Why the Social Sciences Won't Become High-Consensus, Rapid-Discovery Science

Author(s): Randall Collins

Reviewed work(s):
Source: Sociological Forum, Vol. 9, No. 2, Special Issue: What's Wrong with Sociology? (Jun.,
1994), pp. 155-177
Published by: Springer
Stable URL: http://www.jstor.org/stable/685040 .
Accessed: 30/06/2012 12:00

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

Springer is collaborating with JSTOR to digitize, preserve and extend access to Sociological Forum.

http://www.jstor.org
Sociological Forum, VoL 9, No. 2, 1994

Why the Social Sciences Won't Become

High-Consensus, Rapid-Discovery Science
Randall Collins1

A researchfront of rapid discovery,leaving a trail of cognitive consensus behind

it, is characteristicof natural sciences since about the 17th century in Europe.
The basis of this high-consensus, rapid-discoveryscience is not empiricism,
since empirical research existed in the natural sciences before the 17th century.
The key is appropriation of genealogies of research technologies, which are
pragmatically manipulated and modified to produce new phenomena; high
consensus results because there is higher social prestige in moving ahead to
new research discoveries than by continuing to dispute the interpretation of
older discoveries. The social sciences have not acquired this pattern of rapid
discovery with high consensus behind the research front. Their fundamental
disability is not lack of empirical research, nor failure to adhere to a scientific
epistemology, nor the greater ideological controversy that surrounds social
topics. What is fundamentally lacking in the social sciences is a genealogy of
research technology, whose manipulation reliably produces new phenomena
and a rapidly moving research front. Unless the social sciences invent new
research hardware, they will likely never acquire much consensus or rapid
discovery. Possibilities may existfor such development stemming from research
technologies in microsociology and in artificial intelligence.
KEY WORDS: science; research technology; networks; methodology.

INTRODUCTION

In the postmodernist atmosphere of the late 20th century, it has be-

come unfashionable to ask about the scientific prospects of sociology and

'Department of Sociology, University of California, Riverside, Riverside, California 92521-

0419.

155

0884-8971/94/0600-{)155$07.00/0 O 1994 Plenum Publishing Corporation

156 Collins

the other social disciplines. The natural sciences are regarded by many in-
tellectuals as authoritarian and destructive; hence in the social disciplines
we should not try to become a science even if we could. On the other side,
science was once considered a liberalizing and enlightening movement in
Western culture, and some persons continue to work at exploring the social
realm in the same spirit. These contentions are moot, if there is no possi-
bility that the social disciplines can be sciences in the same sense as the
natural sciences. Let us step back from the polemics and consider the issue
using the resources of the sociology of science.
I stress the social organization of intellectual fields in order to avoid
the ideological and philosophical terms in which debates over the scientific
character of the social "sciences" have often been carried out. Philosophical
efforts at demarcation between science and nonscience have a long history
of difficulties. The logical positivist program of reducing true scientific
knowledge to observation statements plus the tautological calculus of
mathematics foundered on the inability to make sense of its own metas-
tatements, as well as on paradoxes in the foundations of mathematics. His-
torians of science and social ethnographers of laboratory life have shown
that criteria such as precision, verifiability, and falsifiability are not often
used in practice, that they are ideals or after-the-fact reconstructions. Lat-
ter-day philosophers from the positivist tradition, such as Quine, hold that
no theory is conclusively verified or defeated by any particular empirical
observation or logical inconsistency. At best science is justified by a global
pragmatism; some scientific theories work better than others, on the whole
and in the long run, while others break down and undergo Kuhnian para-
digm shifts. Why this is so continues to be debated by philosophers of sci-
ence, thus far without definite conclusions.
Does this mean that all disciplines are the same, that physics and
theology, literary criticism, and sociology are all on the same plane? Post-
modernists revel in dissolving boundaries, and the failure of positivist phi-
losophy seems to give grounds for regarding every discipline's knowledge
claims as dubious and relativistic. Nevertheless, there is an appearance of
sophistry in claiming that all the disciplines are the same. There are social
facts to be accounted for: massive differences in their material equipment,
sources of funding, social prestige, patterns of recruitment and training.
Epistemology is not a good basis for understanding why such differences
exist and how they have come about over the past few centuries. A more
fruitful approach is to ask, What is distinctive about the social organization
of the disciplines that we now take as natural science, and do the social
disciplines have (or can they acquire) the conditions that make possible
that kind of organization?
Social Sciences Won't Become High-Consensus, Rapid-Discovery 157

THE EMERGENCE OF HIGH CONSENSUS AND RAPID

DISCOVERY IN INTELLECTUAL COMMUNITIES

In Europe around 1600 emerged a form of scientific organization

characterized by two distinctive traits: high consensus on what counts as
secure knowledge and rapid-discoveryof a train of new results.2 These traits
did not exist previously, even in the subject matter of natural sciences; it
was not science in general that was created in this period, but the organi-
zation of high-consensus rapid-discovery science. (To avoid the clumsiness
of this term, I will hereafter abbreviate it to rapid-discovery science.) The
adaptation of this organization, usually known as the scientific revolution,
was in fact not so sudden; from a few areas, notably mechanics and mathe-
matics, it spread during the next 300 years to encompass most of the natural
sciences.
High consensus contrasts not only with the pattern of the humanistic
disciplines, which have remained outside the orbit of modern science, but
with the typical mode of intellectual life throughout world history. In every
civilization since the invention of writing, the undifferentiated intellectual
role has been that of the philosopher. In this field, in China and India, in
ancient Greece, and in the medieval and modern West, the pattern has
been for major innovators in philosophy to appear in clumps: that is to
say, rival positions are created at the same time, and opposing schools of
thought maintain themselves over many generations without one of them
establishing ascendancy. The basic pattern of intellectual life is not con-
sensus, but disagreement. This does not mean that philosophy is totally
unstructured; it is structured precisely by its rivalries, by its focus upon is-
sues of contention. The rivalry is structured socially by the intergenerational
networks that connect eminent individual philosophers with each other: in
any period of creative life, there are typically between three and six such
lineages or schools, a pattern that I have labeled the Law of Small Num-
bers. Times when the number of schools exceeds the upper limit of six are
periods of structural instability and transition; during these times there is
a sense of intellectual crisis, followed by restructuring in the following gen-
erations, which reduces the number of lineages to the normal ceiling. The
dynamics of intellectual life in the philosophical mode have been driven
by the struggle to appropriate a portion of this limited attention space, to

