple pages, sewed
together with numbered exits and connectors.
The one-page flow chart for a substantial program becomes
essentially a diagram of program structure, and of phases or steps.
As such it is very handy. Figure 15.1 shows such a subprogram
structure graph.
Of course such a structure graph neither follows nor needs the
painfully wrought ANSI flow-charting standards. All the rules on
box shapes, connectors, numbering, etc. are needed only to give
intelligibility to detailed flow charts.
The detailed blow-by-blow flow chart, however, is an obso-
lete nuisance, suitable only for initiating beginners into algorith-
mic thinking. When introduced by Goldstine and von Neumann,1
the little boxes and their contents served as a high-level language,
grouping the inscrutable machine-language statements into clus-
ters of significance. As Iverson early recognized,2 in a systematic
high-level language the clustering is already done, and each box
contains a statement (Fig. 15.2). Then the boxes themselves
become no more than a tedious and space-hogging exercise in
drafting; they might as well be eliminated. Then nothing is left but
the arrows. The arrows joining a statement to its successor are
redundant; erase them. That leaves only GO TO's. And if one
follows good practice and uses block structure to minimize GO
TO's, there aren't many arrows, but they aid comprehension im-
mensely. One might as well draw them on the listing and eliminate
the flow chart altogether.
In fact, flow charting is more preached than practiced. I have
never seen an experienced programmer who routinely made de-
tailed flow charts before beginning to write programs. Where or-
ganization standards require flow charts, these are almost
invariably done after the fact. Many shops proudly use machine
programs to generate this "indispensable design tool" from the
completed code. I think this universal experience is not an embar-
rassing and deplorable departure from good practice, to be ac-
knowledged only with a nervous laugh. Instead it is the
The Flow-Chart 169
application of good judgment, and it teaches us something about
the utility of flow charts.
The Apostle Peter said of new Gentile converts and the Jewish
law, "Why lay a load on [their] backs which neither our ancestors
nor we ourselves were able to carry?" (Acts 15:10, TEV). I would
say the same about new programmers and the obsolete practice of
flow charting.
Self-Documenting Programs
A basic principle of data processing teaches the folly of trying to
maintain independent files in synchronism. It is far better to com-
bine them into one file with each record containing all the infor-
mation both files held concerning a given key.
Yet our practice in programming documentation violates
our own teaching. We typically attempt to maintain a machine-
readable form of a program and an independent set of human-
readable documentation, consisting of prose and flow charts.
The results in fact confirm our teachings about the folly of
separate files. Program documentation is notoriously poor, and its
maintenance is worse. Changes made in the program do not
promptly, accurately, and invariably appear in the paper.
The solution, I think, is to merge the files, to incorporate the
documentation in the source program. This is at once a powerful
incentive toward proper maintenance, and an insurance that the
documentation will always be handy to the program user. Such
programs are called self-documenting.
Now clearly this is awkward (but not impossible) if flow
charts are to be included. But grant the obsolescence of flow charts
and the dominant use of high-level language, and it becomes
reasonable to combine the program and the documentation.
The use of a source program as a documentation medium
imposes some constraints. On the other hand, the intimate avail-
ability of the source program, line by line, to the reader of the
documentation makes possible new techniques. The time has
170 The Other Face
172 The Other Face
come to devise radically new approaches and methods for program
documentation.
As a principal objective, we must attempt to minimize the
burden of documentation, the burden neither we nor our prede-
cessors have been able to bear successfully.
An approach. The first notion is to use the parts of the program
that have to be there anyway, for programming language reasons,
to carry as much of the documentation as possible. So labels,
declaration statements, and symbolic names are all harnessed to
the task of conveying as much meaning as possible to the reader.
A second notion is to use space and format as much as possible
to improve readability and show subordination and nesting.
The third notion is to insert the necessary prose documenta-
tion into the program as paragraphs of comment. Most programs
tend to have enough line-by-line comments; those programs pro-
duced to meet stiff organizational standards for "good documenta-
tion" often have too many. Evert these programs, however, are
usually deficient in the paragraph comments that really give intel-
ligibility and overview to the whole thing.
Since the documentation is built into the structure, naming,
and formats of the program, much of it must be done when the
program is first written. But that is when it should be written. Since
the self-documentation approach minimizes extra work, there are
fewer obstacles to doing it then.
Some techniques. Figure 15.3 shows a self-documenting PL/I
program.3 The numbers in the circles are not part of it; they are
meta-documentation keyed to the discussion.
1. Use a separate job name for each run, and maintain a run log
showing what was tried, when, and the results. If the name is
composed of a mnemonic part (here QLT) and a numerical
suffix (here 4), the suffix can be used as a run number, tying
listings and log together. This technique requires a new job
card for each run, but they can be made up in batches, dupli-
cating the common information.
Fig. 15.3 A self-documenting program.
174 The Other Face
2. Use a program name that is mnemonic but also contains a
version identifier. That is, assume there will be several ver-
sions. Here the index is the low order digit of the year 1967.
3. Incorporate the prose description as comments to PROCE-
DURE.
4. Refer to standard literature to document basic algorithms
wherever possible. This saves space, usually points to a much
fuller treatment than one would provide, and allows the
knowledgeable reader to skip it with confidence that he un-
derstands you.
5. Show the relationship to the book algorithm:
a) changes b) specialization c) representation
6. Declare all variables. Use mnemonic names. Use comments to
convert DECLARE into a complete legend. Note that it already
contains names and structural descriptions, it needs only to be
augmented with descriptions of purpose. By doing so here, one
can avoid repeating the names and structural descriptions in
a separate treatment.
7. Mark the initialization by a label.
8. Label statements in groups to show correspondences to the
statements in the algorithm description in the literature.
9. Use indenting to show structure and grouping.
10. Add logical flow arrows to the listing by hand. They are very
helpful in debugging and changing. They may be incorporated
in the right margin of the comments space, and made part of
the machine-readable text.
11. Use line comments or remark anything that is not obvious. If
the techniques above have been used, these will be short and
fewer in number than is customary.
12. Put multiple statements on one line, or one statement on sev-
eral lines to match thought-grouping and to show correspon-
dence to other algorithm description.
Why not? What are the drawbacks of such an approach to docu-
mentation? There are several, which have been real but are becom-
ing imaginary with changing times.
Self-Documenting Programs 175
The most serious objection is the increase in the size of the
source code that must be stored. As the discipline moves more and
more toward on-line storage of source code, this has become a
growing consideration. I find myself being briefer in comments to
an APL program, which will live on a disk, than on a PL/I one that
I will store as cards.
Yet simultaneously we are moving also toward on-line storage
of prose documents for access and for updating via computerized
text-editing. As shown above, amalgamating prose and program
reduces the total number of characters to be stored.
A similar answer applies to the argument that self-document-
ing programs require more keystrokes. A typed document requires
at least one keystroke per character per draft. A self-documenting
program has fewer total characters and also fewer strokes per
character, since drafts aren't retyped.
How about flow charts and structure graphs? If one uses only
a highest-level structure graph, it might safely be kept as a sepa-
rate document, for it is not subject to frequent change. But it can
certainly be incorporated into the source program as a comment,
and that seems wise.
To what extent are the techniques used above applicable to
assembly language programs? I think the basic approach of self-
documentation is thoroughly applicable. Space and formats are
less free, and thus cannot be so flexibly used. Names and structural
declarations can surely be exploited. Macros can help a great deal.
The extensive use of paragraph comments is good practice in any
language.
But the self-documentation approach is stimulated by the use
of high-level languages and finds its greatest power and its great-
est justification in high-level languages used with on-line systems,
whether batch or interactive. As I have argued, such languages and
systems help programmers in very powerful ways. Since machines
are made for people, not people for machines, their use makes
every form of sense, economic and human.
16
No Silver Bullet-
Essence and Accident
in Software Engineering
16
No Silver Bullet-
Essence and Accident
in Software Engineering
There is no single development, in either technology or
management technique, which by itself promises even
one order-of-magnitude improvement within a decade in
productivity, in reliability, in simplicity.
The Werewolf of Eschenbach, Germany: line engraving, 1685.
Courtesy of The Grainger Collection, New York.
179
180 No Silver Bullet
Abstract1
All software construction involves essential tasks, the fashion-
ing of the complex conceptual structures that compose the ab-
stract software entity, and accidental tasks, the representation of
these abstract entities in programming languages and the map-
ping of these onto machine languages within space and speed
constraints. Most of the big past gains in software productivity
have come from removing artificial barriers that have made the
accidental tasks inordinately hard, such as severe hardware con-
straints, awkward programming languages, lack of machine
time. How much of what software engineers now do is still de-
voted to the accidental, as opposed to the essential? Unless it is
more than 9/10 of all effort, shrinking all the accidental activities
to zero time will not give an order of magnitude improvement.
Therefore it appears that the time has come to address the
essential parts of the software task, those concerned with fash-
ioning abstract conceptual structures of great complexity. I sug-
gest:
• Exploiting the mass market to avoid constructing what can
be bought.
• Using rapid prototyping as part of a planned iteration in es-
tablishing software requirements.
• Growing software organically, adding more and more func-
tion to systems as they are run, used, and tested.
• Identifying and developing the great conceptual designers
of the rising generation.
Introduction
Of all the monsters who fill the nightmares of our folklore, none
terrify more than werewolves, because they transform unex-
pectedly from the familiar into horrors. For these, we seek bul-
lets of silver that can magically lay them to rest.
The familiar software project has something of this character
(at least as seen by the nontechnical manager), usually innocent
Does It Have to Be Hard?—Essential Difficulties 181
and straightforward, but capable of becoming a monster of
missed schedules, blown budgets, and flawed products. So we
hear desperate cries for a silver bullet, something to make soft-
ware costs drop as rapidly as computer hardware costs do.
But, as we look to the horizon of a decade hence, we see no
silver bullet. There is no single development, in either technol-
ogy or management technique, which by itself promises even
one order of magnitude improvement in productivity, in reli-
ability, in simplicity. In this chapter we shall try to see why, by
examining both the nature of the software problem and the
properties of the bullets proposed.
Skepticism is not pessimism, however. Although we see no
startling breakthroughs, and indeed, believe such to be incon-
sistent with the nature of software, many encouraging innova-
tions are under way. A disciplined, consistent effort to develop,
propagate, and exploit them should indeed yield an order-of-
magnitude improvement. There is no royal road, but there is a
road.
The first step toward the management of disease was re-
placement of demon theories and humours theories by the germ
theory. That very step, the beginning of hope, in itself dashed
all hopes of magical solutions. It told workers that progress
would be made stepwise, at great effort, and that a persistent,
unremitting care would have to be paid to a discipline of clean-
liness. So it is with software engineering today.
Does It Have to Be Hard?—Essential Difficulties
Not only are there no silver bullets now in view, the very nature
of software makes it unlikely that there will be any—no inven-
tions that will do for software productivity, reliability, and sim-
plicity what electronics, transistors, and large-scale integration
did for computer hardware. We cannot expect ever to see two-
fold gains every two years.
First, we must observe that the anomaly is not that software
progress is so slow but that computer hardware progress is so
182 No Silver Bullet
fast. No other technology since civilization began has seen six
orders of magnitude price-performance gain in 30 years. In no
other technology can one choose to take the gain in either im-
proved performance or in reduced costs. These gains flow from
the transformation of computer manufacture from an assembly
industry into a process industry.
Second, to see what rate of progress we can expect in soft-
ware technology, let us examine its difficulties. Following Aris-
totle, I divide them into essence—the difficulties inherent in the
nature of the software—and accidents—those difficulties that to-
day attend its production but that are not inherent.
The accidents I discuss in the next section. First let us con-
sider the essence.
The essence of a software entity is a construct of interlocking
concepts: data sets, relationships among data items, algorithms,
and invocations of functions. This essence is abstract, in that the
conceptual construct is the same under many different represen-
tations. It is nonetheless highly precise and richly detailed.
/ believe the hard part of building software to be the specification,
design, and testing of this conceptual construct, not the labor of repre-
senting it and testing the fidelity of the representation. We still make
syntax errors, to be sure; but they are fuzz compared to the con-
ceptual errors in most systems.
If this is true, building software will always be hard. There
is inherently no silver bullet.
Let us consider the inherent properties of this irreducible
essence of modern software systems: complexity, conformity,
changeability, and invisibility.
Complexity. Software entities are more complex for their size
than perhaps any other human construct, because no two parts
are alike (at least above the statement level). If they are, we
make the two similar parts into one, a subroutine, open or
closed. In this respect software systems differ profoundly from
computers, buildings, or automobiles, where repeated elements
abound.
Does It Have to Be Hard?—Essential Difficulties 183
Digital computers are themselves more complex than most
things people build; they have very large numbers of states.
This makes conceiving, describing, and testing them hard. Soft-
ware systems have orders of magnitude more states than com-
puters do.
Likewise, a scaling-up of a software entity is not merely a
repetition of the same elements in larger size; it is necessarily an
increase in the number of different elements. In most cases, the
elements interact with each other in some nonlinear fashion,
and the complexity of the whole increases much more than lin-
early.
The complexity of software is an essential property, not an
accidental one. Hence descriptions of a software entity that ab-
stract away its complexity often abstract away its essence. Math-
ematics and the physical sciences made great strides for three
centuries by constructing simplified models of complex phe-
nomena, deriving, properties from the models, and verifying
those properties experimentally. This worked because the com-
plexities ignored in the models were not the essential properties
of the phenomena. It does not work when the complexities are
the essence.
Many of the classical problems of developing software prod-
ucts derive from this essential complexity and its nonlinear in-
creases with size. From the complexity comes the difficulty
of communication among team members, which leads to prod-
uct flaws, cost overruns, schedule delays. From the complexity
comes the difficulty of enumerating, much less understanding,
all the possible states of the program, and from that comes the
unreliability. From the complexity of the functions comes the
difficulty of invoking those functions, which makes programs
hard to use. From complexity of structure comes the difficulty
of extending programs to new functions without creating side
effects. From complexity of structure comes the unvisualized
states that constitute security trapdoors.
Not only technical problems but management problems as
well come from the complexity. This complexity makes over-
184 No Silver Bullet
view hard, thus impeding conceptual integrity. It makes it hard
to find and control all the loose ends. It creates the tremendous
learning and understanding burden that makes personnel turn-
over a disaster.
Conformity. Software people are not alone in facing complex-
ity. Physics deals with terribly complex objects even at the "fun-
damental" particle level. The physicist labors on, however, in a
firm faith that there are unifying principles to be found, whether
in quarks or in unified field theories. Einstein repeatedly argued
that there must be simplified explanations of nature, because
God is not capricious or arbitrary.
No such faith comforts the software engineer. Much of the
complexity he must master is arbitrary complexity, forced with-
out rhyme or reason by the many human institutions and sys-
tems to which his interfaces must conform. These differ from
interface to interface, and from time to time, not because of ne-
cessity but only because they were designed by different people,
rather than by God.
In many cases the software must conform because it has
most recently come to the scene. In others it must conform be-
cause it is perceived as the most conformable. But in all cases,
much complexity comes from conformation to other interfaces;
this cannot be simplified out by any redesign of the software
alone.
Changeability. The software entity is constantly subject to
pressures for change. Of course, so are buildings, cars, com-
puters. But manufactured things are infrequently changed after
manufacture; they are superseded by later models, or essential
changes are incorporated in later serial-number copies of the
same basic design. Call-backs of automobiles are really quite in-
frequent; field changes of computers somewhat less so. Both are
much less frequent than modifications to fielded software.
Partly this is because the software in a system embodies its
function, and the function is the part that most feels the pres-
sures of change. Partly it is because software can be changed
Does It Have to Be Hard?—Essential Difficulties 185
more easily—it is pure thought-stuff, infinitely malleable. Build-
ings do in fact get changed, but the high costs of change, under-
stood by all, serve to dampen the whims of the changers.
All successful software gets changed. Two processes are at
work. As a software product is found to be useful, people try it
in new cases at the edge of, or beyond, the original domain. The
pressures for extended function come chiefly from users who
like the basic function and invent new uses for it.
Second, successful software also survives beyond the nor-
mal life of the machine vehicle for which it is first written. If not
new computers, then at least new disks, new displays, new
printers come along; and the software must be conformed to its
new vehicles of opportunity.
In short, the software product is embedded in a cultural ma-
trix of applications, users, laws, and machine vehicles. These all
change continually, and their changes inexorably force change
upon the software product.
Invisibility. Software is invisible and unvisualizable. Geometric
abstractions are powerful tools. The floor plan of a building
helps both architect and client evaluate spaces, traffic flows,
views. Contradictions become obvious, omissions can be caught.
Scale drawings of mechanical parts and stick-figure models of
molecules, although abstractions, serve the same purpose. A
geometric reality is captured in a geometric abstraction.
The reality of software is not inherently embedded in space.
Hence it has no ready geometric representation in the way that
land has maps, silicon chips have diagrams, computers have
connectivity schematics. As soon as we attempt to diagram soft-
ware structure, we find it to constitute not one, but several,
general directed graphs, superimposed one upon another. The
several graphs may represent the flow of control, the flow of
data, patterns of dependency, time sequence, name-space rela-
tionships. These are usually not even planar, much less hier-
archical. Indeed, one of the ways of establishing conceptual
control over such structure is to enforce link cutting until one or
186 No Silver Bullet
more of the graphs becomes hierarchical.2
In spite of progress in restricting and simplifying the struc-
tures of software, they remain inherently unvisualizable, thus
depriving the mind of some of its most powerful conceptual
tools. This lack not only impedes the process of design within
one mind, it severely hinders communication among minds.
Past Breakthroughs Solved Accidental Difficulties
If we examine the three steps in software technology that have
been most fruitful in the past, we discover that each attacked a
different major difficulty in building software, but they have
been the accidental, not the essential, difficulties. We can also
see the natural limits to the extrapolation of each such attack.
High-level languages. Surely the most powerful stroke for
software productivity, reliability, and simplicity has been the
progressive use of high-level languages for programming. Most
observers credit that development with at least a factor of five
in productivity, and with concomitant gains in reliability, sim-
plicity, and comprehensibility.
What does a high-level language accomplish? It frees a pro-
gram from much of its accidental complexity. An abstract pro-
gram consists of conceptual constructs: operations, datatypes,
sequences, and communication. The concrete machine program
is concerned with bits, registers, conditions, branches, chan-
nels, disks, and such. To the extent that the high-level language
embodies the constructs wanted in the abstract program and
avoids all lower ones/it eliminates a whole level of complexity
that was never inherent in the program at all.
The most a high-level language can do is to furnish all the
constructs the programmer imagines in the abstract program. To
be sure, the level of our sophistication in thinking about data
structures, data types, and operations is steadily rising, but at
an ever-decreasing rate. And language development approaches
closer and closer to the sophistication of users.
Moreover, at some point the elaboration of a high-level Ian-
Past Breakthroughs Solved Accidental Difficulties 187
guage becomes a burden that increases, not reduces, the intel-
lectual task of the user who rarely uses the esoteric constructs.
Time-sharing. Most observers credit time-sharing with a major
improvement in the productivity of programmers and in the
quality of their product, although not so large as that brought
by high-level languages.
Time-sharing attacks a distinctly different difficulty. Time-
sharing preserves immediacy, and hence enables us to maintain
an overview of complexity. The slow turnaround of batch pro-
gramming means that we inevitably forget the minutiae, if not
the very thrust, of what we were thinking when we stopped
programming and called for compilation and execution. This in-
terruption of consciousness is costly in time, for we must re-
fresh. The most serious effect may well be the decay of grasp of
all that is going on in a complex system.
Slow turn-around, like machine-language complexities, is
an accidental rather than an essential difficulty of the software
process. The limits of the contribution of time-sharing derive di-
rectly. The principal effect is to shorten system response time.
As it goes to zero, at some point it passes the human threshold
of noticeability, about 100 milliseconds. Beyond that no benefits
are to be expected.
Unified programming environments. Unix and Interlisp, the
first integrated programming environments to come into wide-
spread use, are perceived to have improved productivity by in-
tegral factors. Why?
They attack the accidental difficulties of using programs
together, by providing integrated libraries, unified file formats,
and pipes and filters. As a result, conceptual structures that in
principle could always call, feed, and use one another can in-
deed easily do so in practice.
This breakthrough in turn stimulated the development of
whole toolbenches, since each new tool could be applied to any
programs using the standard formats.
Because of these successes, environments are the subject of
188 No Silver Bullet
much of today's software engineering research. We will look at
their promise and limitations in the next section.
Hopes for the Silver
Now let us consider the technical developments that are most
often advanced as potential silver bullets. What problems do
they address? Are they the problems of essence, or are they re-
mainders of our accidental difficulties? Do they offer revolution-
ary advances, or incremental ones?
Ada and other high-level language advances. One of the most
touted recent developments is the programming language Ada,
a general-purpose, high-level language of the 1980s. Ada indeed
not only reflects evolutionary improvements in language con-
cepts but embodies features to encourage modern design and
modularization concepts. Perhaps the Ada philosophy is more
of an advance than the Ada language, for it is the philosophy of
modularization, of abstract data types, of hierarchical structur-
ing. Ada is perhaps over-rich, the natural product of the process
by which requirements were laid on its design. That is not fatal,
for subset working vocabularies can solve the learning problem,
and hardware advances will give us the cheap MIPS to pay for
the compiling costs. Advancing the structuring of software sys-
tems is indeed a very good use for the increased MIPS our dol-
lars will buy. Operating systems, loudly decried in the 1960s for
their memory and cycle costs, have proved to be an excellent
form in which to use some of the MIPS and cheap memory bytes
of the past hardware surge.
Nevertheless, Ada will not prove to be the silver bullet that
slays the software productivity monster. It is, after all, just an-
other high-level language, and the biggest payoff from such lan-
guages came from the first transition, up from the accidental
complexities of the machine into the more abstract statement
of step-by-step solutions. Once those accidents have been re-
moved, the remaining ones are smaller, and the payoff from
their removal will surely be less.
Hopes for the Silver 189
I predict that a decade from now, when the effectiveness of
Ada is assessed, it will be seen to have made a substantial dif-
ference, but not because of any particular language feature, nor
indeed because of all of them combined. Neither will the new
Ada environments prove to be the cause of the improvements.
Ada's greatest contribution will be that switching to it occa-
sioned training programmers in modern software design tech-
niques.
Object-oriented programming. Many students of the art hold
out more hope for object-oriented programming than for any of
the other technical fads of the day.3 I am among them. Mark
Sherman of Dartmouth notes that we must be careful to distin-
guish two separate ideas that go under that name: abstract data
types and hierarchical types, also called classes. The concept of
the abstract data type is that an object's type should be defined
by a name, a set of proper values, and a set of proper opera-
tions, rather than its storage structure, which should be hidden.
Examples are Ada packages (with private types) or Modula's
modules.
Hierarchical types, such as Simula-67's classes, allow the
definition of general interfaces that can be further refined by
providing subordinate types. The two concepts are orthogo-
nal—there may be hierarchies without hiding and hiding with-
out hierarchies. Both concepts represent real advances in the art
of building software.
Each removes one more accidental difficulty from the pro-
cess, allowing the designer to express the essence of his design
without having to express large amounts of syntactic material
that add no new information content. For both abstract types
and hierarchical types, the result is to remove a higher-order
sort of accidental difficulty and allow a higher-order expression
of design.
Nevertheless, such advances can do no more than to re-
move all the accidental difficulties from the expression of the de-
sign. The complexity of the design itself is essential; and such
190 No Silver Bullet
attacks make no change whatever in that. An order-of-magni-
tude gain can be made by object-oriented programming only if
the unnecessary underbrush of type specification remaining to-
day in our programming language is itself responsible for nine-
tenths of the work involved in designing a program product. I
doubt it.
Artificial intelligence. Many people expect advances in artifi-
cial intelligence to provide the revolutionary breakthrough that
will give order-of-magnitude gains in software productivity and
quality.41 do not. To see why, we must dissect what is meant by
"artificial intelligence" and then see how it applies.
Parnas has clarified the terminological chaos:
Two quite different definitions of AI are in common use today.
AI-1: The use of computers to solve problems that previously could
only be solved by applying human intelligence. AI2: The use of a
specific set of programming techniques known as heuristic or rule-
based programming. In this approach human experts are studied to
determine what heuristics or rules of thumb they use in solving
problems. . . . The program is designed to solve a problem the way
that humans seem to solve it.
The first definition has a sliding meaning. . . . Something can fit
the definition of AI-1 today but, once we see how the program
works and understand the problem, we will not think of it as AI
anymore. . . . Unfortunately I cannot identify a body of technology
that is unique to this field. . . . Most of the work is problem-
specific, and some abstraction or creativity is required to see how to
transfer it.5
I agree completely with this critique. The techniques used
for speech recognition seem to have little in common with those
used for image recognition, and both are different from those
used in expert systems. I have a hard time seeing how image
recognition, for example, will make any appreciable difference
in programming practice. The same is true of speech recogni-
Hopes for the Silver 191
tion. The hard thing about building software is deciding what to
say, not saying it. No facilitation of expression can give more
than marginal gains.
Expert systems technology, AI-2, deserves a section of its
own.
Expert systems. The most advanced part of the artificial intel-
ligence art, and the most widely applied, is the technology for
building expert systems. Many software scientists are hard at
work applying this technology to the software-building environ-
ment.5 What is the concept, and what are the prospects?
An expert system is a program containing a generalized in-
ference engine and a rule base, designed to take input data and
assumptions and explore the logical consequences through the
inferences derivable from the rule base, yielding conclusions
and advice, and offering to explain its results by retracing its rea-
soning for the user. The inference engines typically can deal
with fuzzy or probabilistic data and rules in addition to purely
deterministic logic.
Such systems offer some clear advantages over programmed
algorithms for arriving at the same solutions to the same prob-
lems:
• Inference engine technology is developed in an application-
independent way, and then applied to many uses. One can
justify much more effort on the inference engines. Indeed,
that technology is well advanced.
• The changeable parts of the application-peculiar materials
are encoded in the rule base in a uniform fashion, and tools
are provided for developing, changing, testing, and docu-
menting the rule base. This regularizes much of the com-
plexity of the application itself.
Edward Feigenbaum says that the power of such systems
does not come from ever-fancier inference mechanisms, but
rather from ever-richer knowledge bases that reflect the real
world more accurately. I believe the most important advance of-
fered by the technology is the separation of the application com-
192 No Silver Bullet
plexity from the program itself.
How can this be applied to the software task? In many ways:
suggesting interface rules, advising on testing strategies, remem-
bering bug-type frequencies, offering optimization hints, etc.