2Detailed documentation cannot be presented in this paper. The analysis derives from a
comparative study of long-term intergenerational networks of philosophers, including their
overlap with European scientists and mathematicians in the period between 1500 and 1900.
Sources of data are given in my manuscript in progress, The Sociology of Philosophies: A
Global Theory of Intellectual Chlange. Portions are available in Collins (1987, 1989).
158 Collins

be one of the small number of contending parties. Success does not consist
in being a sole hegemon, but in finding a place within the action of a mul-
tisided conflict.
Prior to the scientific revolution and the adoption of the rapid-dis-
covery mode, the fields of natural science also generally had this structure
of disagreement. Often natural science was pursued by the general-purpose
intellectual, the philosopher; thus there were Aristotelian and Stoic physics,
various brands of Taoist and Confucian cosmology, and so forth. Even when
there were specialists in a field that would be considered a natural science
today, such as medicine or astronomy, there were typically rival schools:
for example, in Greece between 300 BC and 200 AD a variety of systems
of astronomy were used; there were five in medieval India; throughout an-
cient and medieval China, there were between three and five rival astrono-
mies (Neugebauer, 1957; Jones, 1991; Dictionary of Scientific Biography,
Vol. 15:533-632; Needham, 1959:171-436; Siven, 1969). Greek medicine
was divided in a fluctuating alignment among as many as half a dozen
schools; Chinese medicine was divided among a variety of Taoist and non-
Taoist schools (Frede, 1987:236-260; Welch and Seidel, 1979).
In this perspective, there is no abnormality in the proliferation of
schools and the lack of consensus that characterizes 20th-century sociology
or anthropology. These fields are much more typical of the form intellectual
life has taken throughout history than are the rapid-discovery sciences of
the last few centuries. Social scholars tend to regard their disagreements
as a pathology of their discipline; but they are merely operating according
to the dynamics of the Law of Small Numbers, dividing the attention space
among factions who get their topics and their energies from mutual points
of contention. Modern philosophy, too, continues to operate as it always
did, through the clash of factional disagreement. That a large group of phi-
losophers has taken as their topic the methods and epistemology of the
natural sciences has not meant that philosophy itself has become a high-
consensus, rapid-discovery science; the focus on science has given these phi-
losophers a particular content but not a new social structure for their own
field. One might say that the nonconsensus fields are in the normal con-
dition of the intellectual world; only the rapid-discovery sciences have be-
come deviant.
The modern natural sciences are unusual in their high consensus.
Structurally, this means that they have found a mode of organization that
evades the Law of Small Numbers, the focus of intellectual life upon rivalries
and disagreements. How this has happened will become clearer when we
examine the other key feature of modern science -rapid discovery.
In a field with a fast-moving research front, there is a chain of new
results following one another with such regularity that innovation becomes
Social Sciences Won't Become Iligh-Consensus, Rapid-Discovery 159

expectable. This phenomenon became socially recognized in several fields

by the mid-1600s. The speed of this research front has apparently not been
constant. Data from the 20th century based on the obsolescence of the
research literature suggests a turnover of attention in some fields of as little
as five years or less.3 But even in physics this extremely rapid turnover may
have set in only in the 1920s (Griffith, 1988) and we lack good comparative
studies of just how fast the foci of research changed in various fields (e.g.,
astronomy, biology, chemistry) during the past four centuries. Generally we
can say that the rapidness of discovery became noticeably distinctive in
some natural sciences in the 1600s, and that the rapid-discovery mode has
speeded up since that time and spread to many scientific fields.
In contrast, the normal mode of intellectual life has not been char-
acterized by rapid discovery. The natural sciences before the rapid discovery
revolution, both in Europe and elsewhere, tended to maintain the same
theories and bodies of recorded facts over long periods of time.4 And in
modern Europe as elsewhere, philosophy and the humanistic disciplines
have not had a fast-moving research forefront. This is not to say there are
never any changes in topics and conceptions in these fields; also there is
the propensity to return repeatedly to the same issues -as in philosophy's
classic epistemological questions, reinterpretations of the same authors by
literary critics, or historians' repeated explanations of the same historical
events-whereas the rapid-discovery sciences move on to find new topics
about which to make their discoveries. The nonrapid discovery disciplines
alone have revivals (bringing back Aristotle or Marx or Nietzsche, experi-
encing neo-Kantian or neo-Hegelian movements, reviving the reputation of
past literary figures). Another facet is the study of the classic text, writing
commentaries on Confucius, Hobbes, or Max Weber, whereas the re-
searcher on the forefront of the rapid discovery science does not study
Galileo or Newton, but leaves this to the humanistic discipline of historians
of science.
The contrasts that Thomas Kuhn and Derek Price initially drew be-
tween high- and low-consensus disciplines (paradigm and nonparadigm
fields), in the one case, and between fast-moving/high-immediacy and stag-