Consider an imaginary testing advisor, for example. In its
most rudimentary form, the diagnostic expert system is very like
a pilot's checklist, fundamentally offering suggestions as to
possible causes of difficulty. As the rule base is developed, the
suggestions become more specific, taking more sophisticated
account of the trouble symptoms reported. One can visualize a
debugging assistant that offers very generalized suggestions at
first, but as more and more system structure is embodied in the
rule base, becomes more and more particular in the hypotheses
it generates and the tests it recommends. Such an expert system
may depart most radically from the conventional ones in that its
rule base should probably be hierarchically modularized in the
same way the corresponding software product is, so that as the
product is modularly modified, the diagnostic rule base can be
modularly modified as well.
The work required to generate the diagnostic rules is work
that will have to be done anyway in generating the set of test
cases for the modules and for the system. If it is done in a suit-
ably general manner, with a uniform structure for rules and a
good inference engine available, it may actually reduce the total
labor of generating bring-up test cases, as well as helping in life-
long maintenance and modification testing. In the same way, we
can postulate other advisors—probably many of them and prob-
ably simple ones—for the other parts of the software construc-
tion task.
Many difficulties stand in the way of the early realization of
useful expert advisors to the program developer. A crucial part
of our imaginary scenario is the development of easy ways to
get from program structure specification to the automatic or
semiautomatic generation of diagnostic rules. Even more diffi-
cult and important is the twofold task of knowledge acquisition:
Hopes for the Silver 193
finding articulate, self-analytical experts who know why they do
things; and developing efficient techniques for extracting what
they know and distilling it into rule bases. The essential prereq-
uisite for building an expert system is to have an expert.
The most powerful contribution of expert systems will
surely be to put at the service of the inexperienced programmer
the experience and accumulated wisdom of the best program-
mers. This is no small contribution. The gap between the best
software engineering practice and the average practice is very
wide—perhaps wider than in any other engineering discipline.
A tool that disseminates good practice would be important.
"Automatic" programming. For almost 40 years, people have
been anticipating and writing about "automatic programming,"
the generation of a program for solving a problem from a state-
ment of the problem specifications. Some people today write
as if they expected this technology to provide the next break-
through.7
Parnas implies that the term is used for glamour and not se-
mantic content, asserting,
In short, automatic programming always has been a euphemism
for programming with a higher-level language than was presently
available to the programmer.3
He argues, in essence, that in most cases it is the solution
method, not the problem, whose specification has to be given.
Exceptions can be found. The technique of building gener-
ators is very powerful, and it is routinely used to good advan-
tage in programs for sorting. Some systems for integrating
differential equations have also permitted direct specification of
the problem. The system assessed the parameters, chose from a
library of methods of solution, and generated the programs.
These applications have very favorable properties:
e
The problems are readily characterized by relatively few pa-
rameters.
• There are many known methods of solution to provide a li-
brary of alternatives.
194 No Silver Bullet
• Extensive analysis has led to explicit rules for selecting so-
lution techniques, given problem parameters.
It is hard to see how such techniques generalize to the wider
world of the ordinary software system, where cases with such
neat properties are the exception. It is hard even to imagine how
this breakthrough in generalization could conceivably occur.
Graphical programming. A favorite subject for Ph.D. disser-
tations in software engineering is graphical, or visual, program-
ming, the application of computer graphics to software design.9
Sometimes the promise of such an approach is postulated from
the analogy with VLSI chip design, where computer graphics
plays so fruitful a role. Sometimes the approach is justified by
considering flowcharts as the ideal program design medium,
and providing powerful facilities for constructing them.
Nothing even convincing, much less exciting, has yet emerged
from such efforts. I am persuaded that nothing will.
In the first place, as I have argued elsewhere, the flow chart
is a very poor abstraction of software structure.10 Indeed, it is
best viewed as Burks, von Neumann, and Goldstine's attempt
to provide a desperately needed high-level control language for
their proposed computer. In the pitiful, multipage, connection-
boxed form to which the flow chart has today been elaborated,
it has proved to be essentially useless as a design tool—pro-
grammers draw flow charts after, not before, writing the pro-
grams they describe.
Second, the screens of today are too small, in pixels, to
show both the scope and the resolution of any serious detailed
software diagram. The so-called "desktop metaphor" of today's
workstation is instead an "airplane-seat" metaphor. Anyone
who has shuffled a lapful of papers while seated in coach
between two portly passengers will recognize the difference—
one can see only a very few things at once. The true desktop
provides overview of and random access to a score of pages.
Moreover, when fits of creativity run strong, more than one pro-
grammer or writer has been known to abandon the desktop for
Hopes for the Silver 195
the more spacious floor. The hardware technology will have to
advance quite substantially before the scope of our scopes is suf-
ficient to the software design task.
More fundamentally, as I have argued above, software is
very difficult to visualize. Whether we diagram control flow,
variable scope nesting, variable cross-references, data flow, hi-
erarchical data structures, or whatever, we feel only one dimen-
sion of the intricately interlocked software elephant. If we
superimpose all the diagrams generated by the many relevant
views, it is difficult to extract any global overview. The VLSI
analogy is fundamentally misleading—a chip design is a layered
two-dimensional object whose geometry reflects its essence. A
software system is not.
Program verification. Much of the effort in modern program-
ming goes into the testing and repair of bugs. Is there perhaps
a silver bullet to be found by eliminating the errors at the source,
in the system design phase? Can both productivity and product
reliability be radically enhanced by following the profoundly
different strategy of proving designs correct before the immense
effort is poured into implementing and testing them?
I do not believe we will find the magic here. Program veri-
fication is a very powerful concept, and it will be very important
for such things as secure operating system kernels. The tech-
nology does not promise, however, to save labor. Verifications
are so much work that only a few substantial programs have
ever been verified.
Program verification does not mean error-proof programs.
There is no magic here, either. Mathematical proofs also can be
faulty. So whereas verification might reduce the program-test-
ing load, it cannot eliminate it.
More seriously, even perfect program verification can only
establish that a program meets its specification. The hardest part
of the software task is arriving at a complete and consistent
specification, and much of the essence of building a program is
in fact the debugging of the specification.
196 No Silver Bullet
Environments and tools. How much more gain can be expected
from the exploding researches into better programming envi-
ronments? One's instinctive reaction is that the big-payoff prob-
lems were the first attacked, and have been solved: hierarchical
file systems, uniform file formats so as to have uniform program
interfaces, and generalized tools. Language-specific smart edi-
tors are developments not yet widely used in practice, but the
most they promise is freedom from syntactic errors and simple
semantic errors.
Perhaps the biggest gain yet to be realized in the program-
ming environment is the use of integrated database systems to
keep track of the myriads of details that must be recalled accu-
rately by the individual programmer and kept current in a group
of collaborators on a single system.
Surely this work is worthwhile, and surely it will bear some
fruit in both productivity and reliability. But by its very nature,
the return from now on must be marginal.
Workstations. What gains are to be expected for the software
art from the certain and rapid increase in the power and mem-
ory capacity of the individual workstation? Well, how many
MIPS can one use fruitfully? The composition and editing of
programs and documents is fully supported by today's speeds.
Compiling could stand a boost, but a factor of 10 in machine
speed would surely leave think-time the dominant activity in
the programmer's day. Indeed, it appears to be so now.
More powerful workstations we surely welcome. Magical
enhancements from them we cannot expect.
Promising Attacks on the Conceptual Essence
Even though no technological breakthrough promises to give
the sort of magical results with which we are so familiar in the
hardware area, there is both an abundance of good work going
on now, and the promise of steady, if unspectacular progress.
All of the technological attacks on the accidents of the soft-
Promising Attacks on the Conceptual Essence 197
ware process are fundamentally limited by the productivity
equation:
If, as I believe, the conceptual components of the task are
now taking most of the time, then no amount of activity on the
task components that are merely the expression of the concepts
can give large productivity gains.
Hence we must consider those attacks that address the
essence of the software problem, the formulation of these com-
plex conceptual structures. Fortunately, some of these are very
promising.
Buy versus build. The most radical possible solution for con-
structing software is not to construct it at all.
Every day this becomes easier, as more and more vendors
offer more and better software products for a dizzying variety of
applications. While we software engineers have labored on pro-
duction methodology, the personal computer revolution has
created not one, but many, mass markets for software. Every
newsstand carries monthly magazines which, sorted by ma-
chine type, advertise and review dozens of products at prices
from a few dollars to a few hundred dollars. More specialized
sources offer very powerful products for the workstation and
other Unix markets. Even software tools and environments can
be bought off-the-shelf. I have elsewhere proposed a market-
place for individual modules.
Any such product is cheaper to buy than to build afresh.
Even at a cost of $100,000, a purchased piece of software is cost-
ing only about as much as one programmer-year. And delivery
is immediate! Immediate at least for products that really exist,
products whose developer can refer the prospect to a happy
user. Moreover, such products tend to be much better docu-
mented and somewhat better maintained than homegrown soft-
ware.
The development of the mass market is, I believe, the most
profound long-run trend in software engineering. The cost of
198 No Silver Bullet
software has always been development cost, not replication
cost. Sharing that cost among even a few users radically cuts the
per-user cost. Another way of looking at it is that the use of n
copies of a software system effectively multiplies the productiv-
ity of its developers by n. That is an enhancement of the pro-
ductivity of the discipline and of the nation.
The key issue, of course, is applicability. Can I use an avail-
able off-the-shelf package to do my task? A surprising thing has
happened here. During the 1950s and 1960s, study after study
showed that users would not use off-the-shelf packages for pay-
roll, inventory control, accounts receivable, etc. The require-
ments were too specialized, the case-to-case variation too high.
During the 1980s, we find such packages in high demand and
widespread use. What has changed?
Not really the packages. They may be somewhat more gen-
eralized and somewhat more customizable than formerly, but
not much. Not really the applications, either. If anything, the
business and scientific needs of today are more diverse, more
complicated than those of 20 years ago.
The big change has been in the hardware/software cost ra-
tio. The buyer of a $2-million machine in 1960 felt that he could
afford $250,000 more for a customized payroll program, one that
slipped easily and nondisruptively into the computer-hostile so-
cial environment. Buyers of $50,000 office machines today can-
not conceivably afford customized payroll programs; so they
adapt their payroll procedures to the packages available. Com-
puters are now so commonplace, if not yet so beloved, that the
adaptations are accepted as a matter of course.
There are dramatic exceptions to my argument that the gen-
eralization of the software packages has changed little over the
years: electronic spreadsheets and simple database systems.
These powerful tools, so obvious in retrospect and yet so late
appearing, lend themselves to myriad uses, some quite unor-
thodox. Articles and even books now abound on how to tackle
unexpected tasks with the spreadsheet. Large numbers of ap-
plications that would formerly have been written as custom pro-
Promising Attacks on the Conceptual Essence 199
grams in Cobol or Report Program Generator are now routinely
done with these tools.
Many users now operate their own computers day in and
day out on varied applications without ever writing a program.
Indeed, many of these users cannot write new programs for
their machines, but they are nevertheless adept at solving new
problems with them.
I believe the single most powerful software productivity
strategy for many organizations today is to equip the computer-
naive intellectual workers on the firing line with personal com-
puters and good generalized writing, drawing, file, and spread-
sheet programs, and turn them loose. The same strategy, with
generalized mathematical and statistical packages and some
simple programming capabilities, will also work for hundreds of
laboratory scientists.
Requirements refinement and rapid prototyping. The hardest
single part of building a software system is deciding precisely
what to build. No other part of the conceptual work is so diffi-
cult as establishing the detailed technical requirements, includ-
ing all the interfaces to people, to machines, and to other
software systems. No other part of the work so cripples the re-
sulting system if done wrong. No other part is more difficult to
rectify later.
Therefore the most important function that software build-
ers do for their clients is the iterative extraction and refinement
of the product requirements. For the truth is, the clients do not
know what they want. They usually do not know what ques-
tions must be answered, and they almost never have thought of
the problem in the detail that must be specified. Even the simple
answer—"Make the new software system work like our old
manual information-processing system"—is in fact too simple.
Clients never want exactly that. Complex software systems are,
moreover, things that act, that move, that work. The dynamics
of that action are hard to imagine. So in planning any software
activity, it is necessary to allow for an extensive iteration be-
tween the client and the designer as part of the system definition.
200 No Silver Bullet
I would go a step further and assert that it is really impos-
sible for clients, even those working with software engineers, to
specify completely, precisely, and correctly the exact require-
ments of a modern software product before having built and
tried some versions of the product they are specifying.
Therefore one of the most promising of the current techno-
logical efforts, and one which attacks the essence, not the accidents,
of the software problem, is the development of approaches and
tools for rapid prototyping of systems as part of the iterative
specification of requirements.
A prototype software system is one that simulates the im-
portant interfaces and performs the main functions of the in-
tended system, while not being necessarily bound by the same
hardware speed, size, or cost constraints. Prototypes typically
perform the mainline tasks of the application, but make no at-
tempt to handle the exceptions, respond correctly to invalid in-
puts, abort cleanly, etc. The purpose of the prototype is to make
real the conceptual structure specified, so that the client can test
it for consistency and usability.
Much of present-day software acquisition procedures rests
upon the assumption that one can specify a satisfactory system
in advance, get bids for its construction, have it built, and install
it. I think this assumption is fundamentally wrong, and that
many software acquisition problems spring from that fallacy.
Hence they cannot be fixed without fundamental revision, one
that provides for iterative development and specification of pro-
totypes and products.
Incremental development—grow, not build, software. I still
remember the jolt I felt in 1958 when I first heard a friend talk
about building a program, as opposed to writing one. In a flash
he broadened my whole view of the software process. The met-
aphor shift was powerful, and accurate. Today we understand
how like other building processes the construction of software
is, and we freely use other elements of the metaphor, such as
specifications, assembly of components, and scaffolding.
Promising Attacks on the Conceptual Essence 201
The building metaphor has outlived its usefulness. It is time
to change again. If, as I believe, the conceptual structures we
construct today are too complicated to be accurately specified in
advance, and too complex to be built faultlessly, then we must
take a radically different approach.
Let us turn to nature and study complexity in living things,
instead of just the dead works of man. Here we find constructs
whose complexities thrill us with awe. The brain alone is intri-
cate beyond mapping, powerful beyond imitation, rich in diver-
sity, self-protecting, and self-renewing. The secret is that it is
grown, not built.
So it must be with our software systems. Some years ago
Harlan Mills proposed that any software system should be
grown by incremental development.11 That is, the system should
first be made to run, even though it does nothing useful except
call the proper set of dummy subprograms. Then, bit by bit it is
fleshed out, with the subprograms in turn being developed into
actions or calls to empty stubs in the level below.
I have seen the most dramatic results since I began urging
this technique on the project builders in my software engineer-
ing laboratory class. Nothing in the past decade has so radically
changed my own practice, or its effectiveness. The approach ne-
cessitates top-down design, for it is a top-down growing of the
software. It allows easy backtracking. It lends itself to early pro-
totypes. Each added function and new provision for more com-
plex data or circumstances grows organically out of what is
already there.
The morale effects are startling. Enthusiasm jumps when
there is a running system, even a simple one. Efforts redouble
when the first picture from a new graphics software system ap-
pears on the screen, even if it is only a rectangle. One always
has, at every stage in the process, a working system. I find that
teams can grow much more complex entities in four months than
they can build.
The same benefits can be realized on large projects as on my
small ones.12
202 No Silver Bullet
Great designers. The central question of how to improve the
software art centers, as it always has, on people.
We can get good designs by following good practices instead
of poor ones. Good design practices can be taught. Program-
mers are among the most intelligent part of the population, so
they can learn good practice. Thus a major thrust in the United
States is to promulgate good modern practice. New curricula,
new literature, new organizations such as the Software Engi-
neering Institute, all have come into being in order to raise the
level of our practice from poor to good. This is entirely
Nevertheless, I do not believe we can make the next step
upward in the same way. Whereas the difference between poor
conceptual designs and good ones may lie in the soundness of
design method, the difference between good designs and great
ones surely does not. Great designs come from great designers.
Software construction is a creative process. Sound methodology
can empower and liberate the creative mind; it cannot enflame
or inspire the drudge.
The differences are not minor—it is rather like Salieri and
Mozart, Study after study shows that the very best designers
produce structures that are faster, smaller, simpler, cleaner, and
produced with less effort. The differences between the great and
the average approach an order of magnitude.
A little retrospection shows that although many fine, useful
software systems have been designed by committees and built
by multipart projects, those software systems that have excited
passionate fans are those that are the products of one or a few
designing minds, great designers. Consider Unix, APL, Pascal,
Modula, the Smalltalk interface, even Fortran; and contrast with
Cobol, PL/I, Algol, MVS/370, and MS-DOS (Fig. 16.1).
Hence, although I strongly support the technology transfer
and curriculum development efforts now underway, I think the
most important single effort we can mount is to develop ways
to grow great designers.
No software organization can ignore this challenge. Good
managers, scarce though they be, are no scarcer than good de-
Promising Attacks on the Conceptual Essence 203
Yes No
Unix Cobol
APL PL/1
Pascal Algol
Modulo MVS/370
Smalltalk MS-DOS
Fortran
Fig. 16.1 Exciting products
signers. Great designers and great managers are both very rare.
Most organizations spend considerable effort in finding and cul-
tivating the management prospects; I know of none that spends
equal effort in finding and developing the great designers upon
whom the technical excellence of the products will ultimately
depend.
My first proposal is that each software organization must
determine and proclaim that great designers are as important to
its success as great managers are, and that they can be expected
to be similarly nurtured and rewarded. Not only salary, but the
perquisites of recognition—office size, furnishings, personal
technical equipment, travel funds, staff support—must be fully
equivalent.
How to grow great designers? Space does not permit a
lengthy discussion, but some steps are obvious:
• Systematically identify top designers as early as possible.
The best are often not the most experienced.
• Assign a career mentor to be responsible for the develop-
ment of the prospect, and keep a careful career file.
• Devise and maintain a career development plan for each
prospect, including carefully selected apprenticeships with
top designers, episodes of advanced formal education, and
short courses, all interspersed with solo design and techni-
cal leadership assignments.
• Provide opportunities for growing designers to interact with
and stimulate each other.
17
"No Silver Bullet" Refired
17
"No Silver Bullet" Refired
Every bullet has its billet.
WILLIAM III OF ENGLAND, PRINCE OF ORANGE
Whoever thinks a faultless piece to see,
Thinks what ne'er was, nor is, nor e'er shall be.
ALEXANDER POPE, AN ESSAY ON CRITICISM
Assembling a structure from ready-made parts, 1945
The Bettman Archive
207
208 "No Silver Bullet" Refired
On Werewolves and Other Legendary Terrors
"No Silver Bullet—Essence and Accidents of Software Engineer-
ing" (now Chapter 16) was originally an invited paper for the
IFIP '86 conference in Dublin, and it was published in those pro-
ceedings.1 Computer magazine reprinted it, behind a gothic
cover, illustrated with stills from films such as The Werewolf of
London.2 They also provided an explanatory sidebar "To Slay the
Werewolf," setting forth the (modern) legend that only silver
bullets will do. I was not aware of the sidebar and illustrations
before publication, and I never expected a serious technical pa-
per to be so embellished.
Computer's editors were expert in achieving their desired ef-
fect, however, and many people seem to have read the paper. I
have therefore chosen yet another werewolf picture for that
chapter, an ancient depiction of an almost comical creature. I
hope this less garish picture will have the same salutary effect.
There is Too a Silver Bullet—AND HERE IT IS!
"No Silver Bullet" asserts and argues that no single software en-
gineering development will produce an order-of-magnitude im-
provement in programming productivity within ten years (from
the paper's publication in 1986). We are now nine years into
that decade, so it is timely to see how this prediction is hold-
ing up.
Whereas The Mythical Man-Month generated many citations
but little argument, "No Silver Bullet" has occasioned rebuttal
papers, letters to journal editors, and letters and essays that con-
tinue to this day.3 Most of these attack the central argument that
there is no magical solution, and my clear opinion that there
cannot be one. Most agree with most of the arguments in
"NSB," but then go on to assert that there is indeed a silver bul-
let for the software beast, which the author has invented. As I
reread the early responses today, I can't help noticing that the
nostrums pushed so vigorously in 1986 and 1987 have not had
the dramatic effects claimed.
Obscure Writing Will Be Misunderstood 209
I buy hardware and software chiefly by the "happy user"
test—conversations with bona fide cash-paying customers who
use the product and are pleased. Likewise, I shall most readily
believe a silver bullet has materialized when a bona fide indepen-
dent user steps forth and says, "I used this methodology, tool,
or product, and it gave me a tenfold improvement in software
productivity."
Many correspondents have made valid emendations or clar-
ifications. Some have undertaken point-by-point analysis and
rebuttal, for which I am grateful. In this chapter, I shall share
the improvements and address the rebuttals.
Obscure Writing Will Be Misunderstood
Some writers show that I failed to make some arguments clear.
Accident. The central argument of "NSB" is as clearly stated in
the Abstract to Chapter 16 as I know how to put it. Some have
been confused, however, by the terms accident and accidental,
which follow an ancient usage going back to Aristotle.4 By acci-
dental, I did not mean occurring by chance, nor misfortunate, but
more nearly incidental, or appurtenant.
I would not denigrate the accidental parts of software con-
struction; instead I follow the English dramatist, detective story
writer, and theologian Dorothy Sayers in seeing all creative ac-
tivity to consist of (1) the formulation of the conceptual con-
structs, (2) implementation in real media, and (3) interactivity
with users in real uses.5 The part of software building I called
essence is the mental crafting of the conceptual construct; the
part I called accident is its implementation process.
A question of fact. It seems to me (although not to everyone)
that the truthfulness of the central argument boils down to a
question of fact: What fraction of total software effort is now as-
sociated with the accurate and orderly representation of the con-
ceptual construct, and what fraction is the effort of mentally
crafting the constructs? The finding and fixing of flaws falls
210 "No Silver Bullet" Retired
partly in each fraction, according to whether the flaws are con-
ceptual, such as failing to recognize some exception, or repre-
sentational, such as a pointer mistake or a memory allocation
mistake.
It is my opinion, and that is all, that the accidental or rep-
resentational part of the work is now down to about half or less
of the total. Since this fraction is a question of fact, its value
could in principle be settled by measurement.6 Failing that, my
estimate of it can be corrected by better informed and more cur-
rent estimates. Significantly, no one who has written publicly or
privately has asserted that the accidental part is as large as 9/10.
"NSB" argues, indisputably, that if the accidental part of the
work is less than 9/10 of the total, shrinking it to zero (which
would take magic) will not give an order of magnitude produc-
tivity improvement. One must attack the essence.
Since "NSB," Bruce Blum has drawn my attention to the
1959 work of Herzberg, Mausner, and Sayderman.7 They find
that motivational factors can increase productivity. On the other
hand, environmental and accidental factors, no matter how pos-
itive, cannot; but these factors can decrease productivity when
negative. "NSB" argues that much software progress has been
the removal of such negative factors: stunningly awkward ma-
chine languages, batch processing with long turnaround times,
poor tools, and severe memory constraints.
Are the essential difficulties therefore hopeless? An excellent
1990 paper by Brad Cox, "There Is a Silver Bullet," argues elo-
quently for the reusable, interchangeable component approach
as an attack on the conceptual essence of the problem.81 enthu-
siastically agree.
Cox however misunderstands "NSB" on two points. First,
he reads it as asserting that software difficulties arise "from
some deficiency in how programmers build software today."
My argument was that the essential difficulties are inherent in
the conceptual complexity of the software functions to be de-
signed and built at any time, by any method. Second, he (and
Obscure Writing Will Be Misunderstood 211
others) read "NSB" as asserting that there is no hope of attack-
ing the essential difficulties of software building. That was not
my intent. Crafting the conceptual construct does indeed have
as inherent difficulties complexity, conformity, changeability,
and invisibility. The troubles caused by each of these difficulties
can, however, be ameliorated.
Complexity is by levels. For example, complexity is the most
serious inherent difficulty, but not all complexity is inevitable.
Much, but not all, of the conceptual complexity in our software
constructs comes from the arbitrary complexity of the applica-
tions themselves. Indeed, Lars S0dahl of MYSIGMA S0dahl and
Partners, a multinational management consulting firm, writes:
In my experience most of the complexities which are encountered in
systems work are symptoms of organizational malfunctions. Trying
to model this reality with equally complex programs is actually to
conserve the mess instead of solving the problems.
Steve Lukasik of Northrop argues that even organizational
complexity is perhaps not arbitrary but may be susceptible to or-
dering principles:
/ trained as a physicist and thus see "complex" things as susceptible
to description in terms of simpler concepts. Now you may be right;
I will not assert that all complex things are susceptible to ordering
principles. . . . by the same rules of argument you cannot assert
that they can not be.
. . . Yesterday's complexity is tomorrow's order. The complexity of
molecular disorder gave way to the kinetic theory of gases and the
three laws of thermodynamics. Now software may not ever reveal
those kinds of ordering principles, but the burden is on you to ex-
plain why not. I am not being obtuse or argumentative. I believe
that someday the "complexity" of software will be understood in
terms of some higher order notions (invariants to the physicist).
212 "No Silver Bullet" Refired
I have not undertaken the deeper analysis Lukasik properly
calls for. As a discipline, we need an extended information the-
ory that quantifies the information content of static structures,
just as Shannon's theory does for communicated streams. That
is quite beyond me. To Lukasik I simply respond that system
complexity is a function of myriad details that must each be
specified exactly, either by some general rule or detail-by-detail,
but not just statistically. It seems very unlikely that uncoordi-
nated works of many minds should have enough coherence to
be exactly described by general rules.
Much of the complexity in a software construct is, however,
not due to conformity to the external world but rather to the im-
plementation itself—its data structures, its algorithms, its con-
nectivity. Growing software in higher-level chunks, built by
someone else or reused from one's own past, avoids facing
whole layers of complexity. "NSB" advocates a wholehearted at-
tack on the problem of complexity, quite optimistic that progress
can be made. It advocates adding necessary complexity to a soft-
ware system:
" Hierarchically, by layered modules or objects
• Incrementally, so that the system always works.
Harel's Analysis
David Harel, in the 1992 paper, "Biting the Silver Bullet," un-
dertakes the most careful analysis of "NSB" that has been pub-
lished.9
Pessimism vs. optimism vs. realism. Harel sees both "NSB"
and Parnas's 1984 "Software Aspects of Strategic Defense Sys-
tems,"10 as "far too bleak." So he aims to illuminate the brighter
side of the coin, subtitling his paper "Toward a Brighter Future
for System Development." Cox as well as Harel reads "NSB" as
pessimistic, and he says, "But if you view these same facts from
a new perspective, a more optimistic conclusion emerges." Both
misread the tone.
Harel's Analysis 213
First, my wife, my colleagues, and my editors find me to err
far more often in optimism than in pessimism. I am, after all, a
programmer by background, and optimism is an occupational
disease of our craft.