3Price (1986). Subsequent work (Cole, 1983; Cozzens, 1989; Leydesdorff and Amsterdamska,
1990) suggests citation analysis is not the best way to measure movement of a research front,
since citations are used for a number of different purposes that vary across fields. Differences
between the sciences and humanities continue to be found, especially in consensus (Cole el
al. 1978; Hargens and Hagstrom, 1982). Rapidity of a researclh front may best be judged
qualitatively, in terms of changing content of ideas and focus on neNvdiscoveries.
4We lack comparative studies of the rate of movement in particular periods when sciences
were changing-e.g., in ancient Greek or Chinese astronomy-so we do not know just
how the non-European sciences at their most innovative compared vith the rates of change
in Europe since the 1500s.
160 Collins

nant/classics-oriented research literatures, in the other, now appear overly

simple. There are aspects of dissensus in the rapid-discovery sciences, and
of consensus in the humanities; and there are various forms of innovation
which take place in the nonsciences intermingled with the several aspects
in which their current publication forefront does not have a rapid discovery
character. But the basic intuitions were on the right track; the high-con-
sensus, rapid-discovery mode has made a crucial division in the social or-
ganization of the intellectual world.
Rapid discovery science does not escape the Law of Small Numbers
that structures intellectual life. It too continues to be driven by rivalries
and the energizing focus upon disagreements. But these disagreements are
confined to the research frontier. As Latour (1987) points out, science has
two cognitive aspects, and these correspond to two modes of social organi-
zation. At the research forefront, what Latour calls "science-in-the-mak-
ing," there are disagreements among rival positions that are similar to those
found in the humanistic disciplines. Just as in the pattern of rivalries found
in philosophy, there is evidence of around five competing research groups
in a scientific specialty (Price, 1986:130-133), similar to what would be ex-
pected under the Law of Small Numbers.
Here, too, studies of scientists' argumentative discourse shows them
using rhetorical tactics to elevate their own position and denigrate those
of their opponents, not unlike clashes of humanistic ideologies (Mulkay,
1985). Once a topic is no longer on the research front, it passes into the
area of consensus that Latour calls "science-already-made." Old results are
taken for granted, elevated to the status of objective facts. One form of
dissensus in science that has been widely researched are priority disputes.
These do not reflect the rivalry and disagreement at the research frontier,
but questions of assigning credit for discoveries that are regarded as having
already been achieved. A priority dispute is possible because there is con-
sensus on what counts as an item of knowledge. It is indicative of the dif-
ference in organization that disciplines that are not rapid-discovery sciences
do not usually have priority disputes.
Rapid discovery science does not so much overthrow the Law of Small
Numbers as dynamicize it, by abandoning old controversies in order to get
on to new ones. It is the existence of the rapid discovery research front
that makes consensus possible on old results. When scientists have confi-
dence they have a reliable method of discoveiy, they are attracted by the
greater payoff in moving to a new problem than in continuing to expound
old positions. The research forefront upstages all older controversies in the
struggle for attention. Because the field is moving rapidly, prestige goes to
the group associated with a lineage of innovations, which carries the implicit
promise of being able to produce still further discoveries in the future.
Social Sciences Won't Become Iligh-Conisenisus,Rapid-Discovery 161

Rapid discovery and consensus are part of the same complex; what makes
something regarded as a discovery rather than as a phenomenon subject
to multiple interpretations is that it soon passes into the realm of consensus,
and that depends upon the social motivation to move onward to fresh phe-
nomena.

Inadequate Explanations of Rapid Discovery Science

What is it in the activity of scientists, emerging at a particular time

in history, that gave rise to this new mode of cognitive and social organi-
zation? Let us briefly review several traditional candidate causes of the sci-
entific revolution; these turn out to be inadequate to explain rapid discovery
science.

Empliricism

One explanation is to attribute the scientific revolution to a turning

away from philosophical speculation and textual tradition, and toward the
direct study of nature. This explanation was touted by many scientists them-
selves at the time, in attacking the scholastic and humanistic traditions that
preceded them. But it fails to stand up to historical comparisons. There is
a good deal of empirical observation in the sciences prior to and outside
of Europe in 1600. The careful collection of astronomical observations had
gone on for many centuries in Greece, China, India, and elsewhere, without
giving rise to either consensus or rapid discovery. Medicine, biology, and
mineralogy had long been studied empirically, and collections of naturalistic
observations were incorporated in major works of the philosophical tradi-
tion ranging from Aristotle to Albertus Magnus. By itself, empiricism has
no dynamism that leads either to a consensus on conceptual schemes and
theoretical explanations, or to a rapid-moving research front.

Measurnementand Mathematization

Was the scientific revolution due to greater precision, or more par-

ticularly, to a new emphasis upon measurement, allowing the formulation
of mathematical laws? Again the generalization does not hold. Some dis-
coveries did take this form (Galileo's experiments with the inclined plane
to formulate the law of acceleration of gravity); but much scientific work
at the period of the scientific revolution, and even later, was not very pre-
cise (e.g., experiments with the mercury tube at various altitudes, or the
162 Collins

pioneering air pump experiments); and many areas of discovery did not
involve measurement at all (such as Galileo's best-known discovery, of the
moons of Jupiter by means of the telescope). If the scientific revolution
was not necessarily measurement-oriented and mathematical, conversely
there was a good deal of mathematical science in the period before rapid
discovery science. A telling instance was Greek astronomy. One highly so-
phisticated version, Ptolemy's planetary system, was the classic exemplar of
a nonmoving research front. Moreover, in comparative perspective, mathe-
matics is not intrinsically a rapid discovery discipline or even one necessarily
characterized by high consensus. The ancient Greeks had rival forms of
mathematics, oriented respectively toward geometric proofs and toward
number theory carrying religious significance; in China, the official mathe-
matical textbooks often disagreed with one another, and advanced methods
discovered by particular individuals were often lost in subsequent periods
(Ho, 1985). The introduction of mathematics into science does not guar-
antee a shift to the rapid-discovery mode.

Experimental Method

Positivist philosophers have often epitomized scientific method as the

design of experimental tests, either in the form of direct manipulation of
conditions or comparison of naturally produced variations. But this method
is not the common denominator of rapid discovery scientific research, as
we see by many examples of nonexperimental and noncomparative findings,
ranging from Galileo's telescopic observations to 19th-century physiology.
Nor is the inductive logic of Mill's Canons unique to modern science; it
has been used in the modern social sciences without making them into
rapid-discovery sciences, and it was used on occasion in Greek and Chinese
science. Again we find the explanation is neither necessary nor sufficient.

Rapid Discovery Science is Produced by Genealogies

of Research Technologies

Modern rapid discovery science hinges on the mode in which scientists

became organized around their research technologies. Galileo's discoveries
in the early 1600s were often considered the takeoff of the scientific revo-
lution. His crucial innovation was not so much a set of new ideas as the
practice of adapting or inventing teclhnologies for purposes of research
(Price, 1986:237-253). Galileo tried out various combinations of lenses-
available as eyeglasses since the 1200s -to adapt them to astronomical ob-
servation; he made use of a pendulum for measuring time in calculations
Social Sciences Won't Become Iligh-Consensus, Rapid-Discovery 163

of motion; he adapted inclined planes not for traditional mechanics but as

a way to make observations on the effects of gravity on balls rolling along
them. Galileo's techniques may have had earlier precedents, but now they
spread to constitute a research forefront. His followers invented barometers
and thermometers, and modified the telescope into a different configuration
of lenses, yielding the microscope and opening up further areas for investi-
gation. Other researchers, away from Galileo's personal network, took up
the example; Guericke and Boyle, for instance, recognized that pumps,
which are already available from the technology of mining, could be modi-
fied for scientific experiments with pressure, temperature, and the vacuum.
In the late 1700s, another wave of research techniques was set off by the
invention of the electric battery, giving rise to far-flung research areas in-
cluding the discovery of new chemical elements by electrolysis of fluids, and
by techniques of electromagnetic and subatomic particle research in the fol-
lowing generations. In the same way, the history of modern nuclear physics
hinges on a genealogy of accelerators, each evolved from the last; and the
progress of astronomical discovery has been the production of new phenom-
ena by the successive developments involving larger optical telescopes, their
use in combination with spectroscopy, and the analogous development of
radio telescopes. Instead of regarding these technologies as transparent me-
dia between scientists' brains and the phenomena that are discovered, we
should see the main dynamic in the research technologies.
Genealogies of research technologies can be regarded as the core of
the rapid discovery revolution because they produce the fast-moving re-
search forefront as well as the tendencies to consensus and imputed ob-
jectivity of results. What Galileo and his followers hit upon was not so much
the value of any particular technique, as the practical sense that manipu-
lating the technologies that were already available would result in pre-
viously unobserved phenomena. Such a stream of new observables in turn
gave rich materials for interpretation in scientific theories. What was dis-
covered was a method of discovery; confidence soon built up that tech-
niques could be modified and recombined endlessly, with new discoveries
guaranteed continually along the way. And the research technologies gave
a strong sense of the objectivity of the phenomena, since they were physi-
cally demonstrable. The practical activity of perfecting each technique con-
sisted in modifying it until it would reliably repeat the phenomena at will.
The theory of the phenomenon, and the research technology that produces
the phenomenon, became socially objectified simultaneously, when enough
practical manipulation had been built into the machinery so that its effects
were routinized.
In the early years of the scientific revolution, the practice of making
research equipment capable of repeating results took considerable time.
164 Collins

Boyle's generation of air pumps produced consistent results only after about
15 years (Shapin and Shaffer, 1985:274-276). Even in the late 20th century,
too quick an announcement of findings leads to controversy and embarrass-
ment when phenomena like cold fusion cannot be routinely evoked. In gen-
eral, the speeding up of the rapid-discovery front has probably been due to
the growing backlog of reliable equipment available to be tinkered with.
Research technologies are not necessarily the brain-children of scien-
tific theorists -ideas embodied in physical apparatus. The Kuhnian para-
digm, insofar as it is a cognitive model of the topic under investigation,
gives a misleading sense of what scientists on a research forefront are doing.
The scientists may have quite irrelevant ideas of how their equipment is
operating. What guides them is on a nonintellectual level, a sense of what
kinds of physical manipulations have resulted in interesting phenomena in
the past, and of what sorts of modifications might be tried that will produce
yet further phenomena. This is not to say that scientists' ideas are super-
fluous; scientists can bring their equipment-generated discoveries into the
social community of scientists only by interpreting them in concepts related
to the going currency of intellectual discussion. Research scientists lead a
double life: as intellectuals in the game of argument for theoretical posi-
tions and as possessors of a genealogy of machines.
Just as the intellectual community consists in networks that pass ideas,
problems and social validation along from one person to another, research
technologies comprise a parallel set of networks. One machine gives rise
to another in a genealogical succession: by modifying the past machine, or
by cloning it from another in the same laboratory, or by a kind of sexual
reproduction recombining parts from several existing pieces of equipment.
Human and machine networks develop symbiotically; a machine embodies
the results of the human activity that went into making it work in a par-
ticular way, while these human skills are typically tacit (H. Collins, 1974)
and cannot be conveyed to another person except by hands-on experience
at the machine. There is some evidence that research technologies on the
active cutting edge can only be made to operate successfully by persons
trained on the immediately preceding generation of machines. For instance,
Boyle's vacuum pump diffused to other researchers only through a network
of persons who had used an earlier exemplar (Shapin and Shaffer,
1985:229-230, 281). Sometimes there are bigger breaks in the genealogy
of research equipment, as when one scientist recognizes that something may
be made analogous to the kind of equipment used in another line of re-
search, perhaps even the word of mouth that a particular kind of equipment
has yielded interesting results.
Two implications of this genealogical structure of equipment are im-
portant for the social organization of rapid discovery science. One is that
Social Sciences Won't Become High-Consensus, Rapid-Discovery 165