"NSB" says explicitly "As we look to the horizon of a decade
hence, we see no silver bullet. . . . Skepticism is not pessimism,
however. . . . There is no royal road, but there is a road." It
forecasts that the innovations under way in 1986, if developed
and exploited, would together indeed achieve an order-of-mag-
nitude improvement in productivity. As the 1986-1996 decade
proceeds, this prediction appears, if anything, too optimistic
rather than too gloomy.
Even if "NSB" were universally seen as pessimistic, what is
wrong with that? Is Einstein's statement that nothing can travel
faster than the speed of light "bleak" or "gloomy?" How about
Godel's result that some things cannot be computed? "NSB" un-
dertakes to establish that "the very nature of software makes
it unlikely that there will ever be any silver bullets." Turski, in
his excellent response paper at the IFIP Conference, said elo-
quently:
Of all misguided scientific endeavours, none are more pathetic than
the search for the philosophers' stone, a substance supposed to
change base metals into gold. The supreme object of alchemy, ar-
dently pursued by generations of researchers generously funded by
secular and spiritual rulers, is an undiluted extract of wishful
thinking, of the common assumption that things are as we would
like them to be. It is a very human belief. It takes a lot of effort to
accept the existence of insoluble problems. The wish to see a way
out, against all odds, even when it is proven that it does not exist,
is very, very strong. And most of us have a lot of sympathy for
these courageous souls who try to achieve the impossible. And so
it continues. Dissertations on squaring a circle are being written.
Lotions to restore lost hair are concocted and sell well. Meth-
ods to improve software productivity are hatched and sell very
well.
214 "No Silver Bullet" Refired
All too often we are inclined to follow our own optimism (or exploit
the optimistic hopes of our sponsors). All too often we are willing
to disregard the voice of reason and heed the siren calls of panacea
pushers.n
Turski and I both insist that pipe-dreaming inhibits forward prog-
ress and wastes effort.
"Gloom" themes. Harel perceives gloom in "NSB" to arise
from three themes:
• Sharp separation into essence and accident
• Treatment of each silver bullet candidate in isolation
• Predicting for only 10 years, instead of a long enough time
in which "to expect any significant improvement."
As to the first, that is the whole point of the paper. I still
believe this separation is absolutely central to understanding
why software is hard. It is a sure guide as to what kinds of at-
tacks to make.
As to treating candidate bullets in isolation, "NSB" does so
indeed. The various candidates have been proposed one by one,
with extravagant claims for each one by itself. It is fair to assess
them one by one. It is not the techniques I oppose, it is expect-
ing them to work magic. Glass, Vessey, and Conger in their 1992
paper offer ample evidence that the vain search for a silver bullet
has not yet ended.12
As to choosing 10 years versus 40 years as a prediction pe-
riod, the shorter period was in part a concession that our pre-
dictive powers have never been good beyond a decade. Which
of us in 1975 predicted the microcomputer revolution of the
1980s?
There are other reasons for the decade limit: the claims
made for candidate bullets all have had a certain immediacy
about them. I recollect none that said "Invest in my nostrum,
and you will start winning after 10 years." Moreover, hardware
performance/price ratios have improved by perhaps a hundred-
fold per decade, and the comparison, though quite invalid, is
Harel's Analysis 215
subconsciously inevitable. We will surely make substantial prog-
ress over the next 40 years; an order of magnitude over 40 years
is hardly magical.
Harel's thought experiment. Harel proposes a thought exper-
iment in which he postulates "NSB" as having been written in
1952, instead of 1986, but asserting the same propositions. This
he uses as a reducto ad absurdum to argue against attempting to
separate essence from accident.
The argument doesn't work. First, "NSB" begins by assert-
ing that the accidental difficulties grossly dominated the essen-
tial ones in 1950s programming, that they no longer do so, and
that eliminating them has effected orders-of-magnitude im-
provements. Translating that argument back 40 years is un-
reasonable; one can hardly imagine asserting in 1952 that the
accidental difficulties do not occasion a major part of the effort.
Second, the state of affairs Harel imagines to have prevailed
in the 1950s is inaccurate:
That was the time when instead of grappling with the design of
large, complex systems, programmers were in the business of de-
veloping conventional one-person programs (which would be on the
order of 100—200 lines in a modern programming language) that
were to carry out limited algorithmic tasks. Given the technology
and methodology available then, such tasks were similarly formi-
dable. Failures, errors, and missed deadlines were all around.
He then describes how the postulated failures, errors, and
missed deadlines in the conventional little one-person programs
were improved by an order of magnitude over the next 25 years.
But the state of the art in the 1950s was not in fact small one-
person programs. In 1952, the Univac was at work processing
the 1950 census with a complex program developed by about
eight programmers.13 Other machines were doing chemical dy-
namics, neutron diffusion calculations, missile performance
calculations, etc.14 Assemblers, relocating linkers and loaders,
floating-point interpretive systems, etc. were in routine use.15
216 "No Silver Bullet" Refired
By 1955 people were building 50 to 100 man-year business pro-
grams.16 By 1956 General Electric had in operation a payroll sys-
tem in its Louisville appliance plant with more than 80,000
words of program. By 1957, the SAGE ANFSQ/7 air defense
computer had been running two years, and a 75,000 instruction
communications-based, fail-safe-duplexed real-time system was
in operation in 30 sites.17 One can hardly maintain that it is evo-
lution of techniques for one-person programs that chiefly de-
scribes software engineering efforts since 1952.
AND HERE IT IS. Harel goes on to offer his own silver bullet,
a modeling technique called "The Vanilla Framework." The ap-
proach itself is not described in enough detail for evaluation, but
reference is given to a paper, and to a technical report to appear
in book form in due time.18 Modeling does address the essence,
the proper crafting and debugging of concepts, so it is possible
that the Vanilla Framework will be revolutionary. I hope so. Ken
Brooks reports he found it a helpful methodology when he tried
it for a real task.
Invisibility. Harel argues strongly that much of the conceptual
construct of software is inherently topological in nature and
these relationships have natural counterparts in spatial/graphi-
cal representations:
Using appropriate visual formalisms can have a spectacular effect
on engineers and programmers. Moreover, this effect is not limited
to mere accidental issues; the quality and expedition of their very
thinking was found to be improved. Successful system develop-
ment in the future will revolve around visual representations. We
will first conceptualize, using the "proper" entities and relation-
ships, and then formulate and reformulate our conceptions as a se-
ries of increasingly more comprehensive models represented in an
appropriate combination of visual languages. A combination it
must be, since system models have several facets, each of which con-
jures up different kinds of mental images.
Jones's Point—Productivity Follows Quality 217
. . . . Some aspects of the modeling process have not been as forth-
coming as others in lending themselves to good visualization. Al-
gorithmic operations on variables and data structures, for example,
will probably remain textual.
Harel and I are quite close. What I argued is that software struc-
ture is not embedded in three-space, so there is no natural single
mapping from a conceptual design to a diagram, whether in two
dimensions or more. He concedes, and I agree, that one needs
multiple diagrams, each covering some distinct aspect, and that
some aspects don't diagram well at all.
I completely share his enthusiasm for using diagrams as
thought and design aids. I have long enjoyed asking candidate
programmers, "Where is next November?" If the question is too
cryptic, then, "Tell me about your mental model of the calen-
dar." The really good programmers have strong spatial senses;
they usually have geometric models of time; and they quite
often understand the first question without elaboration. They
have highly individualistic models.
Jones's Point—Productivity Follows Quality
Capers Jones, writing first in a series of memoranda and later in
a book, offers a penetrating insight, which has been stated by
several of my correspondents. "NSB," like most writings at the
time, was focused on productivity, the software output per unit
of input. Jones says, "No. Focus on quality, and productivity will
follow."19 He argues that costly and late projects invest most of
the extra work and time in finding and repairing errors in spec-
ification, in design, in implementation. He offers data that show
a strong correlation between lack of systematic quality controls
and schedule disasters. I believe it. Boehm points out that pro-
ductivity drops again as one pursues extreme quality, as in
IBM's space-shuttle software.
Coqui similarly argues that systematic software develop-
ment disciplines were developed in response to quality concerns
218 "No Silver Bullet" Refired
(especially avoidance of major disasters) rather than productiv-
ity concerns.
But note: the goal of applying Engineering principles to Software
production in the 1970s was to increase the Quality, Testability,
Stability, and Predictability of software products—not necessarily
the efficiency of Software production.
The driving force to use Software Engineering principles in soft-
ware production was the fear of major accidents that might be
caused by having uncontrollable artists responsible for the devel-
opment of ever more complex systems.20
So What Has Happened to Productivity?
Productivity numbers. Productivity numbers are very hard to
define, hard to calibrate, and hard to find. Capers Jones believes
that for two equivalent COBOL programs written 10 years
apart, one without structured methodology and one with, the
gain is a factor of three.
Ed Yourdon says, "I see people getting a fivefold improve-
ment due to workstations and software tools." Tom DeMarco
believes "Your expectation of an order-of-magnitude improve-
ment in 10 years, due to the whole basket of techniques, was
optimistic. I haven't seen organizations making an order-of-
magnitude improvement."
Shrink-wrapped software—Buy; don't build. One 1986 assess-
ment in "NSB" has, I think, proved to be correct: "The devel-
opment of the mass market is . . . the most profound long-run
trend in software engineering." From the discipline's viewpoint,
the mass-market software is almost a new industry compared to
that of the development of custom software, whether in-house
or out-house. When packages sell in the millions—or even the
thousands—quality, timeliness, product performance, and sup-
port cost become dominant issues, rather than the development
cost that is so crucial for custom
So What Has Happened to Productivity? 219
Power tools for the mind. The most dramatic way to improve
the productivity of management information systems (MIS) pro-
grammers is to go down to your local computer store and buy
off the shelf what they would have built. This is not ridiculous;
the availability of cheap, powerful shrink-wrapped software has
met many needs that formerly would have occasioned custom
packages. These power tools for the mind are more like electric
drills, saws, and sanders than they are like big complex produc-
tion tools. The integration of these into compatible and cross-
linked sets such as Microsoft Works and the better-integrated
ClarisWorks give immense flexibility. And like the homeowner's
collection of power hand tools, frequent use of a small set, for
many different tasks, develops familiarity. Such tools must em-
phasize ease of use for the casual user, not the professional.
Ivan Selin, Chairman of American Management Systems,
Inc., wrote me in 1987:
I quibble with your statement that packages have not really changed
that much. . . : I think you too lightly throw off the major impli-
cations of your observation that, [the software packages] "may be
somewhat more generalized and somewhat more customizable than
formerly, but not much." Even accepting this statement at face
value, I believe that the users see the packages as being both more
generalized and easier to customize, and that this perception leads
the users to be much more amenable to packages. In most cases that
my company finds, it is the [end] users, not the software people,
who are reluctant to use packages because they think they will lose
essential features or functions, and hence the prospect of easy cus-
tomization is a big selling point to them.
I think Selin is quite right—I underestimated both the degree of
package customizability and its importance.
Object-Oriented Programming—Will a Brass Bullet Do?
Building with bigger pieces. The illustration opening this
chapter reminds us that if one assembles a set of pieces, each of
220 "No Silver Bullet" Retired
which may be complex, and all of which are designed to have
smooth interfaces, quite rich structures go together rapidly.
One view of object-oriented programming is that it is a dis-
cipline that enforces modularity and clean interfaces. A second
view emphasizes encapsulation, the fact that one cannot see,
much less design, the inner structure of the pieces. Another
view emphasizes inheritance, with its concomitant hierarchical
structure of classes, with virtual functions. Yet another view em-
phasizes strong abstract data-typing, with its assurance that a par-
ticular data-type will be manipulated only by operations proper
to it.
Now any of these disciplines can be had without taking the
whole Smalltalk or C + + package—many of them predated ob-
ject-oriented technology. The attractiveness of object-oriented
approach is that of a multivitamin pill: in one fell swoop (that is,
programmer retraining), one gets them all. It is a very promising
concept.
Why has object-oriented technique grown slowly? In the nine
years since "NSB," the expectancy has steadily grown. Why has
growth been slow? Theories abound. James Coggins, author for
four years of the column, "The Best of comp.lang.c + + " in The
C++ Report, offers this explanation:
The problem is that programmers in O-O have been experimenting
in incestuous applications and aiming low in abstraction, instead
of high. For example, they have been building classes such as
linked-list or set instead of classes such as user-interface or ra-
diation beam or finite-element model. Unfortunately the self-
same strong type-checking in C++ that helps programmers to
avoid errors also makes it hard to build big things out of little
ones.21
He goes back to the basic software problem, and argues that one
way to address unmet software needs is to increase the size of
the intelligent workforce by enabling and coopting our clients.
This argues for top-down design: ._._
Object-Oriented Programming—Will a Brass Bullet Do? 221
// we design large-grained classes that address concepts our clients
are already working with, they can understand and question the
design as it grows, and they can cooperate in the design of test
cases. My ophthalmology collaborators don't care about stacks; they
do care about Legendre polynomial shape descriptions of corneas.
Small encapsulations yield small benefits.
David Parnas, whose paper was one of the origins of object-ori-
ented concepts, sees the matter differently. He writes me:
The answer is simple. It is because [O-O] has been tied to a variety
of complex languages. Instead of teaching people that O-O is a type
of design, and giving them design principles, people have taught
that O-O is the use of a particular tool. We can write good or bad
programs with any tool. Unless we teach people how to design, the
languages matter very little. The result is that people do bad de-
signs with these languages and get very little value from them. If
the value is small, it won't catch on.
Front-loaded costs, down-stream benefits. My own belief is
that object-oriented techniques have a peculiarly severe case of
a malady that characterizes many methodological improve-
ments. The up-front costs are very substantial—primarily re-
training programmers to think in a quite new way, but also extra
investment in fashioning functions into generalized classes. The
benefits, which I think are real and not merely putative, occur
all along the development cycle; but the big benefits pay off dur-
ing successor building, extension, and maintenance activities.
Coggins says, "Object-oriented techniques will not make the
first project development any faster, or the next one. The fifth
one in that family will go blazingly fast."22
Betting real up-front money for the sake of projected but iffy
benefits later is what investors do every day. In many program-
ming organizations, however, it requires real managerial cour-
age, a commodity much scarcer than technical competence or
administrative proficiency. I believe the extreme degree of cost
front-loading and benefit back-loading is the largest single factor
222 "No Silver Bullet" Refired
slowing the adoption of O-O techniques. Even so, C + + seems
to be steadily replacing C in many communities.
What About Reuse?
The best way to attack the essence of building software is not to
build it at all. Package software is only one of the ways of doing
this. Program reuse is another. Indeed, the promise of easy
reuse of classes, with easy customization via inheritance, is one
of the strongest attractions of object-oriented techniques.
As is so often the case, as one gets some experience with a
new way of doing business the new mode is not so simple as
first appears.
Of course, programmers have always reused their own han-
diwork. Jones says,
Most experienced programmers have private libraries which allow
them to develop software with about 30% reused code by volume.
Reusability at the corporate level aims for 75% reused code by vol-
ume, and requires special library and administrative support. Cor-
porate reusable code also implies changes in project accounting and
measurement practices to give credit for reusability.23
W. Huang proposed organizing software factories with a matrix
management of functional specialists, so as to harness the nat-
ural propensity of each to reuse his own code.24
Van Snyder of JPL points out to me that the mathematical
software community has a long tradition of reusing software:
We conjecture that barriers to reuse are not on the producer side,
but on the consumer side. If a software engineer, a potential con-
sumer of standardized software components, perceives it to be more
expensive to find a component that meets his need, and so verify,
than to write one anew, a new, duplicative component will be writ-
ten. Notice we said perceives above. It doesn't matter what the true
cost of reconstruction is.
What About- Reuse? 223
Reuse has been successful for mathematical software for two rea-
sons: (I) It is arcane, requiring an enormous intellectual input per
line of code; and (2) there is a rich and standard nomenclature,
namely mathematics, to describe the functionality of each compo-
nent. Thus the cost to reconstruct a component of mathematical
software is high, and the cost to discover the functionality of an
existing component is low. The long tradition of professional jour-
nals publishing and collecting algorithms, and offering them at
modest cost, and commercial concerns offering very high quality
algorithms at somewhat higher but still modest cost, makes discov-
ering a component that meets one's need simpler than in many
other disciplines, where it is sometimes not possible to specify one's
need precisely and tersely. These factors collaborate to make it more
attractive to reuse rather than to reinvent mathematical software.
The same reuse phenomenon is found among several com-
munities, such as those that build codes for nuclear reactors, cli-
mate models, and ocean models, and tor the same reasons. The
communities each grew up with the same textbooks and stan-
dard notations.
How does corporate-level reuse fare today? Lots of study; rel-
atively little practice in the United States; anecdotal reports of
more reuse abroad.25
Jones reports that all of his firm's clients with over 5000 pro-
grammers have formal reuse research, whereas fewer than 10
percent of the clients with under 500 programmers do.26 He re-
ports that in industries with the greatest reuse potential, reus-
ability research (not deployment) "is active and energetic, even
if not yet totally successful." Ed Yourdon reports a software
house in Manila that has 50 of its 200 programmers building
only reusable modules for the rest to use; "I've seen a few
cases—adoption is due to organizational factors such as the re-
ward structure, not technical factors."
DeMarco tells me that the availability of mass-market pack-
ages and their suitability as providers of generic functions such
as database systems has substantially reduced both the pressure
224 "No Silver Bullet" Refired
and the marginal utility of reusing modules of one's application
code. "The reusable modules tended to be the generic functions
anyway."
Parnas writes,
Reuse is something that is far easier to say than to do. Doing it
requires both good design and very good documentation. Even
when we see good design, which is still infrequently, we won't see
the components reused without good documentation.
Ken Brooks comments on the difficulty of anticipating which
generalization will prove necessary: "I keep having to bend
things even on the fifth use of my own personal user-interface
library."
Real reuse seems to be just beginning. Jones reports that a
few reusable code modules are being offered on the open mar-
ket at prices between 1 percent and 20 percent of the normal de-
velopment costs.27 DeMarco says,
I am becoming very discouraged about the whole reuse phenome-
non. There is almost a total absence of an existence theorem for
reuse. Time has confirmed that there is a big expense in making
things reusable.
Yourdon estimates the big expense: "A good rule of thumb
is that such reusable components will take twice the effort of a
'one-shot' component."281 see that expense as exactly the effort
of productizing the component, discussed in Chapter 1. So my
estimate of the effort ratio would be threefold.
Clearly we are seeing many forms and varieties of reuse, but
not nearly so much of it as we had expected by now. There is
still a lot to learn.
Learning Large Vocabularies—A Predictable but
Unpredicted Problem for Software Reuse
The higher the level at which one thinks, the more numerous
the primitive thought-elements one has to deal with. So pro-
Learning Large Vocabularies—A Problem for Software Reuse 225
gramming languages are much more complex than machine
languages, and natural languages are more complex still.
Higher-level languages have larger vocabularies, more complex
syntax, and richer semantics.
As a discipline, we have not pondered the implications of
this fact for program reuse. To improve quality and productivity,
we want to build programs by composing chunks of debugged
function that are substantially higher than statements in pro-
gramming languages. Therefore, whether we do this by object
class libraries or procedure libraries, we must face the fact that
we are radically raising the sizes of our programming vocabu-
laries. Vocabulary learning constitutes no small part of the in-
tellectual barrier to reuse.
So today people have class libraries with over 3000 mem-
bers. Many objects require specification of 10 to 20 parameters
and option variables. Anyone programming with that library
must, learn the syntax (the external interfaces) and the semantics
(the detailed functional behavior) of its members if they are to
achieve all of the potential reuse.
This task is far from hopeless. Native speakers routinely use
vocabularies of over 10,000 words, educated people far more.
Somehow we learn the syntax and very subtle semantics. We
correctly differentiate among giant, huge, vast, enormous, mam-
moth; people just do not speak of mammoth deserts or vast ele-
phants.
We need research to appropriate for the software reuse
problem the large body of knowledge as to how people acquire
language. Some of the lessons are immediately obvious:
• People learn in sentence contexts, so we need to publish
many examples of composed products, not just libraries of
parts.
• People do not memorize anything but spelling. They learn
syntax and semantics incrementally, in context, by use.
People group word composition rules by syntactic classes,
not by compatible subsets of objects.
226 "No Silver Bullet" Refired
Net on Bullets—Position Unchanged
So we come back to fundamentals. Complexity is the business
we are in, and complexity is what limits us. R. L. Glass, writing
in 1988, accurately summarizes my 1995 views:
So what, in retrospect, have Parnas and Brooks said to us? That
software development is a conceptually tough business. That magic
solutions are not just around the corner. That it is time for the prac-
titioner to examine evolutionary improvements rather than to
wait—or hope—for revolutionary ones.
Some in the software field find this to be a discouraging picture.
They are the ones who still thought breakthroughs were near at
hand.
But some of us—those of us crusty enough to think that we are
realists—see this as a breath of fresh air. At last, we can focus on
something a little more viable than pie in the sky. Now, perhaps,
we can get on with the incremental improvements to software pro-
ductivity that are possible, rather than waiting for the break-
throughs that are not likely to ever come.29
18
Propositions of The
Mythical Man-Month:
True or False?
18
Propositions of The
Mythical Man-Month
True or False?
For brevity is very good,
Where we are, or are not understood.
SAMUEL BUTLER. Hudibras
Brooks asserting a proposition, 1967
Photo by J. Alex Langley for Fortune Magazine
229
230 Propositions of The Mythical Man-Month: True or False?
Much more is known today about software engineering than
was known in 1975. Which of the assertions in the original 1975
edition have been supported by data and experience? Which
have been disproved? Which have been obsoleted by the chang-
ing world? To help you judge, here in outline form is the essence
of the 1975 book—assertions I believed to be true: facts and
ruleof-thumb-type generalizations from experience—extracted
without change of meaning. (You might ask, "If this is all the
original book said, why did it take 177 pages to say it?") Com-
ments in brackets are new.
Most of these propositions are operationally testable. My
hope in putting them forth in stark form is to focus readers'
thoughts, measurements, and comments.
Chapter 1. The Tar Pit
1.1 A programming systems product takes about nine times
as much effort as the component programs written sepa-
rately for private use. I estimate that productizing im-
poses a factor of three; and that designing, integrating,
and testing components into a coherent system imposes a
factor of three; and that these cost components are essen-
tially independent of each other.
1.2 The craft of programming "gratifies creative longings
built deep within us and delights sensibilities we have in
common with all men," providing five kinds of joys:
• The joy of making things
The joy of making things that are useful to other
people
9
The fascination of fashioning puzzle-like objects of in-
terlocking moving parts
• The joy of always learning, of a nonrepeating task
« The delight of working in a medium so tractable—
pure thought-stuff—which nevertheless exists, moves,
and works in a way that word-objects do not.
Chapter 2. The Mythical Man-Month 231
1.3 Likewise the craft has special woes inherent in it.
• Adjusting to the requirement of perfection is the
hardest part of learning to program.
" Others set one's objectives and one must depend
upon things (especially programs) one cannot control;
the authority is not equal to the responsibility.
« This sounds worse than it is: actual authority comes
from momentum of accomplishment.
• With any creativity come dreary hours of painstaking
labor; programming is no exception.
* The programming project converges more slowly the
nearer one gets to the end, whereas one expects it to
converge faster as one approaches the end.
* One's product is always threatened with obsolescence
before completion. The real tiger is never a match for
the paper one, unless real use is wanted.
Chapter 2. The Mythical Man-Month
2.1 More programming projects have gone awry for lack of
calendar time than for all other causes combined.
2.2 Good cooking takes time; some tasks cannot be hurried
without spoiling the result.
2.3 All programmers are optimists: "All will go well."
2.4 Because the programmer builds with pure thought-stuff,
we expect few difficulties in implementation.
2.5 But our ideas themselves are faulty, so we have bugs.
2.6 Our estimating techniques, built around cost-accounting,
confuse effort and progress. The man-month is a fallacious
and dangerous myth, for it implies that men and months are in-
terchangeable.
2.7 Partitioning a task among multiple people occasions extra
communication effort—training and intercommunication.
2.8 My rule of thumb is 1/3 of the schedule for design, 1/6
for coding, 1/4 for component testing, and 1/4 for system
testing.
232 Propositions of The Mythical Man-Month: True or False?
2.9 As a discipline, we lack estimating data.
2.10 Because we are uncertain about our scheduling estimates,
we often lack the courage to defend them stubbornly
against management and customer pressure.
2.11 Brooks's Law: Adding manpower to a late software proj-
ect makes it later.
2.12 Adding people to a software project increases the total ef-
fort necessary in three ways: the work and disruption of
repartitioning itself, training the new people, and added
intercommunication.
Chapter 3. The Surgical Team
3.1 Very good professional programmers are ten times as pro-
ductive as poor ones, at same training and two-year ex-
perience level. (Sackman, Grant, and Erickson)
3.2 Sackman, Grant, and Erickson's data showed no correla-
tion whatsoever between experience and performance. I
doubt the universality of that result.
3.3 A small sharp team is best—as few minds as possible.
3.4 A team of two, with one leader, is often the best use of
minds. [Note God's plan for marriage.]
3.5 A small sharp team is too slow for really big systems.
3.6 Most experiences with really large systems show the
brute-force approach to scaling up to be costly, slow, in-
efficient, and to produce systems that are not conceptu-
ally integrated.
3.7 A chief-programmer, surgical-team organization offers a
way to get the product integrity of few minds and the to-
tal productivity of many helpers, with radically reduced
communication.
Chapter 4. Aristocracy, Democracy, and System Design
4.1 "Conceptual integrity is the most important consideration
in system design."
I Chapter 5. The Second-System Effect 233
4.2 "The ratio of function to conceptual complexity is the ul-
I timate test of system design," not just the richness of
function. [This ratio is a measure of ease of use, valid over
both simple and difficult uses.]
4.3 To achieve conceptual integrity, a design must proceed
from one mind or a small group of agreeing minds.
4.4 "Separation of architectural effort from implementation is
a very powerful way of getting conceptual integration on
very large projects." [Small ones, too.]
4.5 "If a system is to have conceptual integrity, someone
must control the concepts. That is an aristocracy that
needs no apology."
4.6 Discipline is good for art. The external provision of an ar-
chitecture enhances, not cramps, the creative style of an
implementing group.
4.7 A conceptually integrated system is faster to build and to
test.
4.8 Much of software architecture, implementation, and re-
alization can proceed in parallel. [Hardware and software
design can likewise proceed in parallel.]
Chapter 5. The Second-System Effect
5.1 Early and continuous communication can give the archi-
tect good cost readings and the builder confidence in the
design, without blurring the clear division of responsibil-
ities.