access to the forefront research technology tends to be socially concentrated;

since new discoveries are produced by machines based on those that had
made the previous discoveries, the new discoverers are likely to be persons
who worked intimately with the previous round of equipment. It is not un-
usual for a person admitted to the scientific network as a laboratory tech-
nician to become a notable discoverer in his or her own right -as Hooke
did by working with Boyle and his equipment, or the instrument maker Watt
who developed a steam engine after working for the chemist Black. Limited
access to forefront research equipment is one of the features that reduces
competition among rival scientific theories and helps bring about consensus.
The other result is that research equipment, once routinized, may be
exported from the laboratory, where they stand as physical reminders of the
scientific theories that are associated with them. The triumphs of applied
science are not necessarily the application of a scientific theory to a layper-
son's problem. They often consist in a practical activity developed in the
laboratory and then reproduced in the field; for instance, the pasteurization
of milk to eradicate disease consisted in transforming the dairy into a tech-
nological routine resembling that of the medical laboratory (Latour, 1988).
We noted above that scientific research equipment may begin by being im-
ported into the laboratory after existing, perhaps for hundreds of years, in
the lay world. After modifications at the hands of scientists, the equipment
may become commercially viable when reintroduced into the lay world. Thus
the research device for producing and detecting electromagnetic waves could
be adopted for lay use as the radio. Once this happens, the research process
is socially legitimized to a high degree: not merely on the level of ideology
(which may wax and wane), but in the taken-for-granted practices of every-
day life. Rapid-discovery science generates a strong sense of its objectivity
and factuality because its research technology produces many allies.
In these ways, genealogies of research technologies play a key role
in every aspect of modern rapid discovery science: making possible rapid
movement along a research front, creating a sense that new discoveries are
always possible, turning attention away from old controversies in order to
get on with still newer discoveries, and promoting consensus and the sense
of objectivity of past results.

DO THE SOCIAL DISCIPLINES HAVE RAPID DISCOVERY

RESEARCH TECHNOLOGIES?

If this is correct, sociology and the other social disciplines have severe
obstacles to being rapid-discovery sciences. The most glaring lack within
social science is a self-generating lineage of research technologies. Let us
166 Collins

briefly follow the preceding argument with a parallel consideration of the

conditions of social research. In the interests of space, I will concentrate
on sociology with occasional glances at the other social disciplines.
The problem of the social disciplines is not their youth. They have
not lacked sizable bodies of practitioners and institutional bases of support
during the past several centuries at least. Yet they have had neither much
consensus -sociology in particular is famous for its battles among oppos-
ing schools of thought -nor a pattern of rapid discovery.
As long as we believed that the problem was in the area of theoretical
approaches or methodological canons, it was possible to believe sociology
might be made scientific. It should be necessary only for enough sociologists
to adopt the right approach and to cease using unscientific methods. In
fact, the latter part of the prescription would be redundant, since in a rapid
discovery science those who clung to methods that did not bring new results
would soon be unstaged by the achievements of those who were moving
ahead.
Comparative evidence does not support the picture that a field be-
comes a rapid-discovery science by adopting particular ideas or methodo-
logical canons. That is not to say that these theories and methods are
worthless; they may yield some good results without bringing about con-
sensus and rapid discovery. It should be noted that if sociology is not a
rapid-discovery science, it does not follow that it is not a science of any
kind; it may well resemble the natural sciences before the rapid-discovery
revolution, something like Chinese astronomy or Greek mathematics.
The usual prescriptions for scientific sociology are much the same as
those that are erroneously credited with causing the scientific revolution.
Empiricism is not a panacea for sociology. Although some areas of
the field are not empirical (or at any rate not very systematically empirical),
increasing our degree of empiricism would have little effect in bringing
about consensus or speeding up the rate of discovery. In sociology, there
has been a good deal of systematic collection of data since the statistical
compilations of Quetelet in the 1830s and the field studies of LePlay in
1850s, accelerating since the establishment of academic sociology depart-
ments from the 1890s onward. By itself, empirical research does not lead
to consensus or rapid discovery; the results are just as likely to be sets of
local descriptions repeated in other times or places, epitomized by the com-
pilations of demographic data that have been made over many decades.
Nor does quantitative precision and mathematization result in rapid-
discovery science. These were not responsible for that effect in the natural
sciences. In the social sciences, there have been areas of considerable sta-
tistical precision and of mathematical treatment without bringing about a
social scientific revolution. Economic theory became extensively mathemati-
Social Sciences Won't Become Iligh-Conseiisus, Rapid-Discovery 167

cized since the marginal utility paradigm was created in the 1870s; on the
empirical side, econometrics dates from the 1920s. Nevertheless, it is not
clear that economics has either a rapid-discovery forefront or a high degree
of consensus, especially in macroeconomics. In many ways economics is a
conservative, classics-oriented discipline, much like sociology, with neo-
Smithian and other revivals and a continuing focus upon paradigms over
100 years old. It is true that large areas of economic theory have been
mathematized to a degree that resembles the natural sciences much more
closely than any other social science. But mathematical theory appears as
an encapsulated specialty within economics; it generates its own lineage of
technical refinements, but these are often regarded as an idealized world
bearing little relation to empirical explanation (Whitley, 1984; Rosenberg,
1992). The mathematization of theory in economics has results paralleling
the theory/data split in sociology, and contributing not to consensus but
dissensus about the overall direction of the discipline.
In sociology, statistical and mathematical approaches have become spe-
cialties within the larger discipline, without transforming the whole into a
rapid-discovery science. Mathematical modeling has taken its place as an-
other branch of theory alongside the great variety of sociological theories.
Although it contains some works of considerable creativity, it does not ap-
pear that mathematical sociology has generated any rapid-discovery lineages,
nor achieved any consensus as to a dominating approach. Sociological sta-
tistics may have come closer to approximating the rapid-discovery mode, but
only among its own specialists. Social statisticians have created a progression
of methods, building upon past forms, without much tendency to revive older
methods or to engage in prolonged rivalries between opposing forms of sta-
tistics. But this characterizes not so much the average sociological researcher,
the consumer of statistical methods, as the much smaller community of social
statisticians who produce these methods. The distribution and historical
changes of research methods actually used by sociologists are not known
with much precision; but it seems clear that the progress of statistical meth-
odology has not brought about a high degree of uniformity in the way so-
ciologists do their research, and more generally has not generated consensus
or a rapid discovery forefront on substantive matters.
If there is something like a rapid discovery science within the community
of social statisticians, this may result from a process similar to the genealogy
of research technologies described above. A given statistical method may be
regarded as a kind of machinery, much in the same way that the algorithmic
techniques of pure mathematics have provided sets of procedures that can
be manipulated to produce results in a new form. A takeoff of rapid discovery
occurred within European mathematics in the period between Cardan and
Tartaglia in the 1530s and Descartes in the 1630s. This development hinged
168 Collins

upon the shift from verbal arguments and abbreviations to the invention of
standardized symbols and the formulation of rules for manipulating systems
of equations. The invention of this mathematical "machinery"set in motion
a series of investigations that paralleled the takeoff in physical science based
on tinkering with research equipment. Successive generations of symbolism
were created and recombined in various ways, with new mathematical results
produced at each step. Mathematics became a rapid discovery science in its
own realm, not as an adjunct to research in natural science; its chain of tech-
niques and discoveries has proceeded by its own dynamics; conversely, the
development of pure mathematics has not centrally determined the process
of empirical discovery in natural science.
Much the same process of development may be seen in the history
of statistical techniques. A genealogy of statistical methods is created by
modifying earlier methods and trying out analogous procedures (as in the
invention of ordinal statistics paralleling those based on interval measure-
ment). There is progress within statistics, taking the form of a succession
of methodological discoveries. The relationship between statistics and its
application in social research, however, remains like that between pure
mathematics and the empirical sciences. By itself, the pure statistical (or
mathematical) side can far outrun what is happening in empirical research,
and the application of the pure technique to the empirical enterprise does
not make the latter a rapid-discovery science. In sociology, it may even
lead to a growing sense of divergence, between the refinement of the tech-
nique and the noncumulative quality of the substantive results.
The experimental method was not responsible for rapid-discovery sci-
ence in the natural sciences, nor has it had that effect in social science. It
is true that the experimental method lends itself to programs of research
exploring a clearly delineated topic, and several such programs in sociology
have accumulated series of results (e.g., Willer's Elementary Theory, 1987,
and the Expectation States program of Wagner and Berger, 1984). Never-
theless, these experimental programs have not generated widespread con-
sensus, but have become several more strands among the many rival
positions within sociology. A broader analogue to the logic of experimental
design has been used elsewhere in sociology, in the form of systematic com-
parison. Historical sociologists have often explicitly used the comparative
method, with cumulative results on macroissues such as the theory of revo-
lution (Moore, 1966; Paige, 1975; Skocpol, 1979; Goldstone, 1991; see
Collins, 1993, for an overview of continuity in this area). Other comparative
work has been done by accumulating available studies and synthesizing their
findings into a larger system of generalizations. Such projects have gone
on in sociology since the beginning of the 20th century. This method was
the basis of Durkheim's project in the Anine'eSociologique and played a
Social Sciences Won't Become Iligh-Coiisenstis, Rapid-Discovery 169

large part in Weber's encyclopedic treatment of economy and society; sub-

sequent instances include Lipset (1960) and Etzioni (1975).
The comparative method is particularly useful for producing explana-
tory theory, since it focuses our attention upon generalizable conditions
and on bringing to light the most central causal processes. These virtues
are also shared by experimental research programs in the narrower sense.
Why then has this work not transformed sociology into a rapid-discovery
science? The paradox, I suggest, is that what can be found in the intellectual
content of sociology clashes with the way the field is socially organized.
Substantively, sociology has accumulated a good deal of knowledge on the
causal and structural conditions explaining the major variations in social
behavior and organization; in this sense sociology possesses a moderate
amount of scientific knowledge. I have argued elsewhere that within soci-
ology can be found fair approximations of the basic principles determining
microinteraction, the major forms of formal organization, the dynamics of
conflicts and social movements, some aspects of stratification, and some
macrosociological patterns, especially in the realm of politics. In Collins
(1975) I held, more optimistically than now, that sociology already had the
basis of a science in the coherent patterns demonstrable by synthesis of
previous researches. But this was an argument based on the cognitive con-
tent of sociology, ignoring its social organization as a discipline.
At the same time, the social organization of the discipline disperses
attention onto a wide variety of theories and researches. Some of these
have no social motivation for scientific consensus and cumulation, while
others, which are consciously oriented toward scientific knowledge, do not
generate a strong enough attraction to sustain a rapid-discovery forefront.
The method of comparative research produces impressive results, but it
does not give off a chain of such researches that cumulate in the way that
research in the physical sciences gives off a research front.
The primary social tendency of intellectuals is to split into contending
positions to fill the attention space, to follow the Law of Small Numbers.
As we have seen, this tendency is overcome only by special circumstances
in the rapid-discovery sciences; it is only because there are greater payoffs
in plunging forward into new discoveries that these scientists forego the
traditional intellectual mode of maximizing attention by making their po-
sitions distinctive from each other. Lacking genealogies of research tech-
nologies, sociology and the other social sciences are condemned to the
endless contention of positions. Along with this goes the ideology that no
scientific knowledge exists in these fields. The reality that is being reflected
upon here is the social structure of the field, rather than a description of
its cognitive contents. It is feasible, on the epistemological level, to apply
canons of generalizability, explanatory power, and coherence, and to extract
170 Collins