5.2 How an architect can successfully influence implementa-
tion:
« Remember that the builder has the creative responsi-
bility for implementation; the architect only suggests.
• Always be ready to suggest a way of implementing
anything one specifies; be prepared to accept any
other equally good way.
• Deal quietly and privately in such suggestions.
9
Be ready to forgo credit for suggested improvements.
234 Propositions of The Mythical Man-Month: True or False?
« Listen to the builder's suggestions for architecture im-
provements.
5.3 The second is the most dangerous system a person ever
designs; the general tendency is to over-design it.
5.4 OS/360 is a good example of the second-system effect.
[Windows NT seems to be a 1990s example.]
5.5 Assigning a priori values in bytes and microseconds to
functions is a worthwhile discipline.
Chapter 6. Passing the Word
6.1 Even when a design team is large, the results must be re-
duced to writing by one or two, in order that the mini-
decisions be consistent.
6.2 It is important to explicitly define the parts of an architec-
ture that are not prescribed as carefully as those that are.
6.3 One needs both a formal definition of a design, for pre-
cision, and a prose definition for comprehensibility.
6.4 One of the formal and prose definitions must be standard,
and the other derivative. Either definition can serve in
either role.
6.5 An implementation, including a simulation, can serve as
an architectural definition; such use has formidable dis-
advantages.
6.6 Direct incorporation is a very clean technique for enforc-
ing an architectural standard in software. [In hardware,
too—consider the Mac WIMP interface built into ROM.]
6.7 An architectural "definition will be cleaner and the [ar-
chitectural] discipline tighter if at least two implementa-
tions are built initially."
6.8 It is important to allow telephone interpretations by an ar-
chitect in response to implementers' queries; it is imper-
ative to log these and publish them. [Electronic mail is
now the medium of choice.]
6.9 "The project manager's best friend is his daily adversary,
the independent product-testing organization."
Chapter 7. Why Did the Tower of Babel Fail? 235
Chapter 7. Why Did the Tower of Babel Fail?
7.1 The Tower of Babel project failed because of lack of com-
munication and of its consequent, organization.
Communication
7.2 "Schedule disaster, functional misfit, and system bugs all
arise because the left hand doesn't know what the right
hand is doing." Teams drift apart in assumptions.
7.3 Teams should communicate with one another in as many
ways as possible: informally, by regular project meetings
with technical briefings, and via a shared formal project
workbook. [And by electronic mail.]
Project Workbook
7.4 A project workbook is "not so much a separate document
as it is a structure imposed on the documents that the
project will be producing anyway."
7.5 "All the documents of the project need to be part of this
[workbook] structure."
7.6 The workbook structure needs to be designed carefully
and early.
7.7 Properly structuring the on-going documentation from
the beginning "molds later writing into segments that fit
into that structure" and will improve the product man-
uals.
7.8 "Each team member should see all the [workbook] mate-
rial." [I would now say, each team member should be able
to see all of it. That is, World-Wide Web pages would suf-
fice.]
7.9 Timely updating is of critical importance.
7.10 The user needs to have attention especially drawn to
changes since his last reading, with remarks on their sig-
nificance.
7.11 The OS/360 Project workbook started with paper and
switched to microfiche.
7.12 Today [even in 1975], the shared electronic notebook is a
236 Propositions of The Mythical Man-Month: True or False?
much better, cheaper, and simpler mechanism for achiev-
ing all these goals.
7.13 One still has to mark the text with [the functional equiv-
alent of] change bars and revision dates. One still needs
a LIFO electronic change summary.
7.14 Parnas argues strongly that the goal of everyone seeing
everything is totally wrong; parts should be encapsulated
so that no one needs to or is allowed to see the internals
of any parts other than his own, but should see only the
interfaces.
7.15 Parnas's proposal is a recipe for disaster. [/ have been quite
convinced otherwise by Parnas, and totally changed my mind.]
Organization
7.16 The purpose of organization is to reduce the amount of
communication and coordination necessary.
7.17 Organization embodies division of labor and specialization of
function in order to obviate communication.
7.18 The conventional tree organization reflects the authority
structure principle that no person can serve two masters.
7.19 The communication structure in an organization is a net-
work, not a tree, so all kinds of special organization mech-
anisms ("dotted lines") have to be devised to overcome
the communication deficiencies of the tree-structured or-
ganization.
7.20 Every subproject has two leadership roles to be filled, that
of the producer and that of the technical director, or architect.
The functions of the two roles are quite distinct and re-
quire different talents.
7.21 Any of three relationships among the two roles can be
quite effective:
• The producer and director can be the same.
• The producer may be boss, and the director the pro-
ducer's right-hand person.
• The director may be boss, and the producer the direc-
tor's right-hand person.
Chapter 8. Calling the Shot 237
Chapter 8. Calling the Shot
8.1 One cannot accurately estimate the total effort or schedule
of a programming project by simply estimating the coding
time and multiplying by factors for the other parts of the
task.
8.2 Data for building isolated small systems are not applicable
to programming systems projects.
8.3 Programming increases goes as a power of program size.
8.4 Some published studies show the exponent to be about
1.5. [Boehm's data do not at all agree with this, but vary from
1.05 to I.2.]1
8.5 Portman's ICL data show full-time programmers applying
only about 50 percent of their time to programming and
debugging, versus other overhead-type tasks.
8.6 Aron's IBM data show productivity varying from 1.5 K
lines of code (KLOC) per man-year to 10 KLOC/man-year
as a function of the number of interactions among system
parts.
8.7 Harr's Bell Labs data show productivities on operating-
systems-type work to run about 0.6 KLOC/man-year and
on compiler-type work about 2.2 KLOC/man-year for fin-
ished products.
8.8 Brooks's OS/360 data agrees with Harr's: 0.6-0.8 KLOC/
man-year on operating systems and 2-3 KLOC/man-year
on compilers.
8.9 Corbatd's MIT Project MULTICS data show productivity
of 1.2 KLOC/man-year on a mix of operating systems and
compilers, but these are PL/I lines of code, whereas all the
other data are assembler lines of code!
8.10 Productivity seems constant in terms of elementary state-
ments.
8.11 Programming productivity may be increased as much as
five times when a suitable high-level language is used.
238 Propositions of The Mythical Man-Month: True or False?
Chapter 9. Ten Pounds in a Five-Pound Sack
9.1 Aside from running time, the memory space occupied by a
program is a principal cost. This is especially true for op-
erating systems, where much is resident all the time.
9.2 Even so, money spent on memory for program residence
may yield very good functional value per dollar,^ better
than other ways of investing in configuration. Program
size is not bad; unnecessary size is.
9.3. The software builder must set size targets, control size,
and devise size-reduction techniques, just as the hard-
ware builder does for components.
9.4 Size budgets must be explicit not only about resident size
but also about the disk accesses occasioned by program
fetches.
9.5 Size budgets have to be tied to function assignments; de-
fine exactly what a module must do when you specify
how big it must be.
9.6 On large teams, subteams tend to suboptimize to meet
their own targets rather than think about the total effect
on the user. This breakdown in orientation is a major haz-
ard of large projects.
9.7 All during implementation, the system architects must
maintain constant vigilance to ensure continued system
integrity.
9.8 Fostering a total-system, user-oriented attitude may well
be the most important function of the programming man-
ager.
9.9 An early policy decision is to decide how fine-grained the
user choice of options will be, since packaging them in
clumps saves memory space [and often marketing costs].
9.10 The size of the transient area, hence of the amount of pro-
gram per disk fetch, is a crucial decision, since perfor-
mance is a super-linear function of that size. [This whole
decision has been obsoleted, first by virtual memory, then
Chapter 10. The Documentary Hypothesis 239
by cheap real memory. Users now typically buy enough
real memory to hold all the code of major applications.]
9.11 To make good space-time tradeoffs, a team needs to be
trained in the programming techniques peculiar to a par-
ticular language or machine, especially a new one.
9.12 Programming has a technology, and every project needs
a library of standard components.
9.13 Program libraries should have two versions of each com-
ponent, the quick and the squeezed. [This seems obsolete
today.]
9.14 Lean, spare, fast programs are almost always the result of
strategic breakthrough, rather than tactical cleverness.
9.15 Often such a breakthrough will be a new algorithm.
9.16 More often, the breakthrough will come from redoing the
representation of the data or tables. Representation is the es-
sence of programming.
Chapter 10. The Documentary Hypothesis
10.1 "The hypothesis: Amid a wash of paper, a small number
of documents become the critical pivots around which
every project's management revolves. These are the man-
ager's chief personal tools."
10.2 For a computer development project, the critical docu-
ments are the objectives, manual, schedule, budget, or-
ganization chart, floorspace allocation, and the estimate,
forecast, and prices of the machine itself.
10.3 For a university department, the critical documents are
similar: the objectives, degree requirements, course de-
scriptions, research proposals, class schedule and teach-
ing plan, budget, floorspace allocation, and assignments
of staff and graduate assistants.
10.4 For a software project, the needs are the same: the objec-
tives, user manual, internals documentation, schedule,
budget, organization chart, and floorspace allocation.
10.5 Even on a small project, therefore, the manager should
240 Propositions of The Mythical Man-Month: True or False?
from the beginning formalize such a set of documents.
10.6 Preparing each document of this small set focuses
thought and crystallizes discussion. The act of writing re-
quires hundreds of mini-decisions, and it is the existence
of these that distinguish clear, exact policies from fuzzy
ones.
10.7 Maintaining each critical document provides a status sur-
veillance and warning mechanism.
10.8 Each document itself serves as a checklist and a database.
10.9 The project manager's fundamental job is to keep every-
body going in the same direction.
10.10 The project manager's chief daily task is communication,
not decision-making; the documents communicate the
plans and decisions to the whole team.
10.11 Only a small part of a technical project manager's time—
perhaps 20 percent—is spent on tasks where he needs in-
formation from outside his head.
10.12 For this reason, the touted market concept of a "manage-
ment total-information system" to support executives is
not based on a valid model of executive behavior.
Chapter 11. Plan to Throw One Away
11.1 Chemical engineers have learned not to take a process
from the lab bench to the factory in one step, but to build
a pilot plant to give experience in scaling quantities up and
operating in nonprotective environments.
11.2 This intermediate step is equally necessary for program-
ming products, but software engineers do not yet rou-
tinely field-test a pilot system before undertaking to
deliver the real product. [This has now become common
practice, with a beta version. This is not the same as a pro-
totype with limited function, an alpha version, which I
would also advocate.]
11.3 For most projects, the first system built is barely usable:
too slow, too big, too hard to use, or all three.
Chapter 11, Plan to Throw One Away 241
11.4 The discard and redesign may be done in one lump, or
piece-by-pieee, but it will be done.
11.5 Delivering the first system, the throwaway, to users will
buy time, but only at the cost of agony for the user, dis-
traction for the builders supporting it while they do the
redesign, and a bad reputation for the product that will
be hard to live down.
11.6 Hence, plan to throw one away; you will, anyhow.
11.7 "The programmer delivers satisfaction of a user need
rather than any tangible product." (Cosgrove)
11.8 Both the actual need and the user's perception of that
need will change as programs are built, tested, and used.
11.9 The tractability and the invisibility of the software prod-
uct expose its builders (exceptionally) to perpetual changes
in requirements.
11.10 Some valid changes in objectives (and in development
strategies) are inevitable, and it is better to be prepared for
them than to assume that they will not come.
11.11 The techniques for planning a software product for change,
especially structured programming with careful module
interface documentation, are well known but not uni-
formly practiced. It also helps to use table-driven tech-
niques wherever possible. [Modern memory costs and
sizes make such techniques better and better.]
11.12 Use high-level language, compile-time operations, incor-
porations of declarations by reference, and self-docu-
menting techniques to reduce errors induced by change.
11.13 Quantify changes into well-defined numbered versions.
[Now standard practice.]
Plan the Organization for Change
11.14 Programmer reluctance to document designs comes not
so much from laziness as from the hesitancy to undertake
defense of decisions that the designer knows are tenta-
tive. (Cosgrove)
242 - Propositions of The Mythical Man-Month: True or False?
11.15 Structuring an organization for change is much harder
than designing a system for change.
11.16 The project boss must work at keeping the managers and
the technical people as interchangeable as their talents al-
low; in particular, one wants to be able to move people
easily between technical and managerial roles.
11.17 The barriers to effective dual-ladder organization are so-
ciological, and they must be fought with constant vigi-
lance and energy.
11.18 It is easy to establish corresponding salary scales for the
corresponding rungs on a dual ladder, but it requires
strong proactive measures to give them corresponding
prestige: equal offices, equal support services, over-com-
pensating management actions.
11.19 Organizing as a surgical team is a radical attack on all as-
pects of this problem. It is really the long-run answer to
the problem of flexible organization.
Two Steps Forward and One Step Back—
Program Maintenance
11.20 Program maintenance is fundamentally different from
hardware maintenance; it consists chiefly of changes that
repair design defects, add incremental function, or adapt
to changes in the use environment or configuration.
11.21 The total lifetime cost of maintaining a widely used pro-
gram is typically 40 percent or more of the cost of devel-
oping it.
11.22 Maintenance cost is strongly affected by the number of
users. More users find more bugs.
11.23 Campbell points out an interesting drop-and-climb curve
in bugs per month over a product's life.
11.24 Fixing a defect has a substantial (20 to 50 percent) chance
of introducing another.
11.25 After each fix, one must run the entire bank of test cases
previously run against a system to ensure that it has not
Chapter 12. Sharp Tools 243
been damaged in an obscure way.
11.26 Methods of designing programs so as to eliminate or at
least illuminate side effects can have an immense payoff
in maintenance costs.
11.27 So can methods of implementing designs with fewer peo-
ple, fewer interfaces, and fewer bugs.
One Step Forward and One Step Back—System Entropy
Rises over Lifetime
11.28 Lehman and Belady find that the total number of modules
increases linearly with the release number of a large op-
erating system (OS/360), but that the number of modules
affected increases exponentially with the release number.
11.29 All repairs tend to destroy structure, to increase the en-
tropy and disorder of a system. Even the most skillful pro-
gram maintenance only delays the program's subsidence
into unfixable chaos, from which there has to be a
ground-up redesign. [Many of the real needs for upgrad-
ing a program, such as performance, especially attack its
internal structural boundaries. Often the original bound-
aries occasioned the deficiencies that surface later.]
Chapter 12. Sharp Tools
12.1 The manager of a project needs to establish a philosophy
and set aside resources for the building of common tools,
and at the same time to recognize the need for personal-
ized tools.
12.2 Teams building operating systems need a target machine
of their own on which to debug; it needs maximum
memory rather than maximum speed, and a system pro-
grammer to keep the standard software current and
serviceable.
12.3 The debugging machine, or its software, also needs to be
instrumented, so that counts and measurements of all
244 Propositions of The Mythical Man-Month: True or False?
kinds of program parameters can be automatically made.
12.4 The requirement for target machine use has a peculiar
growth curve: low activity followed by explosive growth,
then leveling off.
12.5 System debugging, like astronomy, has always been done
chiefly at night.
12.6 Allocating substantial blocks of target machine time to
one subteam at a time proved the best way to schedule,
much better than interleaving subteam use, despite theory.
12.7 This preferred method of scheduling scarce computers by
blocks has survived 20 years [in 1975] of technology
change because it is most productive. [It still is, in 1995].
12.8 If a target computer is new, one needs a logical simulator
for it. One gets it sooner, and it provides a dependable de-
bugging vehicle even after one has a real machine.
12.9 A master program library should be divided into (1) a set
of individual playpens, (2) a system integration subli-
brary, currently under system test, and (3) a released ver-
sion. Formal separation and progression gives control.
12.10 The tool that saves the most labor in a programming proj-
ect is probably a text-editing system.
12.11 Voluminosity in system documentation does indeed intro-
duce a new kind of incomprehensibility [see Unix, for
example], but it is far preferable to the severe underdoc-
umentation that is so common.
12.12 Build a performance simulator, outside in, top down.
Start it very early. Listen when it speaks.
High-Level Language
12.13 Only sloth and inertia prevent the universal adoption of
high-level language and interactive programming. [And
today they have been adopted universally]
12.14 High-level language improves not only productivity but
also debugging; fewer bugs and easier to findr
12.15 The classical objections of function, object-code space,
Chapter 13. The Whole and the Parts 245
and object-code speed have been made obsolete by the
advance of language and compiler technology.
12.16 The only reasonable candidate for system programming
today is PL/I. [No longer true.]
Interactive Programming
12.17 Interactive systems will never displace batch systems for
some applications. [Still true.]
12.18 Debugging is the hard and slow part of system program-
ming, and slow turnaround is the bane of debugging.
12.19 Limited evidence shows that interactive programming at
least doubles productivity in system programming.
Chapter 13. The Whole and the Parts
13.1 The detailed, painstaking architectural effort implied in
Chapters 4, 5, and 6 not only makes a product easier to
use, it makes it easier to build and reduces the number of
system bugs that have to be found.
13.2 Vyssotsky says "Many, many failures concern exactly
those aspects that were never quite specified."
13.3 Long before any code itself, the specification must be
handed to an outside testing group to be scrutinized for
completeness and clarity. The developers themselves can-
not do this. (Vyssotsky)
13.4 "Wirth's top-down design [by stepwise refinement] is the
most important new programming formalization of the
[1965-1975] decade."
13.5 Wirth advocates using as high-level a notation as possible
on each step.
13.6 A good top-down design avoids bugs in four ways.
13.7 Sometimes one has to go back, scrap a high level, and
start over.
13.8 Structured programming, designing programs whose
control structures consist only of a specified set that gov-
ern blocks of code (versus miscellaneous branching), is a
246 Propositions of The Mythical Man-Month: True or False?
sound way to avoid bugs and is the right way to think.
13.9 Gold's experimental results show three times as much
progress is made in the first interaction of an interactive
debugging session as on subsequent interactions. It still
pays to plan debugging carefully before signing on. [I
think it still does, in 1995.]
13.10 I find that proper use of a good [quick response interac-
tive debugging] system requires two hours at the desk for
each two-hour session on the machine: one hour in
sweeping up and documenting after the session and one
in planning changes and tests for the next time.
13.11 System debugging (in contrast to component debugging)
will take longer than one expects.
13.12 The difficulty of system debugging justifies a thoroughly
systematic and planned approach.
13.13 One should begin system debugging only after the pieces
seem to work (versus-bolt-it-together-and-try in order to
smoke out the interface bugs; and versus starting system
debugging when the component bugs are fully known
but not fixed.) [This is especially true for teams.]
13.14 It is worthwhile to build lots of debugging scaffolding and
test code, perhaps even 50 percent as much as the product
being debugged.
13.15 One must control and document changes and versions,
with team members working in playpen copies.
13.16 Add one component at a time during system debugging.
13.17 Lehman and Belady offer evidence the change quanta
should be large and infrequent or else very small and
frequent. The latter is more subject to instability. [A Mi-
crosoft team makes small frequent quanta work. The
growing system is rebuilt every night.]
Chapter 14. Hatching a Catastrophe
14.1 "How does a project get to be a year late? . . . One day at
a time."
Chapter 14. Hatching a Catastrophe 247
14.2 Day-by-day schedule slippage is harder to recognize,
harder to prevent, and harder to make up than calamities.
14.3 The first step in controlling a big project on a tight sched-
ule is to have a schedule, made up of milestones and dates
for them.
14.4 Milestones must be concrete, specific, measurable events
defined with knife-edge sharpness.
14.5 A programmer will rarely lie about milestone progress, if
the milestone is so sharp he can't deceive himself.
14.6 Studies of estimating behavior by government contractors
on large projects show that activity time estimates revised
carefully every two weeks do not significantly change as
the start time approaches, that during the activity overes-
timates come steadily down; and that underestimates do
not change until about three weeks before scheduled
completion.
14.7 Chronic schedule slippage is a morale-killer. [Jim Mc-
Carthy of Microsoft says, "If you miss one deadline, make
sure you make the next one."2]
14.8 Hustle is essential for great programming teams, just as
for great baseball teams.
14.9 There is no substitute for a critical-path schedule to enable
one to tell which slips matter how much.
14.10 The preparation of a critical-path chart is the most valu-
able part of its use, since laying out the network, identi-
fying the dependencies, and estimating the segments
force a great deal of very specific planning very early in a
project.
14.11 The first chart is always terrible, and one invents and in-
vents in making the next one.
14.12 A critical path chart answers the demoralizing excuse,
"The other piece is late, anyhow."
14.13 Every boss needs both exception information that re-
quires action and a status picture for education and dis-
tant early warning.
14.14 Getting the status is hard, since subordinate managers
248 Propositions of The Mythical Man-Month: True or False?
have every reason not to share it.
14.15 By bad action, a boss can guarantee to squelch full status
disclosure; conversely, carefully separating status reports
and accepting them without panic or preemption will en-
courage honest reporting.
14.16 One must have review techniques by which true status
becomes known to all players. For this purpose a mile-
stone schedule and completion document is the key.
14.17 Vyssotsky: "I have found it handy to carry both 'sched-
uled' (boss's dates) and 'estimated' (lowest-level manag-
er's dates) dates in the milestone report. The project
manager has to keep his fingers off the estimated dates."
14.18 A small Plans and Controls team that maintains the mile-
stone report is invaluable for a large project.
Chapter 15. The Other Face
15.1 For the program product, the other face to the user, the
documentation, is fully as important as the face to the ma-
chine.
15.2 Even for the most private of programs, prose documen-
tation is necessary, for memory will fail the user-author.
15.3 Teachers and managers have by and large failed to instill
in programmers an attitude about documentation that
will inspire for a lifetime, overcoming sloth and schedule
pressure.
15.4 This failure is not due so much to lack of zeal or eloquence
as to a failure to show how to document effectively and
economically.
15.5 Most documentation fails in giving too little overview.
Stand way back and zoom in slowly.
15.6 The critical user documentation should be drafted before
the program is built, for it embodies basic planning deci-
sions. It should describe nine things (see the chapter).
15.7 A program should be shipped with a few test cases, some
Chapter 15. The Other Face 249
for valid input data, some for borderline input data, and
some for clearly invalid input data.
15.8 Documentation of program internals, for the person who
must modify it, also demands a prose overview, which
should contain five kinds of things (see the chapter).
15.9 The flow chart is a most thoroughly oversold piece of pro-
gram documentation; the detailed blow-by-blow flow
chart is a nuisance, obsoleted by written high-level lan-
guages. (A flow chart is a diagrammed high-level lan-
guage.)
15.10 Few programs need more than a one-page flow chart, if
that. [MILSPEC documentation requirements are really
wrong on this point.]
15.11 One does indeed need a program structure graph, which
does not need the ANSI flow-charting standards.
15.12 To keep documentation maintained, it is crucial that it be
incorporated in the source program, rather than kept as a
separate document.
15.13 Three notions are key to minimizing the documentation
burden:
• Use parts of the program that have to be there any-
way, such as names and declarations, to carry as
much of the documentation as possible.
• Use space and format to show subordination and
nesting and to improve readability.
• Insert the necessary prose documentation into the
program as paragraphs of comment, especially as
module headers.
15.14 In documentation for use by program modifiers, tell why
things are like they are, rather than merely how they are.
Purpose is the key to understanding; even high-level lan-
guage syntax does not at all convey purpose.
15.15 Self-documenting programming techniques find their
greatest use and power in high-level languages used with
on-line systems, which are the tools one should be using.
250 Propositions of The Mythical Man-Month: True or False?
Original Epilogue
E.I Software systems are perhaps the most intricate and com-
plex (in terms of number of distinct kinds of parts) of the
things humanity makes.
E.2 The tar pit of software engineering will continue to be
sticky for a long time to come.
19
The Mythical Man-Month
after 20 Years
19
The Mythical Man-Month
after 20 Years
I know no way of judging the future but by the past.
PATRICK HENRY
You can never plan the future by the past.
EDMUND BURKE
Shooting the Rapids
The Bettman Archive
253
254 The Mythical Man-Month after 20 Years
Why Is There a Twentieth Anniversary Edition?
The plane droned through the night toward LaGuardia. Clouds
and darkness veiled all interesting sights. The document I was
studying was pedestrian. I was not, however, bored. The stranger
sitting next to me was reading The Mythical Man-Month, and I
was waiting to see if by word or sign he would react. Finally as
we taxied toward the gate, I could wait no longer:
"How is that book? Do you recommend it?"
"Hmph! Nothing in it I didn't know already."
I decided not to introduce myself.
Why has The Mythical Man-Month persisted? Why is it still
seen to be relevant to software practice today? Why does it have
a readership outside the software engineering community, gen-
erating reviews, citations, and correspondence from lawyers,
doctors, psychologists, sociologists, as well as from software
people? How can a book written 20 years ago about a software-
building experience 30 years ago still be relevant, much less use-
ful?
One explanation sometimes heard is that the software de-
velopment discipline has not advanced normally or properly.
This view is often supported by contrasting computer software
development productivity with computer hardware manufac-
turing productivity, which has multiplied at least a thousand-
fold over the two decades. As Chapter 16 explains, the anomaly
is not that software has been so slow in its progress but rather
that computer technology has exploded in a fashion unmatched
in human history. By and large this comes from the gradual
transition of computer manufacturing from an assembly indus-
try to a process industry, from labor-intensive to capital-inten-
sive manufacturing. Hardware and software development, in
contrast to manufacturing, remain inherently labor-intensive.
A second explanation often advanced is that The Mythical
Man-Month is only incidentally about software but primarily
about how people in teams make things. There is surely some
truth in this; the preface to the 1975 edition says that managing
The Central Argument: Conceptual Integrity 255
a software project is more like other management than most
programmers initially believe. I still believe that to be true. Hu-
man history is a drama in which the stories stay the same, the
scripts of those stories change slowly with evolving cultures,
and the stage settings change all the time. So it is that we see
our twentieth-century selves mirrored in Shakespeare, Homer,
and the Bible. So to the extent The MM-M is about people and
teams, obsolescence should be slow.
Whatever the reason, readers continue to buy the book, and
they continue to send me much-appreciated comments. Nowa-
days I am often asked, "What do you now think was wrong
when written? What is now obsolete? What is really new in the
software engineering world?" These quite distinct questions are
all fair, and I shall address them as best I can. Not in that order,
however, but in clusters of topics. First, let us consider what was
right when written, and still is.
The Central Argument: Conceptual Integrity and
the Architect
Conceptual integrity. A clean, elegant programming product
must present to each of its users a coherent mental model of the
application, of strategies for doing the application, and of the
user-interface tactics to be used in specifying actions and param-
eters. The conceptual integrity of the product, as perceived by
the user, is the most important factor in ease of use. (There are
other factors, of course. The Macintosh's uniformity of user in-
terface across all applications is an important example. More-
over, it is possible to construct coherent interfaces that are
nevertheless quite awkward. Consider MS-DOS.)