from the myriad sociological publications some materials that have a higher
scientific quality than others. On the substantive level, there is no need to
be a relativist. Some of the contents of sociology are scientific. But they
do not become the focus of a social consensus, because they are not linked
into rapid-discovery chains. Lacking that social support, they are merely
one strand among many others.
Without this prize possession of the rapid-discovery sciences, sociology
is unable to produce a stream of discoveries, or the forward-looking structure
of intellectual competition that goes along with them. Social science thereby
also lacks another major support enjoyed by the natural sciences. The prac-
tical applications of the natural sciences typically take the form of exporting
laboratory equipment or its physical products into the lay world. Since social
science does not invent a stream of new equipment, it is cut off from this
form of applicability. When the social sciences offer practical application, it
is in the form of information or advice: we give policymakers descriptions
of a social problem, and sometimes (more common in economics than in
most other social sciences) projections of what should happen if a given
policy is followed. Beyond this, applied social science consists to a large ex-
tent of exhortation - e.g., the study of stratification is ammunition for social
reform movements, and the study of revolution has been an encouragement
to make one. Applied sociology frequently takes the form of intervening on
a particular side in social conflicts, giving an ideological appeal for one in-
terest and thereby alienating the other.
The natural sciences have acquired a deeper basis of legitimacy. Their
ability to spin off physical products from their research equipment has made
the exchange between scientific intellectuals and the lay community appear
more even-sided. In return for support, laypersons receive material benefits,
and often these take the form of equipment that can be individually owned
and used, giving huge numbers of persons a visible stake in the scientific
research enterprise. Television sets and electric lights are personal remind-
ers of the payoffs of natural science, whereas the benefits of social science
are typically collective, as well as remote and hypothetical. The concrete
practicality of the natural science can be appreciated by laypersons without
understanding the intellectual processes that produced it; in contrast, social
science applications, even relatively popular ones such as psychotherapy,
are embraced largely by persons who are committed to the underlying in-
tellectual doctrines. Without the results that flow from a genealogy of rapid-
discovery technologies, the social sciences are left with a weaker base of
societal support.5

5Turner and Turner (1990) have called sociology anl "impossible science" because the
fragmentation of research specialties and theoretical positions, in America during the late
Social Sciences Won't Become Ihigh-Consenstis, Rapid-Discovery 171

PROSPECTS FOR INVENTING RAPID-DISCOVERY RESEARCH

TECHNOLOGIES IN THE SOCIAL SCIENCES

Are there any signs that rapid-discovery research techniques for social
science are on the horizon? Let us review how this has occurred in natural
science. At the outset is a technique like Galileo's lenses. Assembled in
one way they produce a telescope, in another way, a microscope; combined
with mirrors and prisms, they give rise to the analysis of light spectra. This,
in turn, combined with techniques of the chemical laboratory for heating
and separating materials, results in spectral analysis of the elements. Re-
combined with the telescope, the result is the study of stellar spectra, and
so on, apparently endlessly. The key characteristics are a machinery that
produces some new observable phenomenon (thus giving material for theo-
rizing), plus the capacity that by physically tinkering with the machinery,
modifying or recombining its parts or applying it to new uses, it produces
still further phenomena for study. It is a tinkerable research technology
that gives rise to a stream of discoveries.
Do the social sciences have anything like this? We are looking for
something quite specific, not merely general resemblance to the forms of
presentation found in scientific papers. The use of statistics in social re-
search, as I have argued above, is not a tinkerable research technology in
this sense. It is part of the theoretical manipulation of the data, not a
method of producing new data. In general, sociology's repertoire of tech-
niques for data gathering has been quite slow moving. The basic empirical
methods -field observation, questionnaires, experiments, and historical
analysis-have not changed much over the past century. What has
changed, in some cases, has been the kind of data that has been collected
by such methods (e.g., questions about network ties) and the methods by
which data are analyzed (e.g., the changing fashions for statistical analysis
of questionnaire data). Occasionally we have turned up new phenomena
(e.g., the microdetail pioneered by Goffman and others), but there has been
nothing like a stream of rapid discoveries driven by innovations in research
methods. Our research techniques are not tinkerable and recombinable in
the way that those of the natural sciences are. Ultimately we rest upon
native human skills of making observations, and our discoveries are labo-
riously driven by theoretical acumen in where to look for data and in how
to package it conceptually. What we lack is the key possession of the rapid-

twentieth century, is reinforced by decentralization of organizational resource bases. But this

is an inalterable situation only to the extent that sociology does not develop streams of rapid
discovery technologies.
172 Collins