There are many examples of elegant software products de-
signed by a single mind, or by a pair. Most purely intellectual
works such as books or musical compositions are so produced.
Product-development processes in many industries cannot,
however, afford this straightforward approach to conceptual in-
tegrity. Competitive pressures force urgency; in many modern
256 The Mythical Man-Month after 20 Years
technologies the end product is quite complex, and the design
inherently requires many man-months of effort. Software prod-
ucts are both complex and fiercely competitive in schedule.
Any product that is sufficiently big or urgent to require the
effort of many minds thus encounters a peculiar difficulty: the
result must be conceptually coherent to the single mind of the
user and at the same time designed by many minds. How does
one organize design efforts so as to achieve such conceptual in-
tegrity? This is the central question addressed by The MM-M.
One of its theses is that managing large programming projects
is qualitatively different from managing small ones, just because
of the number of minds involved. Deliberate, and even heroic,
management actions are necessary to achieve coherence.
The architect. I argue in Chapters 4 through 7 that the most
important action is the commissioning of some one mind to be
the product's architect, who is responsible for the conceptual in-
tegrity of all aspects of the product perceivable by the user. The
architect forms and owns the public mental model of the prod-
uct that will be used to explain its use to the user. This includes
the detailed specification of all of its function and the means for
invoking and controlling it. The architect is also the user's agent,
knowledgeably representing the user's interest in the inevitable
tradeoffs among function, performance, size, cost, and sched-
ule. This role is a full-time job, and only on the smallest teams
can it be combined with that of the team manager. The architect
is like the director and the manager like the producer of a mo-
tion picture.
Separation of architecture from implementation and realiza-
tion. To make the architect's crucial task even conceivable, it is
necessary to separate the architecture, the definition of the
product as perceivable by the user, from its implementation. Ar-
chitecture versus implementation defines a clean boundary be-
tween parts of the design task, and there is plenty of work on
each side of it.
The Second-System Effect 257
Recursion of architects. For quite large products, one mind
cannot do all of the architecture, even after all implementation
concerns have been split off. So it is necessary for the system
master architect to partition the system into subsystems. The
subsystem boundaries must be at those places where interfaces
between the subsystems are minimal and easiest to define rig-
orously. Then each piece will have its own architect, who must
report to the system master architect with respect to the archi-
tecture. Clearly this process can proceed recursively as required.
Today I am more convinced than ever. Conceptual integrity is
central to product quality. Having a system architect is the most
important single step toward conceptual integrity. These prin-
ciples are by no means limited to software systems, but to the
design of any complex construct, whether a computer, an air-
plane, a Strategic Defense Initiative, a Global Positioning Sys-
tem. After teaching a software engineering laboratory more than
20 times, I came to insist that student teams as small as four peo-
ple choose a manager and a separate architect. Defining distinct
roles in such small teams may be a little extreme, but I have ob-
served it to work well and to contribute to design success even
for small teams.
The Second-System Effect: Featuritis and
Frequency-Guessing
Designing for large user sets. One of the consequences of the
personal computer revolution is that increasingly, at least in the
business data processing community, off-the-shelf packages
are replacing custom applications. Moreover, standard software
packages sell hundreds of thousands of copies, or even millions.
System architects for machine-vendor-supplied software have
always had to design for a large, amorphous user set rather than
for a single, definable application in one company. Many, many
system architects now face this task.
Paradoxically, it is much more difficult to design a general-
258 The Mythical Man-Month after 20 Years
purpose tool than it is to design a special-purpose tool, precisely
because one has to assign weights to the differing needs of the
diverse users.
Featuritis. The besetting temptation for the architect of a gen-
eral purpose tool such as a spreadsheet or a word processor is
to overload the product with features of marginal utility, at the
expense of performance and even of ease of use. The appeal of
proposed features is evident at the outset; the performance pen-
alty is evident only as system testing proceeds. The loss of ease
of use sneaks up insidiously, as features are added in little in-
crements, and the manuals wax fatter and fatter.1
For mass-market products that survive and evolve through
many generations, the temptation is especially strong. Millions
of customers request hundreds of features; any request is itself
evidence that "the market demands it." Frequently, the original
system architect has gone on to greater glories, and the archi-
tecture is in the hands of people with less experience at repre-
senting the user's overall interest in balance. A recent review of
Microsoft Word 6.0 says "Word 6.0 packs in features; update
slowed by baggage. . . . Word 6.0 is also big and slow." It notes
with dismay that Word 6.0 requires 4 MB of RAM, and goes on
to say that the rich added function means that "even a Macin-
tosh Ilfx [is] just barely up to the Word 6 task".2
Defining the user set. The larger and more amorphous the
user set, the more necessary it is to define it explicitly if one is
to achieve conceptual integrity. Each member of the design team
will surely have an implicit mental image of the users, and each
designer's image will be different. Since an architect's image of
the user consciously or subconsciously affects every architec-
tural decision, it is essential for a design team to arrive at a sin-
gle shared image. And that requires writing down the attributes
of the expected user set, including:
* Who they are
* What they need
The Second-System Effect 259
* What they think they need
* What they want
Frequencies. For any software product, any of the attributes of
the user set is in fact a distribution, with many possible values,
each with its own frequency. How is the architect to arrive at
these frequencies? Surveying this ill-defined population is a du-
bious and costly proposition.3 Over the years I have become
convinced that an architect should guess, or, if you prefer, pos-
tulate, a complete set of attributes and values with their fre-
quencies, in order to develop a complete, explicit, and shared
description of the user set.
Many benefits flow from this unlikely procedure. First, the
process of carefully guessing the frequencies will cause the ar-
chitect to think very carefully about the expected user set. Sec-
ond, writing the frequencies down will subject them to debate,
which will illuminate all the participants and bring to the sur-
face the differences in the user images that the several designers
carry. Third, enumerating the frequencies explicitly helps every-
one recognize which decisions depend upon which user set
properties. Even this sort of informal sensitivity analysis is valu-
able. When it develops that very important decisions are hing-
ing on some particular guess, then it is worth the cost to
establish better estimates for that value. (The gIBIS system de-
veloped by Jeff Conklin provides a tool for formally and accu-
rately tracking design decisions and documenting the reasons
for each.4 I have not had opportunity to use it, but I think it
would be very helpful.)
To summarize: write down explicit guesses for the attributes
of the user set. It is far better to be explicit and wrong man to be
vague.
What about the "Second-System Effect"? A perceptive stu-
dent remarked that The Mythical Man-Month recommended a
recipe for disaster: Plan to deliver the second version of any new
system (Chapter 11), which Chapter 5 characterizes as the most
260 The Mythical Man-Month after 20 Years
dangerous system one ever designs. I had to grant him a
"gotcha."
The contradiction is more linguistic than real. The "second"
system described in Chapter 5 is the second system fielded, the
follow-on system that invites added function and frills. The
"second" system in Chapter 11 is the second try at building
what should be the first system to be fielded. It is built under all
the schedule, talent, and ignorance constraints that characterize
new projects—the constraints that exert a slimness discipline.
The Triumph of the WIMP Interface
One of the most impressive developments in software during
the past two decades has been the triumph of the Windows,
Icons, Menus, Pointing interface—or WIMP for short. It is today
so familiar as to need no description. This concept was first
publicly displayed by Doug Englebart and his team from the
Stanford Research Institute at the Western Joint Computer
Conference of 1968.5 From there the ideas went to Xerox Palo
Alto Research Center, where they emerged in the Alto personal
workstation developed by Bob Taylor and team. They were
picked up by Steve Jobs for the Apple Lisa, a computer too slow
to carry its exciting ease-of-use concepts. These concepts Jobs
then embodied in the commercially successful Apple Macintosh
in 1985. They were later adopted in Microsoft Windows for the
IBM PC and compatibles. The Mac version will be my example.6
Conceptual integrity via a metaphor. The WIMP is a superb
example of a user interface that has conceptual integrity,
achieved by the adoption of a familiar mental model, the desk-
top metaphor, and its careful consistent extension to exploit a
computer graphics implementation. For example, the costly but
proper decision to overlay windows instead of tiling them fol-
lows directly from the metaphor. The ability to size and shape
windows is a consistent extension that gives the user the new
powers enabled by the computer graphics medium. Papers on a
The Triumph of the WIMP Interface 261
desktop cannot be so readily sized and shaped. Dragging and
dropping follow directly from the metaphor; selecting icons by
pointing with a cursor is a direct analog of picking things with
the hand. Icons and nested folders are faithful analogs of desk-
top documents; so is the trash can. The concepts of cutting,
copying, and pasting faithfully mirror the things we used to do
with documents on desktops. So faithfully is the metaphor fol-
lowed and so consistent is its extension that new users are pos-
itively jarred by the notion of dragging a diskette's icon to the
trash to eject the disk. Were the interface not almost uniformly
consistent, that (pretty bad) inconsistency would not grate so
much.
Where is the WIMP interface forced to go far beyond the
desktop metaphor? Most notably in two respects: menus and
one-handedness. When working with a real desktop, one does
actions to documents, rather than telling someone or something
to do them. And when one does tell someone to do an action,
one usually generates, rather than selects, the oral or written
imperative verb commands: "Please file this." "Please find the
earlier correspondence." "Please send this to Mary to handle."
Alas, the reliable interpretation of free-form generated En-
glish commands is beyond the present state of the art, whether
commands are written or spoken. So the interface designers
were two steps removed from direct user action on documents.
They wisely picked up from the usual desktop its one example
of command selection—the printed buck slip, on which the user
selects from among a constrained menu of commands whose se-
mantics are standardized. This idea they extended to a horizon-
tal menu of vertical pull-down submenus.
Command utterances and the two-cursor problem. Com-
mands are imperative sentences; they always have a verb and
usually have a direct object.. For any action, one needs to specify
a verb and a noun. The pointing metaphor says, to specify two
things at a time, have two distinguished cursors on the screen,
each driven by a separate mouse—one in the right hand and one
262 The Mythical Man-Month after 20 Years
in the left. After all, on a physical desktop we normally work
with both hands. (But, one hand is often holding things fixed in
place, which happens by default on the computer desktop.) The
mind is certainly capable of two-handed operation; we regularly
use two hands in typing, driving, cooking. Alas, providing one
mouse was already a big step forward for personal computer
makers; no commercial system accommodates two simultane-
ous mouse-cursor actions, one driven with each hand.7
The interface designers accepted reality and designed for
one mouse, adopting the syntactic convention that one points
out (selects) the noun first. One points at the verb, a menu item.
This really gives away a lot of ease-of-use. As I watch users, or
videotapes of users, or computer tracings of cursor movements,
I am immediately struck that one cursor is having to do the work
of two: pick an object in the desktop part of the window; pick a
verb in the menu portion; find or re-find an object in the desk-
top; again pull down a menu (often the same one) and pick a
verb. Back and forth, back and forth the cursor goes, from data-
space to menu-space, each time discarding the useful informa-
tion as to where it was last time it was in this space—altogether,
an inefficient process.
A brilliant solution. Even if the electronics and software could
readily handle two simultaneously active cursors, there are
space-layout difficulties. The desktop in the WIMP metaphor
really includes a typewriter, and one must accommodate its real
keyboard in physical space on the real desktop. A keyboard plus
two mouse-pads uses a lot of the arm's-reach real estate. Well,
the problem of the keyboard can be turned into an opportu-
nity—why not enable efficient two-handed operation by using
one hand on the keyboard to specify verbs and the other hand
on a mouse to pick nouns. Now the cursor stays in the data
space, exploiting the high locality of successive noun picks. Real
efficiency, real user power.
The Triumph of the WIMP Interface 263
User power versus ease of use. That solution, however, gives
away the thing that makes menus so easy to use for novices—
menus present the alternative verbs valid at any particular state.
We can buy a package, bring it home, and start using it without
consulting the manual, merely by knowing what we bought it
for, and experimenting with the different menu verbs.
One of the hardest issues facing software architects is ex-
actly how to balance user power versus ease of use. Do they de-
sign for simple operation for the novice or the occasional user,
or for efficient power for the professional user? The ideal answer
is to provide both, in a conceptually coherent way—that is
achieved in the WIMP interface. The high-frequency menu
verbs each have single-key + command-key equivalents, mostly
chosen so that they can easily be struck as a single chord with
the left hand. On the Mac, for example, the command key (||)
is just below the Z and X keys; therefore the highest-frequency
operations are encoded as
Incremental transition from novice to power user. This dual
system for specifying command verbs not only meets the low-
learning-effort needs of the novice and the efficiency needs of
the power user, it provides for each user to make a smooth tran-
sition between modes. The letter encodings, called short cuts, are
shown on the menus beside the verbs, so that a user in doubt
can pull down the menu to check the letter equivalent, instead
of just picking the menu item. Each novice learns first the short
cuts for his own high-frequency operations. Any short cut he is
doubtful about he can try, since &z will undo any single mis-
take. Alternatively, he can check the menu to see what com-
mands are valid. Novices will pull lots of menus; power users
very few; and in-between users will only occasionally need to
pick from a menu, since each will know the few short-cuts that
make up most of his own operations. Most of us software de-
signers are too familiar with this interface to appreciate fully its
elegance and power.
264 The Mythical Man-Month after 20 Years
The success of direct incorporation as a device for enforcing ar-
chitecture. The Mac interface is remarkable in yet another way.
Without coercion, its designers have made it a standard inter-
face across applications, including the vast majority that are
written by third parties. So the user gains conceptual coherence
at the interface level not only within the software furnished with
the machine but across all applications.
This feat the Mac designers accomplished by building the
interface into the read-only memory, so that it is easier and
faster for developers to use it than to build their own idiosyn-
cratic interfaces. These natural incentives for uniformity pre-
vailed widely enough to establish a de facto standard. The
natural incentives were helped by a total management commit-
ment and a lot of persuasion by Apple. The independent re-
viewers in the product magazines, recognizing the immense
value of cross-application conceptual integrity, have also sup-
plemented the natural incentives by mercilessly criticizing prod-
ucts that do not conform.
This is a superb example of the technique, recommended in
Chapter 6, of achieving uniformity by encouraging others to di-
rectly incorporate one's code into their products, rather than at-
tempting to have them build their own software to one's
specifications.
The fate of WIMP: Obsolescence. Despite its excellencies, I ex-
pect the WIMP interface to be a historical relic in a generation.
Pointing will still be the way to express nouns as we command
our machines; speech is surely the right way to express the
verbs. Tools such as Voice Navigator for the Mac and Dragon for
the PC already provide this capability.
Don't Build One to Throw Away—The Waterfall Model
Is Wrong!
The unforgettable picture of Galloping Gertie, the Tacoma Nar-
rows Bridge, opens Chapter 11, which radically recommends:
Don't Build One to Throw Away 265
"Plan to throw one away; you will, anyhow." This I now per-
ceive to be wrong, not because it is too radical, but because it is
too simplistic.
The biggest mistake in the "Build one to throw away" con-
cept is that it implicitly assumes the classical sequential or wa-
terfall model of software construction. The model derives from
a Gantt chart layout of a staged process, and it is often drawn
as in Figure 19.1. Win ton Royce improved the sequential model
in a classic 1970 paper by providing for
* Some feedback from a stage to its predecessors
* Limiting the feedback to the immediately preceding stage
only, so as to contain the cost and schedule delay it occa-
sions.
He preceded The MM-M in advising builders to "build it twice."8
Chapter 11 is not the only one tainted by the sequential waterfall
model; it runs through the book, beginning with the scheduling
rule in Chapter 2. That rule-of-thumb allocates 1/3 of the sched-
ule to planning, 1/6 to coding, 1/4 to component test, and 1/4 to
system test.
Fig. 19.1 Waterfall model of software construction
266 The Mythical Man-Month after 20 Years
The basic fallacy of the waterfall model is that it assumes a
project goes through the process once, that the architecture is ex-
cellent and easy to use, the implementation design is sound,
and the realization is fixable as testing proceeds. Another way
of saying it is that the waterfall model assumes the mistakes will
all be in the realization, and thus that their repair can be
smoothly interspersed with component and system testing.
"Plan to throw one away" does indeed attack this fallacy
head on. It is not the diagnosis that is wrong; it is the remedy.
Now I did suggest that one might discard and redesign the first
system piece by piece, rather than in one lump. This is fine so
far as it goes, but it fails to get at the root of the problem. The
waterfall model puts system test, and therefore by implication
user testing, at the end of the construction process. Thus one can
find impossible awkwardnesses for users, or unacceptable per-
formance, or dangerous susceptibility to user error or malice,
only after investing in full construction. To be sure, the Alpha
test scrutiny of the specifications aims to find such flaws early,
but there is no substitute for hands-on users.
The second fallacy of the waterfall model is that it assumes
one builds a whole system at once, combining the pieces for an
end-to-end system test after all of the implementation design,
most of the coding, and much of the component testing has
been done.
The waterfall model, which was the way most people
thought about software projects in 1975, unfortunately got en-
shrined into DOD-STD-2167, the Department of Defense speci-
fication for all military software. This ensured its survival well
past the time when most thoughtful practitioners had recog-
nized its inadequacy and abandoned it. Fortunately, the DoD
has since begun to see the light.9
There has to be upstream movement. Like the energetic
salmon in the chapter-opening picture, experience and ideas
from each downstream part of the construction process must
leap upstream, sometimes more than one stage, and affect the
An Incremental-Build Model Is Better 267
upstream activity.
Designing the implementation will show that some archi-
tectural features cripple performance; so the architecture has to
be reworked. Coding the realization will show some functions
to balloon space requirements; so there may have to be changes
to architecture and implementation.
One may well, therefore, iterate through two or more ar-
chitecture-implementation design cycles before realizing any-
thing as code.
An Incremental-Build Model Is Better—
Progressive Refinement
Building an end-to-end skeleton system
Harlan Mills, working in a real-time system environment, early
advocated that we should build the basic polling loop of a real-
time system, with subroutine calls (stubs) for all the functions
(Fig. 19.2), but only null subroutines. Compile it; test it. It goes
round and round, doing literally nothing, but doing it cor-
rectly.10
Subroutines
Fig. 19.2
268 The Mythical Man-Month after 20 Years
Next, we flesh out a (perhaps primitive) input module and
an output module. Voil3! A running system that does some-
thing, however dull. Now, function by function, we incremen-
tally build and add modules. At every stage we have a running
system. If we are diligent, we have at every stage a debugged,
tested system. (As the system grows, so does the burden of
regression-testing each new module against all the previous test
cases.)
After every function works at a primitive level, we refine or
rewrite first one module and then another, incrementally grow-
ing the system. Sometimes, to be sure, we have to change the
original driving loop, and or even its module interfaces.
Since we have a working system at all times
• we can begin user testing very early, and
• we can adopt a build-to-budget strategy that protects abso-
lutely against schedule or budget overruns (at the cost of
possible functional shortfall).
For some 22 years, I taught the software engineering laboratory
at the University of North Carolina, sometimes jointly with Da-
vid Parnas. In this course, teams of usually four students built
in one semester some real software application system. About
halfway through those years, I switched to teaching incremental
development. I was stunned by the electrifying effect on team
morale of that first picture on the screen, that first running sys-
tem.
Parnas Families
David Parnas has been a major thought leader in software en-
gineering during this whole 20-year period. Everyone is familiar
with his information-hiding concept. Rather less familiar, but
very important, is Parnas's concept of designing a software
product as a family of related products.11 He urges the designer
to anticipate both lateral extensions and succeeding versions of
Parnas Families 269
Fig. 19.3
a product, and to define their function or platform differences
so as to construct a family tree of related products (Fig 19.3).
The trick in the design of such a tree is to put near its root
those design decisions that are less likely to change.
Such a design strategy maximizes the re-use of modules.
More important, the same strategy can be broadened to include
not only deliverable products but also the successive interme-
diate versions created in an incremental-build strategy. The
product then grows through its intermediate stages with mini-
mum backtracking.
270 The Mythical Man-Month after 20 Years
Microsoft's "Build Every Night" Approach
James McCarthy described to me a product process used by his
team and others at Microsoft. It is incremental growth carried to
a logical conclusion. He says,
After we first ship, we will be shipping later versions that add more
function to an existing, running product. Why should the initial
building process be different? Beginning at the time of our first
milestone [where the march to first ship has three intermediate
milestones] we rebuild the developing system every night [and run
the test cases]. The build cycle becomes the heartbeat of the project.
Every day one or more of the programmer-tester teams check in
modules with new functions. After every build, we have a running
system. If the build breaks, we stop the whole process until the trou-
ble is found and fixed. At all times everybody on the team knows
the status.
It is really hard. You have to devote lots of resources, but it is
a disciplined process, a tracked and known process. It gives the
team credibility to itself. Your credibility determines your morale,
your emotional state.
Software builders in other organizations are surprised, even
shocked, by this process. One says, "I've made it a practice to
build every week, but I think it would be too much work to
build every night." And that may be true. Bell Northern Re-
search, for example, rebuilds its 12-million-line system every
week.
Incremental-Build and Rapid Prototyping
Since an incremental development process enables early testing
with real users, what is the difference between that and rapid
prototyping? It seems to me that the two are related but sepa-
rate. One can have either without the other.
Harel usefully defines a prototype as
[A version of a program that] reflects only the design decisions
made in the process of preparing the conceptual model, and not de-
cisions driven by implementation concerns.12
Parnas Was Right, and I Was Wrong 271
It is possible to build a prototype that is not at all part of a prod-
uct growing toward shipment. For example, one may build an
interface prototype that has no real program function behind it,
merely the finite-state machine that makes it appear to go
through its paces. One can even prototype and test interfaces
by the Wizard-of-Oz technique, with a concealed human simu-
lating the system's responses. Such prototyping can be very use-
ful for getting early user feedback, but it is quite apart from
testing a product growing toward shipment.
Similarly, implementers may well undertake to build a ver-
tical slice of a product, in which a very limited function set is
constructed in full, so as to let early sunlight into places where
performance snakes may lurk. What is the difference between
the first-milestone-build of the Microsoft process and a rapid
prototype? Function. The first-milestone product may not have
enough function to be of interest to anyone; the shippable prod-
uct is defined as such by its completeness in furnishing a useful
set of functions, and by its quality, the belief that it works ro-
bustly.
Parnas Was Right, and I Was Wrong about
Information Hiding
In Chapter 7 I contrast two approaches to the question of how
much each team member should be allowed or encouraged to
know about each other's designs and code. In the Operating
System/360 project, we decided that all programmers should see
all the material—i.e., each programmer having a copy of the
project workbook, which came to number over 10,000 pages.
Harlan Mills has argued persuasively that "programming
should be a public process," that exposing all the work to every-
body's gaze helps quality control, both by peer pressure to do
things well and by peers actually spotting flaws and bugs.
This view contrasts sharply with David Parnas's teaching
that modules of code should be encapsulated with well-defined
interfaces, and that the interior of such a module should be the
272 The Mythical Man-Month after 20 Years
private property of its programmer, not discernible from out-
side. Programmers are most effective if shielded from, not ex-
posed to, the innards of modules not their own.13
I dismissed Parnas's concept as a "recipe for disaster" in
Chapter 7. Parnas was right, and I was wrong. I am now con-
vinced that information hiding, today often embodied in object-
oriented programming, is the only way of raising the level of
software design.
One can indeed get disasters with either technique. Mills'
technique ensures that programmers can know the detailed se-
mantics of the interfaces they work to by knowing what is on
the other side. Hiding those semantics leads to system bugs. On
the other hand, Parnas's technique is robust under change and
is more appropriate in a design-for-change philosophy.
Chapter 16 argues the following:
• Most past progress in software productivity has come from
eliminating noninherent difficulties such as awkward ma-
chine languages and slow batch turnaround.
• There are not a lot more of these easy pickings.
• Radical progress is going to have to come from attacking the
essential difficulties of fashioning complex conceptual con-
structs.
The most obvious way to do this recognizes that programs
are made up of conceptual chunks much larger than the individ-
ual high-level language statement—subroutines, or modules, or
classes. If we can limit design and building so that we only do
the putting together and parameterization of such chunks from
prebuilt collections, we have radically raised the conceptual
level, and eliminated the vast amounts of work and the copious
opportunities for error that dwell at the individual statement
level.
Parnas's information-hiding definition of modules is the
first published step in that crucially important research pro-
gram, and it is an intellectual ancestor of object-oriented pro-
How Mythical Is the Man-Month? 273
gramming. He defined a module as a software entity with its
own data model and its own set of operations. Its data can only
be accessed via one of its proper operations. The second step
was a contribution of several thinkers: the upgrading of the Par-
nas module into an abstract data type, from which many objects
could be derived. The abstract data type provides a uniform way
of thinking about and specifying module interfaces, and an ac-
cess discipline that is easy to enforce.
The third step, object-oriented programming, introduces
the powerful concept of inheritance, whereby classes (data types)
take as defaults specified attributes from their ancestors in the
class hierarchy.14 Most of what we hope to gain from object-
oriented programming derives in fact from the first step, mod-
ule encapsulation, plus the idea of prebuilt libraries of modules
or classes that are designed and tested for reuse. Many people have
chosen to ignore the fact that such modules are not just pro-
grams, but instead are program products in the sense discussed
in Chapter 1. Some people are vainly hoping for significant
module reuse without paying the initial cost of building prod-
uct-quality modules—generalized, robust, tested, and docu-
mented. Object-oriented programming and reuse are discussed
in Chapters 16 and 17.
How Mythical Is the Man-Month? Boehm's Model
and Data
Over the years, there have been many quantitative studies of
software productivity and the factors affecting it, especially the
trade-offs between project staffing and schedule.
The most substantial study is one done by Barry Boehm of
some 63 software projects, mostly aerospace, with about 25 at
TRW. His Software Engineering Economics contains not only the
results but a useful set of cost models of progressive compre-
hensiveness. Whereas the coefficients in the models are surely
different for ordinary commercial software and for aerospace
software built to government standards, nevertheless his mod-
274 The Mythical Man-Month after 20 Years
els are backed by an immense amount of data. I think the book
will be a useful classic a generation from now.
His results solidly confirm The MM-M's assertion that the
trade-off between men and months is far from linear, that the
man-month is indeed mythical as a measure of productivity. In
particular, he finds:15
• There is a cost-optimum schedule time to first shipment,
1 = 2.5 (MM) 1/3. That is, the optimum time in months goes
as the cube root of the expected effort in man-months, a fig-
ure derived from the size estimate and other factors in his
model. An optimum staffing curve is a corollary.
e
The cost curve rises slowly as the planned schedule gets
longer than the optimum. People with more time take more
time.
9
The cost curve rises sharply as the planned schedule gets
shorter than the optimum.
• Hardly any projects succeed in less than 3/4 of the calculated opti-
mum schedule, regardless of the number of people applied! This
quotable result gives the software manager solid ammuni-
tion when higher management is demanding impossible
schedule commitments.
How true is Brooks's Law? There have even been careful stud-
ies evaluating the truth of Brooks's (intentionally simplistic)
Law, that adding manpower to a late software project makes it
later. The best treatment is that of Abdel-Hamid and Madnick,
in their ambitious and valuable 1991 book, Software Project Dy-
namics: An Integrated Approach.16 The book develops a quantita-
tive model of project dynamics. Their chapter on Brooks's Law
provides more detailed insight into what happens under vari-
ous assumptions as to what manpower is added, and when. To
investigate this, the authors extend their own careful model of
a middle-sized applications project by assuming that new peo-
ple have a learning curve and accounting for the extra commu-
How Mythical Is the Man-Month? 275
nication and training work. They conclude that "Adding more
people to a late project always makes it more costly, but it does
not always cause it to be completed later [italics theirs]." In par-
ticular, adding extra manpower early in the schedule is a much
safer maneuver than adding it later, since the new people al-
ways have an immediate negative effect, which takes weeks to
compensate.
Stutzke develops a simpler model in order to perform a sim-
ilar investigation, with a similar result.17 He develops a detailed
analysis of the process and costs of assimilating the new work-
ers, including explicitly the diversion of their mentors from the
project task itself. He tests his model against an actual project in
which manpower was successfully doubled and the original
schedule achieved, after a mid-project slip. He treats alterna-
tives to adding more programmers, especially overtime. Most
valuable are his many items of practical advice as to how new
workers should be added, trained, supported with tools, etc.,
so as to minimize the disruptive effects of adding them. Espe-
cially noteworthy is his comment that new people added late in
a development project must be team players willing to pitch in
and work within the process, and not attempt to alter or im-
prove the process itself!
Stutzke believes that the added burden of communication in
a larger project is a second-order effect and does not model it. It
is not clear whether and how Abdel-Hamid and Madnick take it
into account. Neither model takes into account the fact that the
work must be repartitioned, a process I have often found to be
non trivial.
The "outrageously simplified" statement of Brooks's Law is
made more useful by these careful treatments of the proper
qualifications. On balance, I stand by the bald statement as the
best zeroth-order approximation to the truth, a rule of thumb to
warn managers against blindly making the instinctive fix to a
late project.
276 The Mythical Man-Month after 20 Years
People Are Everything (Well, Almost Everything)
Some readers have found it curious that The MM-M devotes
most of the essays to the managerial aspects of software engi-
neering, rather than the many technical issues. This bias was
due in part to the nature of my role on the IBM Operating Sys-
tem/360 (now MVS/370). More fundamentally, it sprang from a
conviction that the quality of the people on a project, and their
organization and management, are much more important fac-
tors in success than are the tools they use or the technical ap-
proaches they take.
Subsequent researches have supported that conviction.
Boehm's COCOMO model finds that the quality of the team is
by far the largest factor in its success, indeed four times more
potent than the next largest factor. Most academic research on
software engineering has concentrated on tools. I admire and
covet sharp tools. Nevertheless, it is encouraging to see ongoing
research efforts on the care, growing, and feeding of people,
and on the dynamics of software management.
Peopleware. A major contribution during recent years has
been DeMarco and Lister's 1987 book, Peopleware: Productive
Projects and Teams. Its underlying thesis is that "The major prob-
lems of our work are not so much technological as sociological in
nature." It abounds with gems such as, "The manager's func-
tion is not to make people work, it is to make it possible for peo-
ple to work." It deals with such mundane topics as space,
furniture, team meals together. DeMarco and Lister provide real
data from their Coding War Games that show stunning corre-
lation between performances of programmers from the same or-
ganization, and between workplace characteristics and both
productivity and defect levels.
The top performers' space is quieter, more private, better protected
against interruption, and there is more of it. . . . Does it really
matter to you . . . whether quiet, space, and privacy help your cur-
rent people to do better work or [alternatively] help you to attract
and keep better people?19
The Power of Giving Up Power 277
I heartily commend the book to all my readers.
Moving projects. DeMarco and Lister give considerable atten-
tion to team fusion, an intangible but vital property. I think it is
management's overlooking fusion that accounts for the readi-
ness I have observed in multilocation companies to move a proj-
ect from one laboratory to another.
My experience and observation are limited to perhaps a
half-dozen moves. I have never seen a successful one. One can
move missions successfully. But in every case of attempts to
move projects, the new team in fact started over, in spite of hav-
ing good documentation, some well-advanced designs, and
some of the people from the sending team. I think it is the break-
ing of fusion of the old team that aborts the embryonic product,
and brings about restart.
The Power of Giving Up Power
If one believes, as I have argued at many places in this book,
that creativity comes from individuals and not from structures
or processes, then a central question facing the software man-
ager is how to design structure and process so as to enhance,
rather than inhibit, creativity and initiative. Fortunately, this
problem is not peculiar to software organizations, and great
thinkers have worked on it. E. F. Schumacher, in his classic,
Small is Beautiful: Economics as if People Mattered, proposes a the-
ory of organizing enterprises to maximize the creativity and
joy of the workers. For his first principle he chooses the "Prin-
ciple of Subsidiary Function" from the Encyclical Quadragesima
Anno of Pope Pius XI:
It is an injustice and at the same time a grave evil and disturbance
of right order to assign to a greater and higher association what
lesser and subordinate organizations can do. For every social activ-
ity ought of its very nature to furnish help to the members of the
body social and never destroy and absorb them. . . . Those in com-
mand should be sure that the more perfectly a graduated order is
278 The Mythical Man-Month after 20 Years
preserved among the various associations, in observing the princi-
ple of subsidiary function, the stronger will be the social authority
and effectiveness and the happier and more prosperous the condition
of the State.19
Schumacher goes on to interpret:
The Principle of Subsidiary Function teaches us that the centre will
gain in authority and effectiveness if the freedom and responsibility
of the lower formations are carefully preserved, with the result that
the organization as a whole will be "happier and more prosperous."
How can such a structure be achieved?. . . . The large orga-
nization will consist of many semi-autonomous units, which we
may call quasi-firms. Each of them will have a large amount of
freedom, to give the greatest possible chance to creativity and en-
trepreneurship. . . . Each quasi-firm must have both a profit and
loss account, and^a balance sheet.20
Among the most exciting developments in software engi-
neering are the early stages of putting such organizational ideas
into practice. First, the microcomputer revolution created a new
software industry of hundreds of start-up firms, all of them
starting small, and marked by enthusiasm, freedom, and crea-
tivity. The industry is changing now, as many small companies
are being acquired by larger ones. It remains to be seen if the
larger acquirers will understand the importance of preserving
the creativity of smallness.
More remarkably, high management in some large firms
have undertaken to delegate power down to individual software
project teams, making them approach Schumacher's quasi-firms
in structure and responsibility. They are surprised and de-
lighted at the results.
Jim McCarthy of Microsoft described to me his experience at
emancipating his teams:
Each feature team (30-40 people) owns its feature set, its schedule,
and even its process of how to define, build, ship. The team is made
What's the Biggest New Surprise? Millions of Computers 279
up for four or five specialties, including building, testing, and
writing. The team settles squabbles; the bosses don't. I can't em-
phasize enough the importance of empowerment, of the team being
accountable to itself for its success.
Earl Wheeler, retired head of IBM's software business, told
me his experience in undertaking the downward delegation of
power long centralized in IBM's division managements:
The key thrust [of recent years] was delegating power down. It was
like magic! Improved quality, productivity, morale. We have small
teams, with no central control. The teams own the process, but they
have to have one. They have many different processes. They own
the schedule, but they feel the pressure of the market. This pressure
causes them to reach for tools on their own.
Conversations with individual team members, of course,
show both an appreciation of the power and freedom that is del-
egated, and a somewhat more conservative estimate of how
much control really is relinquished. Nevertheless, the delega-
tion achieved is clearly a step in the right direction. It yields
exactly the benefits Pius XI predicted: the center gains in real
authority by delegating power, and the organization as a whole
is happier and more prosperous.
What's the Biggest New Surprise? Millions of Computers
Every software guru I have talked with admits to being caught
by surprise by the microcomputer revolution and its outgrowth,
the shrink-wrapped software industry. This is beyond doubt the
crucial change of the two decades since The MM-M, It has many
implications for software engineering.
The microcomputer revolution has changed how everybody
uses computers. Schumacher stated the challenge more than
20 years ago:
What is it that we really require from the scientists and technolo-
280 The Mythical Man-Month after 20 Years
gists? I should answer: We need methods and equipment which are
9
cheap enough so that they are accessible to virtually everyone;
9
suitable for small-scale application; and
• compatible with man's need for creativity.21
These are exactly the wonderful properties that the micro-
computer revolution has brought to the computer industry and
to its users, now the general public. The average American can
now afford not only a computer of his own, but a suite of soft-
ware that 20 years ago would have cost a king's salary. Each of
Schumacher's goals is worth contemplating; the degree to which
each has been achieved is worth savoring, especially the last. In
area after area, new means of self-expression are accessible to
ordinary people as well as to professionals.
Partly the enhancement comes in other fields as it has in
software creation—in the removal of accidental difficulties. Writ-
ten manuscripts used to be accidentally rigidified by the time
and cost of retyping to incorporate changes. On a 300-page
work, one might go through retyping every three to six months,
but in between, one just kept marking the manuscript. One
could not easily assess what the changes had done to the flow
of the logic and the rhythm of the words. Now, manuscripts
have become wondrously fluid.22
The computer has brought a similar fluidity to many other
media: artistic drawings, building plans, mechanical drawings,
musical compositions, photographs, video sequences, slide pre-
sentations, multimedia works, and even to spreadsheets. In
each case, the manual method of production required recopying
the bulky unchanged parts in order to see changes in context.
Now we enjoy for each medium the same benefits that time-
sharing brought to software creation—the ability to revise and
to assess instantly the effect without losing one's train of
thought.
Creativity is also enhanced by new and flexible auxiliary
tools. For prose production, as one example, we are now served
by spelling checkers, grammar checkers, style advisors, biblio-
What's the Biggest New Surprise? Millions of Computers 281
graphic systems, and the remarkable ability to see pages con-
currently formatted into final layout. We do not yet appreciate
what instantaneous encyclopedias or the infinite resources of
the World-Wide Web will mean for a writer's impromptu re-
search.
Most important, the new fluidity of the media makes easy
the exploration of many radically different alternatives when a
creative work is just taking form. Here is another case where an
order of magnitude in a quantitative parameter, here change-
time, makes a qualitative difference in how one goes about a
task.
Tools for drawing enable building designers to explore
many more options per hour of creative investment. The con-
nection of computers to synthesizers, with software for auto-
matically generating or playing scores, makes it much easier to
capture keyboard doodles. Digital photograph manipulation, as
with Adobe Photoshop, enables minutes-long experiments that
would take hours in a darkroom. Spreadsheets enable the easy
exploration of dozens of "what if" alternative scenarios.
Finally, wholly new creative media have been enabled by
the ubiquity of the personal computer. Hypertexts, proposed by
Vannevar Bush in 1945, are practical only with computers.23
Multimedia presentations and experiences were big deals—just
too much trouble—before the personal computer and the rich,
cheap software available for it. Virtual-environment systems,
not yet cheap or ubiquitous, will become so, and will be yet an-
other creative medium.
The microcomputer revolution has changed how everybody
builds software. The software processes of the 1970s have
themselves been altered by the microprocessor revolution and
the technology advances that enabled it. Many of the accidental
difficulties of those software building processes have been elim-
inated. Fast individual computers are now the routine tools of
the software developer, so that turnaround time is an almost ob-
solete concept. The personal computer of today is not only faster
282 The Mythical Man-Month after 20 Years
than the supercomputer of I960, it is faster than the Unix work-
station of 1985. All of this means that compilation is fast even
on the humblest machines, and large memories have eliminated
waits for disk-based linking. Large memories also make it rea-
sonable to keep symbol tables in memory with object code, so
high-level debugging without recompilation is routine.
In the last 20 years, we have come almost completely
through the use of time-sharing as the methodology for con-
structing software. In 1975, time-sharing had just replaced batch
computing as the most common technique. The network was
used to give the software builder access both to shared files and
to a shared powerful engine for compilation, linking, and test-
ing. Today, the personal workstation provides the computing
engine, and the network primarily gives shared access to the
files that are the team's developing work-product. Client-server
systems make shared access to check-in, build, and the appli-
cation of test cases a different and simpler process.
Similar advances in user interfaces have occurred. The
WIMP interface provides much more convenient editing of pro-
gram texts as well as of English-language texts. The 24-line,
72-column screen has been replaced by full-page or even two-
page screens, so programmers can see much more context for
changes they are making.
Whole New Software Industry—Shrink-Wrapped Software
Alongside the classical software industry there has exploded an-
other. Product unit sales run in hundreds of thousands, even
millions. Entire rich software packages can be had for less than
the cost of one supported programmer-day. The two industries
are different in many ways, and they coexist.
The classical software industry. In 1975, the software industry
had several identifiable and somewhat different components, all
of which still exist today:
« Computer vendors, who provide operating systems, com-
Shrink-Wrapped Software 283
pilers, and utilities for their products.
• Application users, such as the MIS shops of utilities, banks,
insurance companies, and government agencies, who build
application packages for their own use.
« Custom application builders, who contract to build proprie-
tary packages for users. Many of these contractors specialize
in defense applications, where requirements, standards,
and marketing procedures are peculiar.
• Commercial package developers, who at that time devel-
oped mostly large applications for specialized markets, such
as statistical analysis packages and CAD systems.
Tom DeMarco notes the fragmentation of the classical soft-
ware industry, especially the application-user component:
What I didn't expect: the field has partitioned into niches. How you
do something is much more a function of the niche than it is the use
of general systems analysis methods, general languages, and gen-
eral testing techniques. Ada was the last of the general-purpose lan-
guages, and it has become a niche language.
In the routine commercial application niche, fourth-generation
languages have made powerful contributions. Boehm says,
"Most successful 4GLs are the result of someone's codifying a
piece of an application domain in terms of options and param-
eters." The most pervasive of these 4GLs are application gen-
erators and combined database-communications packages with
inquiry languages.
Operating system worlds have coalesced. In 1975, operating
systems abounded: each hardware vendor had at least one pro-
prietary operating system per product line; many had two. How
different things are today! Open systems are the watchword,
and there are only five significant operating systems environ-
284 The Mythical Man-Month after 20 Years
ments into which people market applications packages (in
chronological order):
• The IBM MVS and VM environments
« The DEC VMS environment
« The Unix environment, in one flavor or another
• The IBM PC environment, whether DOS, OS-2, or Windows
The Apple Macintosh environment.
The shrink-wrapped industry. For the developer in the
shrink-wrapped industry, the economics are entirely different
from those of the classical industry: development cost is divided
by large quantities; packaging and marketing costs loom large.
In the classical in-house application development industry,
schedule and the details of function were negotiable, develop-
ment cost might not be; in the fiercely competitive open market,
schedule and function quite dominate development cost.
As one would expect, the starkly different economics have
given rise to rather different programming cultures. The classi-
cal industry tended to be dominated by large firms with estab-
lished management styles and work cultures. The shrink-
wrapped industry, on the other hand, began as hundreds of
start-ups, freewheeling and fiercely focused on getting the job
done rather than on process. In this climate, there has always
been a much greater recognition of the talent of the individual
programmer, an implicit awareness that great designs come
from great designers. The start-up culture has the capability of
rewarding star performers in proportion to their contributions;
in the classical software industry the sociology of corporations
and their salary-management plans have always made this dif-
ficult. It is not surprising that many of the stars of the new gen-
eration have gravitated to the shrink-wrapped industry.
Buy and Build—Shrink-Wrapped Packages
As Components
Radically better software robustness and productivity are to be
Buy and Build—Shrink-Wrapped Packages As Components 285
had only by moving up a level, and making programs by the
composition of modules, or objects. An especially promising
trend is the use of mass-market packages as the platforms on
which richer and more customized products are built. A truck-
tracking system is built on a shrink-wrapped database and com-
munications package; so is a student information system. The
want ads in computer magazines offer hundreds of HyperCard
stacks and customized templates for Excel, dozens of special
functions in Pascal for MiniCad or functions in AutoLisp for
AutoCad.
Metaprogramming. Building HyperCard stacks, Excel tem-
plates, or MiniCad functions is sometimes called metaprogram-
ming, the construction of a new layer that customizes function
for a subset of a package's users. The metaprogramming concept
is not new, only resurgent and renamed. In the early 1960s,
computer vendors and many big management information sys-
tems (MIS) shops had small groups of specialists who crafted
whole application programming languages out of macros in as-
sembly language. Eastman Kodak's MIS shop had a house ap-
plication language defined on the IBM 7080 macroassembler.
Similarly with IBM's OS/360 Queued Telecommunications Ac-
cess Method, one could read many pages of an ostensibly
assembly-language telecommunications program before en-
countering a machine-level instruction. Now the chunks offered
by the metaprogrammer are many times larger than those mac-
ros. This development of secondary markets is most encourag-
ing—while we have been waiting to see an effective market in
C + + classes develop, a market in reusable metaprograms has
grown up unremarked.
This really does attack essence. Because the build-on-package
phenomenon does not today affect the average MIS program-
mer, it is not yet very visible to the software engineering disci-
pline. Nevertheless, it will grow rapidly, because it does attack
the essence of fashioning conceptual constructs. The shrink-
286 The Mythical Man-Month after 20 Years
wrapped package provides a big module of function, with an
elaborate but proper interface, and its internal conceptual struc-
ture does not have to be designed at all. High-function software
products such as Excel or 4th Dimension are big modules in-
deed, but they serve as known, documented, tested modules
with which to build customized systems. Next-level application
builders get richness of function, a shorter development time, a
tested component, better documentation, and radically lower
cost.
The difficulty, of course, is that the shrink-wrapped soft-
ware package is designed as a stand-alone entity whose
functions and interfaces metaprogrammers cannot change.
Moreover, and more seriously, shrink-wrapped package build-
ers seemingly have little incentive to make their products suit-
able as modules in a larger system. I think that perception is
wrong, that there is an untapped market in providing packages
designed to facilitate metaprogrammer use.
So what is needed? We can identify four levels of users of
shrink-wrapped packages:
4
The as-is user, who operates the application in straightfor-
ward manner, content with the functions and the interface
the designers provided.
• The metaprogrammer, who builds templates or functions on
top of a single application, using the interface provided,
principally to save work for the end user.
e
The external function writer, who hand-codes added func-
tions into an application. These are essentially new appli-
cation language primitives that call out to separate code
modules written in a general-purpose language. One needs
the capability to interface these new functions to the appli-
cation as intercepted commands, as callbacks, or as over-
loaded functions.
e
The metaprogrammer who uses one, or especially several,
applications as components in a larger system. This is the
user whose needs are poorly met today. This is also the use
The State and Future of Software Engineering 287
which promises substantial effectiveness gains in building
new applications.
For this last user, a shrink-wrapped application needs an ad-
ditional documented interface, the metaprogramming interface
(MPI). It needs several capabilities. First, the metaprogram
needs to be in control of an ensemble of applications, whereas
normally each application assumes it is itself in control. The en-
semble must control the user interface, which ordinarily the ap-
plication assumes it is doing. The ensemble must be able to
invoke any application function as if its command string had
come from the user. It should receive output from the applica-
tion as if it is the screen, except that it needs the output parsed
into logical units of suitable datatypes, rather than the text string
that would have been displayed. Some applications, such as
FoxPro, have wormholes that allow one to pass a command
string in, but the information one gets back is skimpy and un-
parsed. The wormhole is an ad hoc fix for a need that demands
a general, designed solution.
It is powerful to have a scripting language for controlling the
interactions among the ensemble of applications. Unix first
provided this kind of function, with its pipes and its standard
ASCII-string file format. Today AppleScript is a rather good ex-
ample.
The State and Future of Software Engineering
I once asked Jim Ferrell, chairman of the Department of Chem-
ical Engineering at North Carolina State University, to relate the
history of chemical engineering, as distinguished from chemis-
try. He thereupon gave a wonderful impromptu hour-long ac-
count, beginning with the existence from antiquity of many
different production processes for many products, from steel to
bread to perfume. He told how Professor Arthur D. Little
founded a Department of Industrial Chemistry at MIT in 1918,
to find, develop, and teach a common base of technique shared
288 The Mythical Man-Month after 20 Years
by all the processes. First came rules of thumb, then empirical
nomograms, then formulas for designing particular compo-
nents, then mathematical models for heat transport, mass trans-
port, momentum transport in single vessels.
As Ferrell's tale unfolded, I was struck by the many parallels
between the development of chemical engineering and that of
software engineering, almost exactly fifty years later. Parnas re-
proves me for writing about software engineering at all. He con-
trasts the software discipline with electrical engineering and
feels it is a presumption to call what we do engineering. He may
be right that the field will never develop into an engineering dis-
cipline with as precise and all-encompassing a mathematical
base as electrical engineering has. After all, software engineer-
ing, like chemical engineering, is concerned with the nonlinear
problems of scaling up into industrial-scale processes, and like
industrial engineering, it is permanently confounded by the
complexities of human behavior.
Nevertheless, the course and timing of chemical engineer-
ing's development leads me to believe that software engineering
at age 27 may be not hopeless but merely immature, as chemical
engineering was in 1945. It was only after WWII that chemical
engineers really addressed the behavior of closed-loop intercon-
nected continuous-flow systems.
The distinctive concerns of software engineering are today
exactly those set forth in Chapter 1:
• How to design and build a set of programs into a system
" How to design and build a program or a system into a ro-
bust, tested, documented, supported product
• How to maintain intellectual control over complexity in large
doses.
The tar pit of software engineering will continue to be sticky for
a long time to come. One can expect the human race to continue
attempting systems just within or just beyond our reach; and
software systems are perhaps the most intricate of man's han-
The State and Future of Software Engineering 289
diworks. This complex craft will demand our continual devel-
opment of the discipline, our learning to compose in larger
units, our best use of new tools, our best adaptation of proven
engineering management methods, liberal application of com-
mon sense, and a God-given humility to recognize our fallibility
and limitations.
Fifty Years of Wonder,
Excitement, and Joy
Still vivid in my mind is the wonder and delight with which I—
then 13 years old—read the account of the August 7, 1944, ded-
ication of the Harvard Mark I computer, an electromechanical
marvel for which Howard Aiken was the architect and IBM
engineers Clair Lake, Benjamin Durfee, and Francis Hamilton
were the implementation designers. Equally wonder-provoking
was the reading of Vannevar Bush's "That We May Think" paper
in the April 1945 Atlantic Monthly, in which he proposed orga-
nizing knowledge as a big hypertext web and giving users ma-
chines for both following existing trails and blazing new trails of
associations.
My passion for computers got another strong boost in 1952,
when a summer job at IBM in Endicott, New York, gave me
hands-on experience in programming the IBM 604 and formal
instruction in programming IBM's 701, its first stored-program
machine. Graduate school under Aiken and Iverson at Harvard
made my career dream a reality, and I was hooked for life. To
only a fraction of the human race does God give the privilege of
earning one's bread doing what one would have gladly pursued
free, for passion. I am very thankful.
It is hard to imagine a more exciting time to have lived as a
computer devotee. From mechanisms to vacuum tubes to tran-
sistors to integrated circuits, the technology has exploded. The
first computer on which I worked, fresh out of Harvard, was the
291
292 Epilogue
IBM 7030 Stretch supercomputer. Stretch reigned as the world's
fastest computer from 1961 to 1964; nine copies were delivered.
My Macintosh Powerbook is today not only faster, with a larger
memory and bigger disk, it is a thousand times cheaper. (Five
thousand times cheaper in constant dollars.) We have seen in
turn the computer revolution, the electronic computer revolu-
tion, the minicomputer revolution, and the microcomputer rev-
olution, each bringing orders-of-magnitude more computers.
The computer-related intellectual discipline has exploded as
has the technology. When I was a graduate student in the mid-
1950s, I could read all the journals and conference proceedings;
I could stay current in all the discipline. Today my intellectual
life has seen me regretfully kissing subdiscipline interests good-
bye one by one, as my portfolio has continuously overflowed
beyond mastery. Too many interests, too many exciting oppor-
tunities for learning, research, and thought. What a marvelous
predicament! Not only is the end not in sight, the pace is not
slackening. We have many future joys.
Notes and References
Chapter 1
1. Ershov considers this not only a woe, but also a part of the
joy. A. P. Ershov, "Aesthetics and the human factor in pro-
gramming/' CACM, 15, 7 (July, 1972), pp. 501-505.
Chapter 2
1. V. A. Vyssotsky of Bell Telephone Laboratories estimates
that a large project can sustain a manpower buildup of 30
percent per year. More than that strains and even inhibits
the evolution of the essential informal structure and its com-
munication pathways discussed in Chapter 7.
F. J. Corbat<5 of MIT points out that a long project must an-
ticipate a turnover of 20 percent per year, and these must be
both technically trained and integrated into the formal struc-
ture.
2. C. Portman of International Computers Limited says,
"When everything has been seen to work, all integrated, you have
four more months work to do." Several other sets of schedule
divisions are given in Wolverton, R. W., "The cost of devel-
oping large-scale software," IEEE Trans, on Computers, C-23,
6 (June, 1974) pp. 615-636.
3. Figures 2.5 through 2.8 are due to Jerry Ogdin, who in quot-
ing my example from an earlier publication of this chapter
much improved its illustration. Ogdin, J. L., "The Mongo-
lian hordes versus superprogrammer," Infosy'stems (Dec.,
1972), pp. 20-23.
293
294 Notes and References
Chapter 3
1. Sackman, H., W. J. Erikson, and E. E. Grant, "Exploratory
experimental studies comparing online and offline pro-
gramming performance," CACM, 11, 1 (Jan., 1968), pp. 3-
11.
2. Mills, H., "Chief programmer teams, principles, and pro-
cedures," IBM Federal Systems Division Report FSC 71-
5108, Gaithersburg, Md., 1971.
3. Baker, F. T., "Chief programmer team management of pro-
duction programming," IBM Sys. J., 11, 1 (1972).
Chapter 4
1. Eschapasse, M., Reims Cathedral, Caisse Narionale des Mon-
uments Historiques, Paris, 1967.
2. Brooks, F. P., "Architectural philosophy," in W. Buchholz
(ed.), Planning A Computer System. New York: McGraw-Hill,
1962.
3. Blaauw, G. A., "Hardware requirements for the fourth gen-
eration," in F. Gruenberger (ed.), Fourth Generation Com-
puters. Englewood Cliffs, N.J.: Prentice-Hall, 1970.
4. Brooks, F. P., and K. E. Iverson, Automatic Data Processing,
System/360 Edition. New York: Wiley, 1969, Chapter 5.
5. Glegg, G. L., The Design of Design. Cambridge: Cambridge
Univ. Press, 1969, says "At first sight, the idea of any rules or
principles being superimposed on the creative mind seems more
likely to hinder than to help, but this is quite untrue in practice.
Disciplined thinking focusses inspiration rather than blinkers it."
6. Conway, R. W., "The PL/C Compiler," Proceedings of a Conf.
on Definition and Implementation of Universal Programming Lan-
guages. Stuttgart, 1970.
7. For a good discussion of the necessity for programming
technology, see C. H. Reynolds, "What's wrong with com-
Notes and References 295
puter programming management?" in G. F, Weinwurm
(ed.), On the Management of Computer Programming. Philadel-
phia: Auerbach, 1971, pp. 35-42.
Chapter 5
1. Strachey, C., "Review of Planning a Computer System," Comp.
/., 5, 2 (July, 1962), pp. 152-153.
2. This applies only to the control programs. Some of the com-
piler teams in the OS/360 effort were building their third or
fourth systems, and the excellence of their products shows
it.
3. Shell, D. L., "The Share 709 system: a cooperative effort";
Greenwald, I. D., andM. Kane, "The Share 709 system: pro-
gramming and modification"; Boehm, E. M., and T. B.
Steel, Jr., "The Share 709 system: machine implementation
of symbolic programming"; all in /ACM, 6, 2 (April, 1959),
pp.123-140.
Chapter6
1. Neustadt, R. E., Presidential Power. New York: Wiley, 1960,
Chapter 2.
2. Backus, J. W., "The syntax and semantics of the proposed
international algebraic language." Proc, Intl. Con/. Inf. Proc.
UNESCO, Paris, 1959, published by R. Oldenbourg, Mu-
nich, and Butterworth, London. Besides this, a whole col-
lection of papers on the subject is contained in T. B. Steel,
Jr. (ed.), Formal Language Description Languages for Computer
Programming. Amsterdam: North Holland, (1966).
3. Lucas, P., and K. Walk, "On the formal description of
PL/I," Annual Review in Automatic Programming Language.
New York: Wiley, 1962, Chapter 2, p. 2.
4. Iverson, K. E., A Programming Language. New York: Wiley,
1962, Chapter 2.
296 Notes and References
5. Falkoff, A. D., K. E. Iverson, E. H. Sussenguth, "A formal
description of System/360," ZBM Systems Journal, 3, 3 (1964),
pp. 198-261.
6. Bell, C. G., and A. Newell, Computer Structures. New York:
McGraw-Hill, 1970, pp. 120-136, 517-541.
7. Bell, C. G., private communication.
Chapter 7
1. Parnas, D. L., "Information distribution aspects of design
methodology," Carnegie-Mellon Univ., Dept. of Computer
Science Technical Report, February, 1971.
2. Copyright 1939, 1940 Street & Smith Publications, Copy-
right 1950, 1967 by Robert A. Heinlein. Published by ar-
rangement with Spectrum Literary Agency.
Chapter 8
1. Sackman, H., W. J. Erikson, and E. E. Grant, "Exploratory
experimentation studies comparing online and offline pro-
gramming performance," CACM, 11, 1 (Jan., 1968), pp. 3-
11.
2. Nanus, B., and L. Farr, "Some cost contributors to large-
scale programs," AFIPS Proc. SJCC, 25 (Spring, 1964), pp.
239-248.
3. Weinwurm, G. F., "Research in the management of com-
puter programming," Report SP-2059, System Development
Corp., Santa Monica, 1965.
4. Morin, L. H., "Estimation of resources for computer pro-
gramming projects," M. S. thesis, Univ. of North Carolina,
Chapel Hill, 1974.
5. Portman, C., private communication.
6. An unpublished 1964 study by E. F. Bardain shows pro-
grammers realizing 27 percent productive time. (Quoted by
Notes and References 297
D. B. Mayer and A. W. Stalnaker, "Selection and evaluation
of computer personnel/' Proc. 23rd ACM. Con/., 1968, p.
661.)
7. Aron, J., private communication.
8. Paper given at a panel session and not included in the
AFIPS Proceedings.
9. Wolverton, R. W., "The cost of developing large-scale soft-
ware," IEEE Trans, on Computers, C-23, 6 (June, 1974) pp.
615-636. This important recent paper contains data on many
of the issues of this chapter, as well as confirming the pro-
ductivity conclusions.
10. Corbato, F. J., "Sensitive issues in the design of multi-use
systems," lecture at the opening of the Honeywell EDP
Technology Center, 1968.
11. W. M. Taliaffero also reports a constant productivity of 2400
statements/year in assembler, Fortran, and Cobol. See
"Modularity. The key to system growth potential," Software,
I, 3 (July 1971) pp. 245-257.
12. E. A. Nelson's System Development Corp. Report TM-3225,
Management Handbook for the Estimation of Computer Program-
ming Costs, shows a 3-to-l productivity improvement for
high-level language (pp. 66-67), although his standard de-
viations are wide.
Chapter 9
1. Brooks, F. P. and K. E. Iverson, Automatic Data Processing,
System/360 Edition. New York: Wiley, 1969, Chapter 6.
2. Knuth, D. E., The Art of Computer Programming, Vols. 1-3.
Reading, Mass.: Addison-Wesley, 1968, ff.
Chapter 10
1. Conway, M. E., "How do committees invent?" Datamation,
14, 4 (April, 1968), pp. 28-31.
298 Notes and References
Chapter 11
1. Speech at Oglethorpe University, May 22, 1932.
2. An illuminating account of Multics experience on two suc-
cessive systems is in F. J. Corbatd, J. H. Saltzer, and C. T.
Clingen, "Multics—the first seven years," AFIPS Proc S/CC,
40 (1972), pp. 571-583.
3. Cosgrove, J., "Needed: a new planning framework," Data-
mation, 17, 23 (Dec., 1971), pp. 37-39.
4. The matter of design change is complex, and I oversimplify
here. See J. H. Saltzer, "Evolutionary design of complex sys-
tems," in D. Eckman (ed.), Systems: Research and Design.
New York: Wiley, 1961. When all is said and done, however,
I still advocate building a pilot system whose discarding is
planned.
5. Campbell, E., "Report to the AEC Computer Information
Meeting," December, 1970. The phenomenon is also dis-
cussed by J. L. Ogdin in "Designing reliable software," Da-
tamation, 18, 7 (July, 1972), pp. 71-78. My experienced
friends seem divided rather evenly as to whether the curve
finally goes down again.
6. Lehman, M., and L. Belady, "Programming system dynam-
ics," given at the ACM SIGOPS Third Symposium on Op-
erating System Principles, October, 1971.
7. Lewis, C. S., Mere Christianity. New York: Macmillan, I960,
p. 54.
Chapter 12
1. See also J. W. Pomeroy, "A guide to programming tools and
techniques," IBM Sys. J., 11, 3 (1972), pp. 234-254.
2. Landy, B., and R. M. Needham, "Software engineering
techniques used in the development of the Cambridge Mul-
tiple-Access System," Software, 1, 2 (April, 1971), pp. 167-
173.
Notes and References 299
3. Corbatd, F. J., "PL/I as a tool for system programming," Da-
tamation, 15, 5 (May, 1969), pp. 68-76.
4. Hopkins, M., "Problems of PL/I for system programming,"
IBM Research Report RC 3489, Yorktown Heights, N.Y.,
August 5, 1971.
5. Corbatd, F. J., J. H. Saltzer, and C. T. Clingen, "Multics—
the first seven years," AFIPS Proc SJCC, 40 (1972), pp. 571-
582. "Only a half-dozen areas which were written in PL/1 have
been receded in machine language for reasons of squeezing out the
utmost in performance. Several programs, originally in machine
language, have been receded in PL/I to increase their maintainabil-
ity."
6. To quote Corbato's paper cited in reference 3: "PL/I is here
now and the alternatives are still untested." But see a quite con-
trary view, well-documented, in Henricksen, J. O. and R. E.
Merwin, "Programming language efficiency in real-time
software systems," AFIPS Proc SJCC, 40 (1972) pp. 155-161.
7. Not all agree. Harlan Mills says, in a private communica-
tion, "My experience begins to tell me that in production program-
ming the person to put at the terminal is the secretary. The idea is
to make programming a more public practice, under common scru-
tiny of many team members, rather than a private art."
8. Harr, J., "Programming Experience for the Number 1 Elec-
tronic Switching System," paper given at the 1969 SJCC.
Chapter 13
1. Vyssotsky, V. A., "Common sense in designing testable
software," lecture at The Computer Program Test Methods
Symposium, Chapel Hill, N.C., 1972. Most of Vyssotsky's
lecture is contained in Hetzel, W. C. (ed.), Program Test
Methods. Englewood Cliffs, N.J.: Prentice-Hall, 1972, pp. 41-
47.
2. Wirth, N., "Program development by stepwise refine-
ment," CACM 14, 4 (April, 1971), pp. 221-227. See also
300 Notes and References
Mills, H. "Top-down programming in large systems," in R.
Rustin (ed.)- Debugging Techniques in Large Systems. Engle-
wood Cliffs, N.J.: Prentice-Hall, 1971, pp. 41-55 and Baker,
F. T., "System quality through structured programming,"
AFIPS Proc FJCC, 41-1 (1972), pp. 339-343.
3. Dahl, O. J., E. W. Dijkstra, and C. A. R. Hoare, Structured
Programming. London and New York: Academic Press, 1972.
This volume contains the fullest treatment. See also Dijk-
stra's germinal letter, "GOTO statement considered harm-
ful," CACM, II, 3 (March, 1968), pp. 147-148.
4. Bohm, C., and A. Jacopini, "Flow diagrams, Turing ma-
chines, and languages with only two formation rules,"
CACM, 9, 5 (May, 1966), pp. 366-371.
5. Codd, E. F., E. S. Lowry, E. McDonough, and C. A. Scalzi,
"Multiprogramming STRETCH: Feasibility considerations,"
CACM, 2,11 (Nov., 1959), pp. 13-17.
6. Strachey, C., "Time sharing in large fast computers," Proc.
Int. Con/, on Info. Processing, UNESCO (June, 1959), pp. 336-
341. See also Codd's remarks on p. 341, where he reported
progress on work like that proposed in Strachey's paper.
7. Corbat<5, F. J., M. Merwin-Daggett, R. C. Daley, "An exper-
imental time-sharing system," AFIPS Proc. SJCC, 2, (1962),
pp. 335-344. Reprinted in S. Rosen, Programming Systems
and Languages. New York: McGraw-Hill, 1967, pp. 683-698.
8. Gold, M. M., "A methodology for evaluating time-shared
computer system usage," Ph.D. dissertation, Carnegie-Mel-
lon University, 1967, p. 100.
9. Gruenberger, F., "Program testing and validating," Data-
mation, 14, 7, (July, 1968), pp. 39^7.
10. Ralston, A., Introduction to Programming and Computer Sci-
ence. New York: McGraw-Hill, 1971, pp. 237-244.
11. Brooks, F. P., and K. E. Iverson, Automatic Data Processing,
System/360 Edition. New York: Wiley, 1969, pp. 296-299.
Notes and References 301
12. A good treatment of development of specifications and of
system build and test is given by F. M. Trapnell, "A system-
atic approach to the development of system programs,"
AFIPS Proc S/CC, 34 (1969) pp. 411-418.
13. A real-time system will require an environment simulator.
See, for example, M. G. Ginzberg, "Notes on testing real-
time system programs," IBM Sys. /., 4, 1 (1965), pp. 58-72.
14. Lehman, M., and L. Belady, "Programming system dynam-
ics," given at the ACM SIGOPS Third Symposium on Op-
erating System Principles, October, 1971.
Chapter 14
1. See C. H. Reynolds, "What's wrong with computer pro-
gramming management?" in G. F. Weinwurm (ed.), On the
Management of Computer Programming. Philadelphia: Auer-
bach, 1971, pp. 35-42.
2. King, W. R., and T. A. Wilson, "Subjective time estimates
in critical path planning—a preliminary analysis," Mgt. Sci.,
13, 5 (Jan., 1967), pp. 307-320, and sequel, W. R. King, D.
M. Witterrongel, K. D. Hezel, "On the analysis of critical
path time estimating behavior," Mgt. Sci., 14, 1 (Sept.,
1967), pp. 79-S4.
3. For a fuller discussion, see Brooks, F. P., and K. E. Iverson,
Automatic Data Processing, System/360 Edition, New York:
Wiley, 1969, pp. 428-430.
4. Private communication.
Chapter 15
1. Goldstine, H. H., and J. von Neumann, "Planning and cod-
ing problems for an electronic computing instrument," Part
II, Vol. 1, report prepared for the U.S. Army Ordinance De-
partment, 1947; reprinted in J. von Neumann, Collected
Works, A. H. Taub (ed.), Vol. v., New York: McMillan, pp.
80-151.
302 Notes and References
2. Private communication, 1957. The argument is published in
Iverson, K. E., "The Use of APL in Teaching," Yorktown,
N.Y.: IBM Corp., 1969.
3. Another list of techniques for PL/I is given by A. B. Walter
and M. Bohl in "From better to best—tips for good program-
ming," Software Age, 3,11 (Nov., 1969), pp. 46-50.
The same techniques can be used in Algol and even For-
tran. D. E. Lang of the University of Colorado has a Fortran
formatting program called STYLE that accomplishes such a
result. See also D. D. McCracken and G. M. Weinberg,
"How to write a readable FORTRAN program," Datamation,
18, 10 (Oct., 1972), pp. 73-77.
Chapter 16
1. The essay entitled "No Silver Bullet" is from Information
Processing 1986, the Proceedings of the IFIP Tenth World
Computing Conference, edited by H.-J. Kugler (1986), pp.
1069-76. Reprinted with the kind permission of IFIP and
Elsevier Science B. V., Amsterdam, The Netherlands.
2. Parnas, D. L., "Designing software for ease of extension and
contraction," IEEE Trans, on SE, 5,2 (March, 1979), pp. 128-
138.
3. Booch, G., "Object-oriented design," in Software Engineering
with Ada. Menlo Park, Calif.: Benjamin/Cummings, 1983.
4. Mostow, J., ed., Special Issue on Artificial Intelligence and
Software Engineering, /£££ Trans, on SE, 11, 11 (Nov.,
1985).
5. Parnas, D. L., "Software aspects of strategic defense sys-
tems," Communications of the ACM, 28, 12 (Dec., 1985), pp.
1326-1335. Also in American Scientist, 73, 5 (Sept.-Oct.,
1985), pp. 432-440.
6. Balzer, R., "A 15-year perspective on automatic program-
ming," in Mostow, op. cit.
7. Mostow, op. cit.
8. Parnas, 1985, op. cit.
Notes and References 303
9. Raeder, G., "A survey of current graphical programming
techniques," in R. B. Grafton and T. Ichikawa, eds., Special
Issue on Visual Programming, Computer, 18, 8 (Aug., 1985),
pp. 11-25.
10. The topic is discussed in Chapter 15 of this book.
11. Mills, H. D., "Top-down programming in large systems,"
Debugging Techniques in Large Systems, R. Rustin, ed., Engle-
wood Cliffs, N.J., Prentice-Hall, 1971.
12. Boehm, B. W., "A spiral model of software development
and enhancement," Computer, 20, 5 (May, 1985), pp. 43-57.
Chapter 17
Material quoted without citation is from personal communica-
tions.
1. Brooks, F. P., "No silver bullet—essence and accidents of
software engineering," in Information Processing 86, H, J.
Kugler, ed. Amsterdam: Elsevier Science (North Holland),
1986, pp. 1069^1076.
2. Brooks, F. P., "No silver bullet—essence and accidents of
software engineering," Computer 20,4 (April, 1987), pp. 10-19.
3. Several of the letters, and a reply, appeared in the July, 1987
issue of Computer.
It is a special pleasure to observe that whereas "NSB" re-
ceived no awards, Bruce M. Skwiersky's review of it was se-
lected as the best review published in Computing Reviews in
1988. E. A. Weiss, "Editorial," Computing Reviews (June,
1989), pp. 283-284, both announces the award and reprints
Skwiersky's review. The review has one significant error:
"sixfold" should be "106."
4. "According to Aristotle, and in Scholastic philosophy, an ac-
cident is a quality which does not belong to a thing by right
of that thing's essential or substantial nature but occurs in it
as an effect of other causes." Webster's New International
Dictionary of the English Language, 2d ed., Springfield, Mass.:
G. C. Merriam, 1960.
304 Notes and References
5. Sayers, Dorothy L., The Mind of the Maker. New York: Har-
court, Brace, 1941.
6. Glass, R. L., and S. A. Conger, "Research software tasks:
Intellectual or clerical?" Information and Management, 23, 4
(1992). The authors report a measurement of software re-
quirements specification to be about 80% intellectual and
20% clerical. Fjelstadt and Hamlen, 1979, get essentially the
same results for application software maintenance. I know
of no attempt to measure this fraction for the whole end-to-
end task.
7. Herzberg, F., B. Mausner, and B. B. Sayderman. The Moti-
vation to Work, 2nd ed. London: Wiley, 1959.
8. Cox, B. J., "There is a silver bullet," Byte (Oct., 1990), pp.
209-218.
9. Harel, D., "Biting the silver bullet: Toward a brighter future
for system development," Computer (Jan., 1992), pp. 8-20.
10. Parnas, D. L., "Software aspects of strategic defense sys-
tems," Communications of the ACM, 28, 12 (Dec., 1985), pp.
1326-1335.
11. Turski, W. M., "And no philosophers' stone, either," in In-
formation Processing 86, H. J. Kugler, ed. Amsterdam: Elsev-
ier Science (North Holland), 1986, pp. 1077-1080.
12. Glass, R. L., and S. A. Conger, "Research Software Tasks:
Intellectual or Clerical?" Information and Management, 23, 4
(1992), pp. 183-192.
13. Review of Electronic Digital Computers, Proceedings of a Joint
AIEE-IRE Computer Conference (Philadelphia, Dec. 10-12,
1951). New York: American Institute of Electrical Engineers,
pp. 13-20.
14. Ibid., pp. 36, 68, 71, 97.
15. Proceedings of the Eastern Joint Computer Conference, (Washing-
ton, Dec. 8-10, 1953). New York: Institute of Electrical En-
gineers, pp. 45-47.
16. Proceedings of the 1955 Western Joint Computer Conference (Los
Notes and References 305
Angeles, March 1-3, 1955). New York: Institute of Electrical
Engineers.
17. Everett, R. R., C. A. Zraket, and H. D. Bennington,
"SAGE—A data processing system for air defense," Proceed-
ings of the Eastern Joint Computer Conference, (Washington,
Dec. 11-13, 1957). New York: Institute of Electrical Engi-
neers.
18. Harel, D., H. Lachover, A. Naamad, A. Pnueli, M. Politi, R.
Sherman, A. Shtul-Trauring, "Statemate: A working envi-
ronment for the development of complex reactive systems,"
IEEE Trans, on SE, 16, 4 (1990), pp. 403^44.
19. Jones, C., Assessment and Control of Software Risks. Engle-
wood Cliffs, N.J.: Prentice-Hall, 1994. p. 619.
20. Coqui, H., "Corporate survival: The software dimension,"
Focus '89, Cannes, 1989.
21. Coggins, James M., "Designing C + + libraries," C+ + Jour-
nal, \, 1 (June, 1990), pp. 25-32.
22. The tense is future; I know of no such result yet reported for
a fifth use.
23. Jones, op. cit., p 604.
24. Huang, Weigiao, "Industrializing software production,"
Proceedings ACM 1988 Computer Science Conference, Atlanta,
1988. I fear the lack of personal job growth in such an ar-
rangement.
25. The entire September, 1994 issue of IEEE Software is on
reuse.
26. Jones, op. cit., p. 323.
27. Jones, op. cit., p. 329.
28. Yourdon, E., Decline and Fall of the American Programmer. En-
glewood Cliffs, N.J.: Yourdon Press, 1992, p. 221.
29. Glass, R. L., "Glass"(column), System Development, (Janu-
ary, 1988), pp. 4-5.
306 Notes and References
Chapter 18
1. Boehm, B. W., Software Engineering Economics, Englewood
Cliffs, N.J.: Prentice-Hall, 1981, pp. 81^84.
2. McCarthy, J., "21 Rules for Delivering Great Software on
Time," Software World USA Conference, Washington (Sept.,
1994).
Chapter 19
Material quoted without citation is from personal communica-
tions.
1. On this painful subject, see also Niklaus Wirth "A plea for
lean software," Computer, 28, 2 (Feb., 1995), pp. 64-68.
2. Coleman, D., 1994, "Word 6.0 packs in features; update
slowed by baggage," MacWeek, 8, 38 (Sept. 26, 1994), p. 1.
3. Many surveys of machine language and programming lan-
guage command frequencies after fielding have been pub-
lished. For example, see J. Hennessy and D. Patterson,
Computer Architecture. These frequency data are very useful
for building successor products, although they never exactly
apply. I know of no published frequency estimates prepared
before the product was designed, much less comparisons of
a priori estimates and a posteriori data. Ken Brooks suggests
that bulletin boards on the Internet now provide a cheap
method of soliciting data from prospective users of a new
product, even though only a self-selected set responds.
4. Conklin, J., and M. Begeman, "gIBIS: A Hypertext Tool for
Exploratory Policy Discussion," ACM Transactions on Office
Information Systems, Oct. 1988, pp. 303-331.
5. Englebart, D., and W. English, "A research center for aug-
menting human intellect," AFIPS Conference Proceedings, Fall
Joint Computer Conference, San Francisco (Dec. 9-11, 1968),
pp. 395-410.
6. Apple Computer, Inc., Macintosh Human Interface Guidelines,
Reading, Mass.: Addison-Wesley, 1992.
Notes and References 307
7. It appears the Apple Desk Top Bus could handle two mice
electronically, but the operating system provides no such
function.
8. Royce, W. W., 1970. "Managing the development of large
software systems: Concepts and techniques," Proceedings,
WESCON (Aug., 1970), reprinted in the ICSE 9 Proceedings.
Neither Royce nor others believed one could go through the
software process without revising earlier documents; the
model was put forth as an ideal and a conceptual aid. See
D. L. Parnas and P. C. Clements, "A rational design pro-
cess: How and why to fake it," IEEE Transactions on Software
Engineering, SE-12, 2 (Feb., 1986), pp. 251-257.
9. A major reworking of DOD-STD-2167 produced DOD-STD-
2167A (1988), which allows but does not mandate more re-
cent models such as the spiral model. Unfortunately, the
MILSPECS that 2167A references and the illustrative exam-
ples it uses are still waterfall-oriented, so most procure-
ments have continued to use the waterfall, Boehm reports.
A Defense Science Board Task Force under Larry Druffel
and George Heilmeyer, in their 1994 "Report of the DSB task
force on acquiring defense software commercially," has ad-
vocated the wholesale use of more modern models.
10. Mills, H. D., "Top-down programming in large systems," in
Debugging Techniques in Large Systems, R. Rustin, ed. Engle-
wood Cliffs, N.J.: Prentice-Hall, 1971.
11. Parnas, D. L., "On the design and development of program
families," IEEE Trans, on Software Engineering, SE-2, 1
(March, 1976), pp. 1-9; Parnas, D. L., "Designing software
for ease of extension and contraction," IEEE Trans, on Soft-
ware Engineering, SE-5, 2 (March, 1979), pp. 128-138.
12. D. Harel, "Biting the silver bullet," Computer (Jan,, 1992),
pp. 8-20.
13. The seminal papers on information hiding are: Parnas,
D. L., "Information distribution aspects of design method-
ology," Carnegie-Mellon, Dept. of Computer Science, Tech-
308 Notes and References
nical Report (Feb., 1971); Parnas, D. L., "A technique for
software module specification with examples," Comm.
ACM, 5, 5 (May, 1972), pp. 330-336; Parnas, D. L. (1972).
"On the criteria to be used in decomposing systems into
modules," Comm. ACM, 5,12 (Dec., 1972), pp. 1053-1058.
14. The ideas of objects were initially sketched by Hoare and
Dijkstra, but the first and most influential development of
them was the Simula-67 language by Dahl and Nygaard.
15. Boehm, B. W., Software Engineering Economics, Englewood
Cliffs, N.J.: Prentice-Hall, 1981, pp. 83-94; 470-472.
16. Abdel-Hamid, T., and S. Madnick, Software Project Dynamics:
An Integrated Approach, ch. 19, "Model enhancement and
Brooks's law." Englewood Cliffs, N.J.: Prentice Hall, 1991.
17. Stutzke, R. D., "A Mathematical Expression of Brooks's
Law." In Ninth International Forum on COCOMO and Cost
Modeling. Los Angeles: 1994.
18. DeMarco, T., and T. Lister, Peopleware: Productive Projects and
Teams. New York: Dorset House, 1987.
19. Pius XI, Encyclical Quadragesima Anno, [Ihm, Claudia Car-
len, ed., The Papal Encyclicals 1903-1939, Raleigh, N.C.:
McGrath, p. 428.]
20. Schumacher, E. E, Small Is Beautiful: Economics as if People
Mattered, Perennial Library Edition. New York: Harper and
Row, 1973, p. 244.
21. Schumacher, op. cit., p. 34.
22. A thought-provoking wall poster proclaims: "Freedom of
the press belongs to him who has one."
23. Bush, V., "That we may think," ^Atlantic Monthly, 176, 1
(April, 1945), pp. 101-108.
24. Ken Thompson of Bell Labs, inventor of Unix, realized early
the importance of big screens for programming. He devised
a way to get 120 lines of code, in two columns, onto his
primitive Tektronix electron-storage tube. He clung to this
terminal through a whole generation of small-window, fast
tubes.
Index
Abdel-Hamid, T., 308 AppleScript, 287
abstract data type, 188, 220, architect, 37, 41, 54, 62, 66, 79,
273 100, 233, 236, 238, 255, 257
accident, 179,182, 209, 214, 272, architecture, 44, 143, 233, 234,
280, 281, 303, viii 245, 266
accounting, 132 archive, chronological, 33
Ada, 188, 283 aristocracy, 39, 44, 46
administrator, 33 Aristotle, 209, 303
Adobe Photoshop, 281 Aron, J., 90, 93, 237, 297
advancement, dual ladder of, ARPA network, 78
119, 242 artificial intelligence, 190, 302
advisor, testing, 192 assembler, 132
Aiken, H. H., 291 authority, 8, 80, 231, 236
airplane-seat metaphor, 194 AutoCad, 285
Algol, 34, 44, 64, 68, 203, 295, AutoLisp, 285
302 automatic programming, 302
algorithm, 102, 239
allocation, dynamic memory, 57 Bach, J. S., 47
alpha test, 142, 245, 266 Backus, J. W., 64, 295
alpha version, 240 Backus-Naur Form, 64
Alto personal workstation, 260 Baker, F. T., 36, 294, 300
ANSI, 168, 249 Balzer, R., 302
APL, 64, 98, 136, 175, 203, 302 Bardain, E. F., 2%
Apple Computer, Inc., 264, 306 barrier, sociological, 119
Apple Desk Top Bus, 307 Begeman, M., 306
Apple Lisa, 260 Belady, L., 122, 123, 150, 243,
Apple Macintosh, 255, 258, 264, 246, 298, 301
284, 291, 306 Bell Northern Research, 270
Note: Bold numerals indicate a relatively substantial discussion of a topic.
309
310 Index
Bell Telephone Laboratories, 90, building a program, 200
119, 133, 137, 142, 158, 237, bullet, brass, 219
293, xi, 308 silver, 179, 207, 212, 214, 226,
Bell, C. G., 64, 296, viii 303, 304
Bengough, W., 107 Burke, E., 253
Bennington, H. D., 305 Burks, A. W., 194
beta version, 240 Burris, R., xii
Bible, 255 Bush, V., 281, 291, 308
Bierly, R., viii Butler, S., 229
Blaauw, G. A., 45, 49, 62, 63, 294 buy versus build, 197
Bloch, E., i
Blum, B., 210 C + + , 220, 285,305
Boehm, B. W., 217, 237, 273, 283, Cambridge Multiple-Access
303, 306, 308, viii System, 298
Boehm, E. M., 295 Cambridge University, 133
Boes, H., ix Campbell, E., 121, 242, 298
Bohl, M., 302 Canova, A., 153
Bohm, C., 144, 300 Capp, A., 80
Booch, G., 302 Carnegie-Mellon University, 78
Boudot-Lamotte, E., 40 CASE statement, 144
brass bullet, 219 Case, R. P., viii, xi
breakthrough, 186 Cashman, T. J., 169
Breughel, P. , the Elder, 73 cathedral, 41
Brooks's Law, 25, 274 change summary, 77, 78
Brooks, F. P. Jr., 102, 226, 229, change, 117
237, 294, 297, 300, 301, 303, i control of, 149
Brooks, K. P., 216, 224, 306, viii design, 166, 241, 298
Brooks, N. G., v, viii organization, 118
Buchanan, B., viii changeability, 117, 184, 241
Buchholz, W., 294 channel, 45
budget, 6,108, 110, 239 chemical engineering, 116, 287,
access, 99, 238 288
size, 100, 238 chief programmer, 32, 232
bug, 142,143, 195, 209, 231, 235, ClarisWorks, 219
242, 243, 244, 245, 272 class, 189, 222, 225, 272
documented, 148 Clements, P. C., 307, 308
Build-every-night approach, 270 clerk, program, 33
build, incremental, 270 client-server system, 282
system, 147, 246, 301 Clingen, C. T., 298, 299
build-to-budget strategy, 268 COBOL, 199, 203, 218
build-up, manpower, 179 Codd, E. R, 146, 300
Index 311
Coding War Games, 276 Coqui, H., 217, 305
coding, 20, 237 Corbatd, F. J., 93, 146, 237, 293,
Coggins, J. M., 220, 221, 305, viii 297, 298, 299, 300, xi
Coleman, D., 306 Cornell University, 47
command key, 263 Cosgrove, J., 117, 118, 241, 298
command, 261, 286, 287, 306 cost, 6, 16, 87, 121, 182, 224, 233,
comment, 172, 249 242, 274
committee, 17, 74, 79 cost, development, 198
communication, 16, 17, 35, 54, front-loaded, 221
61, 73, 78, 79, 88, 100, 111, courage, managerial, 12, 21, 119,
183,232,233,234,235,236, 153, 221, 242, 274
240, 274 court, for design disputes, 66
compatibility, 63, 64, 68 Cox, B. J., 210, 212, 304
compile-time operation, 66 Crabbe, G., 163
compiler, 132 creation, component stages, 15,
complexity, 182, 211, 226, 233, 45, 143
288 creative joy, 7, 120, 280
arbitrary, 184, 211 creative style, 47
conceptual, 210 creative work, 46
component debugging, 144 creativity, 278, 280
component, 223, 230, 239, 284, critical-path schedule, 89,156,
286 158, 247, 301
dummy, 148 Crockwell, D., 87
comprehensibility, 186 Crowley, W. R., 132
computer facility, 128 cursor, 261
conceptual construct, 182, 186, customizability, 219
209 customization, 222
conceptual integrity, 35, 36, 42,
62, 80, 142, 184, 232, 233, d'Orbais, J., 41
255, 257, 260, 264 Dahl, O. J., 300, 308
conceptual structure, 180 Daley, R. C, 300
conference, 66 data base, 108
conformity, 184 data service, 131
Conger, S. A., 214, 304 database, 198, 223, 240, 283, 285
Conklin, J., 259, 306 datatype, abstract, 189
control program, 91, 93 date, estimated, 158
convergence of debugging, 9 scheduled, 158
Conway, M. E., Ill, 297 debugging aid, 128
Conway, R. W., 47, 294 debugging, component, 144
Cooley, J. W., 102 high-level language, 135
copilot, 32 interactive, 34, 136, 146, 245
312 Index
debugging (continued) DOD-STD-2167, 266, 307
on-machine, 145 • DOD-STD-2167A, 307
sequential nature of, 17 Dragon voice recognition system,
system, 147 264
DEC PDP-8, 64 Druffel, L., 307
DEC VMS operating system, 284 dual ladder of advancement, 119,
DECLARE, 174 242
Defense Science Board Task Force dummy component, 148
on Military Software, i, vii, dump, memory 133, 145
viii Durfee, B., 291
Defense Science Board, 307
DeMarco, T., 218, 223, 224, 276, ease of use, 43, 98, 255, 258, 260,
283, 308, viii 262,263
democracy, 44 Eastman Kodak Company, 285
Department of Defense, 266 Eckman, D., 298
dependability of debugging editor, job description for, 33
vehicle, 131 text, 32, 34, 68, 128, 133, 134,
description; See specification, 146
design change, 241 Einstein, A., 213
design-for-change, 272 electronic mail, 234, 235
designer, great, 180, 202, 304 electronic notebook, 78, 235
desktop metaphor, 194, 260, 262 Electronic Switching System, 90
development, incremental, 200 encapsulation, 78, 220, 236, 271
diagram, 216 Engelbart, D. C, 78, 260, 306
difference in judgement, 35 English, W., 306
difference of interest, 35 entropy, 122, 243
Digitek Corporation, 102 environment, 6, 165,196
Dijkstra, E. W., 144, 300, 308 Erikson, W. J., 29, 30, 88, 294,
director, technical, role of, 79, 296
236,256 Ershov, A. P., 293, xi
discipline, 46, 54, 55, 233 Eschapasse, M., 294
Disk Operating System, IBM essence, 179,181,196, 214, 222,
1410-7010, 56, 57, 99 272, 285, 303, viii
display terminal, 78, 129 estimating, 14, 21, 88, 109, 155,
division of labor, 79, 236 231, 237, 239, 247, 301
DO... WHILE, 144 Evans, B. O., v
document, 107, 239 Everett, R. R., 305
documentation system, 134, 244 Excel, 285, 286
documentation, 6, 32, 33, 122, expert system, 191
164, 224, 235, 248 extension, 221, 268, 302
Index 313
Fagg, P., 24 GO TO, 170
Falkoff, A. D., 296 God, 42, 184, 232, 289, 291, £x
family, software product, 268 Godel, 213
Farr, L., 88, 296 Goethe, J. W. von, 163
Fast Fourier Transform, 102 Gold, M. M., 146, 246, 300
featuritis, 257 Goldstine, H. H., 170,194, 301
Feigenbaum, E. A., 191 Gordon, P., mi
Ferrell, J., 287 GOTO, 300
file, dummy, 148 Grafton, R. B., 302
miniature, 148 Grant, E. E., 29, 30, 88, 294, 296
filters, 187 \graph, 185, 216
Fjelstadt, 304 \ structure, 248
floorspace, 239 graphical programming, 194, 302
flow arrow, 174 great designer, 180, 202, 284
flow chart, 167, 185, 194, 248, 300 Greenwald, I. D., 295
forecast, 109, 239 growing software, 180, 200, 212,
formal definition, 63, 234 268
formal document, 111 Gruenberger, R, 147, 294, 300
formal progression of release, 133
formality, of written proposals, Hamilton, F., 291
67 Hamlen, 304
Fortran, 45, 102, 203, 302 hardware, computer, 181
Fortran, H., 99 Hardy, H., 97
FoxPro database, 287 Harel, D. L., 212, 214, 270, 304,
Franklin, B. (Poor Richard), 87 305, 307
Franklin, J. W., 134 Harr, J., 90, 93,137, 237, 299, xi
frequency data, 306 Hayes-Roth, R., viii
frequency guessing, 257, 259 Heilmeyer, G., 307
fusion, 277 Heinlein, R. A., 81, 296
Hennessy, J., 306
Galloping Gertie, Tacoma Henricksen, J. O., 299
Narrows Bridge, 264 Henry, P., 253
Gantt chart, 265 Herzberg, F., 210, 304
General Electric Company, 216 Hetzel, W. C, 299
generator, 193, 283 Hezel, K. D., 301
gIBIS, 259, 306 hierarchical structure, 189, 212,
Ginzberg, M. G., 301 220
Glass, R. L., 214, 226, 304, 305 high-level language, See
Glegg, G. L., 294 language, high-level
Global Positioning System, 257 Hoare, C. A. R., 300, 308
314 Index
Homer, 255 IF...THEN...ELSE, 144
Hopkins, M., 299 Ihm, C. C, 308
Huang, W., 222, 305 implementation, 15, 45, 64, 143,
hustle, 155, 247 209, 233, 234, 238, 256, 266
HyperCard, 285 implementations, multiple, 68
hypertext, 281, 291, 306 implementer, 47, 54, 62, 66
incorporation, direct, 66, 118,
IBM 1401, 45, 65, 130 241, 264
IBM 650, 43, 102 incremental development, 200,
IBM 701, 131 268
IBM 7030 Stretch computer, 44, incremental-build model, 212,
47, 55, 291, 300, i 267,270
IBM 704, 55 indenting, 174
IBM 709, 55, 57, 181 information hiding, 78, 271, 308
IBM 7090, 55, 64 information theory, 212
IBM Corporation, 90, 119, 291, vii inheritance, 220, 222, 273
IBM Harvest computer, i initialization, 174
IBM MVS/370 operating system, input range, 6,165, 248
276, 284 input-output format, 165
IBM Operating System/360, See instrumentation, 129
Operating System/360 integrity, conceptual, 35, 36, 42,
IBM OS-2 operating system, 284 43, 62, 80, 142, 255, 257, 260,
IBM PC computer, 260, 264, 284 264
IBM SAGE ANFSQ/7 data interaction, as part of creation,
processing system, 216, 305 15, 209
IBM System/360 computer family first of session, 146
44, 45, 62, 64 interactive debugging, 34,146
IBM System/360 Model 165, 98 interactive programming, 136,
IBM System/360 Model 30, 45, 47 244, 245, 246
IBM System/360 Model 65, 99 interface, 6, 32, 62, 66, 79, 118,
IBM System/360 Model 75, 47 120, 122, 241, 243, 255, 257,
IBM. System/360 Principles of 264, 271, 282, 286, 306
Operation, 62 metaprogramming, 287
IBM VM/360 operating system, module, 268
284 WIMP, 234, 260, 263
IBSYS operating system for the Interlisp, 187
7090, 56 International Computers Limited,
Ichikawa, T., 302 89, 133, 293, xi
icon, 260 Internet, 306
ideas, as stage of creation, 15 interpreter, for space-saving, 102
IEEE Computer magazine, vii invisibility, 185, 216, 241
Index 315
iteration, 199, 267 class, 225
Iverson, K. E., 64, 102, 170, 291, macro, 34
294, 295, 296, 297, 300, 301, program, 132
302 linkage editor, 56, 282
Lister, T., 276, 308
Jacopini, A., 144, 300 Little, A. D., 287
Jobs, S., 260 Locken, O. S., 76
Jones, C., 217, 218, 222, 223, 224, Lowry, E. S., 300
305 Lucas, P., 295
joys of the craft, 7 Lukasik, S., 211
Kane, M., 295 Macintosh WIMP interface, 234,
keyboard, 262 260, 263
Keys, W. J., 169 Madnick, S., 274, 308
King, W. R., 301 magic, 7, 142, 226
Knight, C. R., 3 maintenance, 118, 242
Knuth, D. E., 102, 297 man-month, 16, 231, 273
Kugler, H. J., 303 management information system
(MIS), 107, 111, 219, 239,
label, 174 240, 285
Lachover, H., 305 manual, 62, 239, 258, 263
Lake, C., 291 System/360, 62
Landy, B., 298 market, mass, 180,197, 218, 223,
Lang, D. E., 302 258
language description, formal, 181 matrix management, 222
language translator, 93 matrix-type organization, 79
language, fourth-generation, 283 Mausner, B., 210, 304
high-level, 118,135, 143, 146, Mayer, D. B., 297
186, 194, 225, 237, 241, 244, McCarthy, J., 247, 270, 278, 306,
245, 248, 249 via
machine, 180, 225 McCracken, D. D., 302
programming, 68, 180, 186, 225 McDonough, E., 300
scripting, 287 Mealy, G., xi
late project, 13, 89, 217, 235, 246, measurement, 222
275, 306 medium of creation, tractable, 7,
lawyer, language, 34 15, 117
Lehman, M., 122, 123, 150, 243, meeting, problem action, 157
246, 298, 301 status review, 75, 157
Lewis, C. S., 123, 298 memory use pattern, 129, 239
library, 187, 222, 239, 244, 272, mentor, 203, 275
305 menu, 260
316 Index
Merwin, R. E., 299 multiple implementations, 68
Merwin-Dagget, M., 300 MVS/370, 203
metaphor, 260
metaprogramming, 285 Naamad, A., 305
microcomputer revolution, 214, Nanus, B., 88, 296
279 Naur, P., 64
microfiche, 77, 235 Needham, R. M., 298
Microsoft Corporation, 246, 270 Nelson, E. A., 297
Microsoft Windows, 260 nesting, as documentation aid,
Microsoft Word 6.0, 258, 306 172
Microsoft Works, 219 network nature of
milestone, 22, 25,154, 158, 247, communication, 79
248, 270 Neustadt, R. E., 295
Mills, H. D., 32, 33, 201, 267, Newell, A., 64, 296
271, 294, 299, 300, 303, 307 Noah, 97
MILSPEC documentation, 248 North Carolina State University,
mini-decision, 63, 111, 234, 240 287
MiniCad design program, 285 notebook, status, 33
MIT, 93, 121, 146, 287, 293, xi system, 147
mnemonic name, 174 Nygaard, 308
model, 255, 256, 274, 307
COCOMO, 273 object 285
incremental-build, 212, 267, object-oriented design, 302
270 object-oriented programming,
spiral, 303, 307 189, 219, 273
waterfall, 264, 307 objective, 8, 75,108,110, 117, 239
Modula, 189, 203 cost and performance, 49
modularity, 118, 188, 220 space and time, 49
module, 101,122, 143, 241, 243, obsolescence, 9, 26,123
245, 269, 271, 273, 285 off-the-shelf package, 198
modules, number of, 122 office space, 242
Mooers, C. N., 44 Ogdin, J. L., 293, 298
Moore, S. E., xii open system, 283
Morin, L. H., 88, 296 operating system, 128, 238, 243,
Mostow, J., 302 283
mouse, 307 Operating System/360, 43, 45, 47,
moving projects, 277 56, 76, 93, 129, 234, 235, 237,
Mozart, W. A. 202 243, 271, 276, 295, i, x
MS-DOS, 203, 255, 284 optimism, 14, 212, 231
Multics, 93, 136, 146, 237, 298, option, 101, 165, 238
- 299,300 Orbais, J. d', 41
Index 317
order-of-magnitude pilot plant, 116,240
improvement, 208, 213, 215, pilot system, 298
281, 291, vii pipes, 187
organization chart, 108, 111, 239 Pisano, A., 127
organization, 74, 78,118, 235, Pius XI, 277, 308
236,242 PL/C language, 47, 294
OS/360 Concepts and Facilities, 134 PL/I, 32, 47, 64, 66, 93, 135, 172,
OS/360 Queued 203, 245, 299, 302
Telecommunications Access planning, 20
Method, 285 Plans and Controls organization,
OS/360, See Operating System/ 160, 248
360 playpen, 133, 149, 244, 246
overlay, 54, 99, 129 Pnueli, A., 305
overview, 165, 248 pointing, 260
Ovid, 55 policed system, 65
Politi, M., 305
Padegs, A., 62 Pomeroy, J, W., 298
paperwork, 108 Poor Richard (Benjamin
Parnas families, 268 Franklin), 87
Parnas, D. L., 78, 190, 193, 212, Pope, Alexander, 207
221, 224, 226, 236, 268, 271, Portman, C, 89, 237, 293, 296, xi
288, 296, 302, 304, 307, 308, power tools for the mind, 219
via power, giving up, 277
partitioning, 16, 231 practice, good software
Pascal programming language, engineering, 193, 202
203, 285 price, 109
Pascal, B., 123 PROCEDURE, 174
pass structure, 166 procedure, catalogued, 34
Patrick, R. L., vii producer, role of, 79, 236, 256
Patterson, D., 306 product test, 69, 142, 234, 245
people, 29, 202, 276, 284 product, exciting, 203
Peopleware: Productive Projects and programming system, 4, 230
Teams, 276 programming, 5, 288
perfection, requirement for, 8 productivity equation, 197
performance simulator, 134 productivity, programming, 21,
performance, 182, 258 30, 88, 94, 135, 181, 186, 208,
PERT chart, 89, 156, 158, 247 213, 217, 237, 244, 245, 273,
pessimism, 212 276, 284
Peter the Apostle, 171 program clerk, 33
philosopher's stone, 304 program library, 132
Piestrasanta, A. M., 160, xi program maintenance, 120
318 Index
program name, 174 Ralston, A., 300
program products, 273 rapid prototyping, 180,199, 270
program structure graph, 170, real-time system, 301
185 realism, 212, 226
program, 4 realization, step in creation, 49,
auxiliary, 149 143, 256, 266
self-documenting, 171 refinement, progressive, 143,
programmer retraining, 220, 221 267, 299
programming environment, 187 requirements, 199
programming language, 221 regenerative schedule disaster, 21
programming product, 5, 116, Reims Cathedral, 41
240 release, program, 121, 185, 217,
programming system, 6 243, 244
programming systems product, reliability, 186
4, 230 remote job entry, 58
programming systems project, repartitioning, 24, 232, 275
237 representation, of information,
programming, automatic, 193 102, 239
graphical, 194 requirements refinement, 199
visual, 194 rescheduling, 24
progressive refinement, 267, 299 responsibility, versus authority,
Project Mercury Real-Time 8,231
System, 56 Restaurant Antoine, 13 /
project workbook, 235 reusable component, 210/
promotion, in rank, 120 reuse, 222, 224, 269, 273, 285
prototyping, rapid, 180,199, 270 Reynolds, C. H., 294, 301
Publilius, 87 role conflict, reducing, 157
purple-wire technique, 149 ROM, read-only memory, 234
purpose, of a program, 165, 249 Roosevelt, F. D., 115, 298
of a variable, 174 Rosen, S., 300
Royce, W. W., 265, 307
Quadragesima Anno, Encyclical, Rustin, R., 300, 303, 307
277,308 Ruth, G. H. (Babe), 87
quality, 217
quantization, of change, 62,118, Sackman, H., 20, 29, 88, 294, 296
150,246 Salieri, A., 202
of demand for change, 117 Saltzer, J. H., 298, 299
Sayderman, B. B., 210, 304
Raeder, G., 302 Sayers, D. L., 15, 209, 303
raise in salary, 120 scaffolding, 34, 148, 246
Index 319
scaling up, 36,116, 240, 288 logic, 65, 131
Scalzi, C A., 300 performance, 99
schedule, 79,108, 111, 154, 239, size, program, 30, 98,129,135,
244, 247, 265, 273, See Late 175
project Skwiersky, B. M., 303
cost-optimum, 274 slippage, schedule, See Late
scheduler, 57 project
scheduling, 14, 129 Sloane, J. C., xii
Schumacher, E. F., 277, 279, Small is Beautiful, 277
308 Smalltalk, 203, 220
screen, 194, 201, 268, 287, 308 Smith, S., 97
second-system effect, 51, 234, snapshot, 145
257,259 Snyder, Van, 222
secretary, 33 sociological barrier, 119
security, 183 Sedahl, L., 211
self-documenting program, 118, Software Engineering Economics,
171, 249 273
Selin, I., 219 Software Engineering Institute,
semantics, 44, 64, 66, 225, 261, 202
272 software industry, 282
Shakespeare, W., 141, 255 Software Project Dynamics, 274
Shannon, E. C. , 212 Sophocles, 153,155
Share 709 Operating System space allocation, 108, 111
(SOS), 295 space, memory, 238
Share Operating System for the office, 203, 276
IBM 709, 57 program, See size, program,
Shell, D. L., 295 specialization of function, 35, 79,
Sherman, M., 189 236
Sherman, R., 305 specification, 195, 200, 245, 256,
short cuts, 263 266, 301, 304
shrink-wrapped software, 218, architectural, 43, 142, 245
219, 223, 257, 279, 282, 284 functional, 32, 48, 49, 62, 75,
Shtul-Trauring, A,, 305 108, 110
side effect, 65, 122, 243 interface, 75
silver bullet, 179, 207, 212, 214, internal, 75
226, 303, 304 performance, 32
simplicity, 44, 181, 186 testing the, 142
Simula-67, 189, 308 speed, program, 30, 98,135
simulator, 234, 244 spiral, pricing-forecasting, 109
environment, 131 spreadsheet, 198, 280, 281
320 Index
staff group, 79 system integration sublibrary, 133
staffing, project, 21, 273 system test, 19, 122,133,147
Stalnaker, A. W., 297 system, large, 31
standard, 75, 168, 249, 283 programming, 6, 288
standard, de facto, 264 System/360 computer family, 296,
Stanford Research Institute, 78, 301
260 systems product, programming,
Stanton, N., ix 4,288
start-up firm, 278, 284
Statemate design tool, 305 Tacoma Narrows Bridge, 115, 264
status control, 108 Taliaffero, W. M., 297
status report, 157, 247 target machine, 129, 243, 244
status review meeting, 157 Taub, A. H., 301
status symbol, 81 Taylor, B., 260
Steel, T. B., Jr., 295 team, small, sharp, 30
Strachey, C, 56, 146, 295, 300 technical director, role of, 79,
straightforwardness, 44 236, 256
Strategic Defense Initiative, 257 technology, programming, 49,
Stretch Operating System, 56, 99 102,128
structured programming, 32,144, telephone log, 68, 234
218, 241, 245 test case, 6, 34, 147, 166, 192,
stub, 267 248, 270
Stutzke, R. D., 275, 308 test, component, 20
subroutine, 182, 272 system, 19, 122,133,147
Subsidiary Function, Principle of, tester, 34
277 testing advisor, 192
superior-subordinate testing, 6
relationship, 35 regression, 122, 242, 268
supervisory program, 146 specification, 142
support cost, 218 TESTRAN debugging facility, 57,
surgical team, 27, 120, 232 146
Sussenguth, E. H., 296 text-editing system, 134, 244
Swift, J., 115 Thompson, K., 308
synchronism in file, 171 throw-away, 116, 241, 264
syntax, 44, 64, 66, 225, 262 time, calendar, 14, 231
abstract, 64 machine, 180
system build, 147, 246, 301 Time-Sharing System, PDP-10,
system debugging, 132, 147 43
System Development Time-Sharing System/360, 136,
Corporation, 88, 297 146
Index 321
time-sharing, 187, 280, 282, 300 Vanilla Framework design
tool, 125,196, 243, 258, 280 methodology, 216
power, for the mind, 219 vehicle machine, 131
toolsmith, 34, 128 verification, 195
top-down design, 220, 134, 143, version, 118, 185, 217, 241, 268,
245 270
top-down programming, 303 alpha, 240
Tower of Babel, 72, 235 beta 240
TRAC language, 44 Vessey, 214
tracing, program 146 virtual environment, 281, i, viii
trade-off, size-function, 101 virtual memory, 238
size-speed, 99, 101 visual programming, 302
training, time for, 18 visual representation, 216
transient area, 101, 238 vocabularies, large, 224
Trapnell, F. M., 301, x, xi Voice Navigator voice recognition
tree organization, 79 system, 264
Truman, H. S., 61 von Neumann, J., 170, 194, 301
TRW, Inc. , 273 Vyssotsky, V. A., 142, 158, 179,
Tukey, J. W., 102 185, 245, 248, 293, 299, xi
turnaround time, 136, 187, 245,
281 Walk, K. 295
turnover, personnel, 184 Walter, A. B., 302
Turski, W. M., 213, 304 Ward, F., viii
two-cursor problem, 261 waterfall model, 264, 307
two-handed operation, 262 Watson, T. J., Jr., v, xi
type, abstract data, 188 Watson, T. J., Sr., 164
type-checking, 220 Weinberg, G. M., 302
Weinwurm, G. F., 88, 295, 296,
Univac computer, 215 301
Unix workstation, 282 Weiss, E. A., 303
Unix, 187, 197, 203, 244, 284, 287, Wells Apocalypse, The, 61
308 werewolf, 178, 180, 208
University of North Carolina at Wheeler, E., 279, viii
Chapel Hill, i William III of England, Prince of
user, 45, 117, 121, 165, 255, 258, Orange, 207
261, 266, 268, 271, 286 Wilson, T. A., 301
novice, 263 WIMP interface, 234, 260, 263
power, 263 window, 260
USSR Academy of Sciences, xi Windows NT operating system,
utility program, 34, 128, 134 234
322 Index
Windows operating system, 284 Xerox Palo Alto Research Center,
Wirth, N., 143, 245, 299, 306 260
Witterrongel, D. M., 301
Wizard-of-Oz technique, 271
Wolverton, R. W., 293, 297 Yourdon, E., 218, 223, 224, 305,
workbook, 75 viii
workstation, 196
World-Wide Web, 235, 281
wormhole, 287 zoom, 165, 248
Wright, W. V., 167 Zraket, C. A., 305