discovery sciences, their nonhuman machinery for generating previously un-

observable phenomena. They have acquired a mechanical device that pro-
duces a rich stream of novel data as rapidly as the machinery can be
manipulated.
The outstanding example of new hardware in social research was the
introduction of the computer. But this has been used since the 1950s, with-
out major effects on the rapidity of the research front. The computer does
not produce data, but assists in analyzing it. Developments of computer
hardware are extrinsic to social research, and for the most part have little
bearing on the phenomena we study or our theoretical consensus in un-
derstanding them. On the other hand, social scientists have made some
contributions in creating software -for instance, algorithms for social net-
works or simulation models for sociological theories. But such activity is
essentially theory driven; not unexpectedly, it has thus far not given rise to
rapid discovery.
There are some possibilities that social research hardware may move
in this direction. In microsociology during the past 30 years, the introduc-
tion of audio and video recording has opened up the field of conversation
analysis; new phenomena have become visible, such as the fine-grained co-
ordination of persons in social interaction (Sacks et al., 1974; Grimshaw,
1990; Clayman, 1993). Is it possible that such research technologies can
turn into a self-sustaining genealogy of new technologies producing a
stream of discoveries? There are some signs that a stream may be emerging.
For example, by combining voice recording with a Fast Fourier Transform
analyzer, a device that distinguishes the component frequencies of a sound,
new social phenomena have been produced (Gregory, 1983; Gregory et al.,
1993). This equipment has brought the discovery of a low-frequency acous-
tic region where conversationalists' voices converge during an interaction,
thus giving a nonverbal measurement of social solidarity. An offshoot of
this technique gives an instrumentation for who is deferring to whom in
an interaction. So far the pace of innovation from one technology of mi-
croconversational research to another has not been particularly rapid, but
perhaps the field is in the early phases of an acceleration that will become
more noticeable in the future. We lack comparison studies on the pace of
technological innovation in research equipment in the early phases of the
natural science revolution; for instance, how long did it take before
Galileo's techniques, or those of the early air-pump experiments, were ex-
tended to new discovery-making techniques? It is possible that in sociology
such technologies as those involved in microconversational research will en-
courage analogous developments outside the realm of conversation and mi-
crointeraction that will lead to a wider stream of discoveries.
Social Sciences Won't Become Ilig,h-Conseiisus, Rapid-Discovery 173

Another possibility exists in the field of artificial intelligence (Al).

The software of Al is an application of theory, and hence would not be
expected to be the source of rapid discovery. What raises a stronger pos-
sibility is that a series of hardware innovations may emerge that is directly
connected to sociological research. Elsewhere (Collins, 1992) I have sug-
gested that an Al that has human-like capacities can be built if it has the
basic competencies of social interaction. The fundamental argument com-
bines theoretical points from Mead, Durkheim, and conversation analysis.
Human thinking is interiorized conversation; and verbal conversation op-
erates on two levels: a cognitive level of symbols which are acquired in
interactions and represent group membership; and a behavioral level of
emotion-laden rituals, which establish membership. A sociological Al can
be built by modeling these processes. An early prototype could be an "in-
fant Al," which gradually builds up symbols through its history of social
interactions. An important requirement is the capacity for emotional at-
tunement with other persons. Along with software design, this requires
building an Al robot with voice sensors (or sensors for other emotional
signals) and also equipment that can communicate emotional gestures to
others. Such a piece of sociological research hardware would be subject to
tinkering, and thus to the production of new phenomena. These would not
be limited to the nonverbal properties of microinteraction; for instance, it
could become possible to simulate and experiment with various interac-
tional situations and forms of social organization. In this scenario, a soci-
ology-built robot could give rise to a stream of research techniques and to
a rapid discovery forefront.
It is beyond the scope of this paper to survey the research technolo-
gies of all the social sciences. In general, sociology, political science, and
anthropology are methodologically in the same boat. Economics has been
discussed above in terms that suggest it too lacks the research technology
of rapid-discovery science. An exception is the research technologies of ar-
cheology and physical anthropology, which draw on those of the natural
sciences in dating artefacts or establishing genetic links. There are indica-
tions that the use of these techniques is splitting anthropology into non-
communicating factions (Morell, 1993). Whether they are giving rise to
rapid discovery science in the hardware-using branches is not clear.
A brief comment on psychology is in order. Psychology became a
laboratory science in the 1870s. But it has remained a typical social science
in key respects: long-standing dissensus among many competing theoretical
programs (both in the period of introspectionism and during the heyday
of behaviorism) and the lack of a rapid discovery front. In recent decades,
part of psychology has closely allied with methods of biological research,
especially genetics and neurophysiology. The result has been a rift with the
174 Collins

cognitive, behavioral, and social-motivational sides of psychology. It remains

for the sociology of science to investigate whether any of these develop-
ments shows signs of bringing a shift to the high-consensus, rapid-discovery
mode. During the time when the battle goes on among biologically ori-
ented, computational, and more traditional methods, the result is lower
rather than higher consensus. Whether one of these areas is becoming a
rapid-discovery science, and what effect that will have on other parts of
the field, needs investigation. One possibility is that the rapid-discovery sec-
tor will drive the others out of existence through the competition for at-
tention and funding. Another is that the discipline may permanently split
into separate enterprises.

Scenarios of Future Social Science

None of the above scenarios for sociology is certain, or perhaps even

likely. The possibility cannot be ruled out that sociology will acquire a self-
sustaining stream of research technologies in the future and thereby move
into the rapid-discovery mode. If so, this will happen in the midst of many
other kinds of sociology that continue to be pursued by traditional methods.
What then will happen to all these other data-gathering techniques - his-
torical comparisons, field observations, interviews -and to other modes of
sociological discourse? Many of these would be unaffected by a research
technology, say, for measuring nonverbal interaction. Nevertheless, we
should bear in mind that the rapid-discovery sciences have developed not
as individual technologies but as streams of research technologies. Earlier
techniques give rise to new ones, sometimes by application to remote topics,
sometimes by the creation of new research technologies by analogy or re-
combination from those existing in another area. In this case, one can imag-
ine that sociology, with a first few pieces of research hardware, would
resemble natural science in the early 1600s. There would be hundreds of
years ahead of us during which research technologies would penetrate and
transform the older branches of social research; and since a genealogy of
research technologies produces new phenomena to be studied, we could
expect that many new branches of social science would be created.
It may be that some areas of social science simply cannot be tech-
nologized, and hence cannot be turned into the rapid-discovery mode. It
is hard to envision, for instance, how historical sociology would be turned
into a hardware-driven research field. And doubtless huge areas of
metatheory and ideological controversy will remain. Social scientists, deal-
ing intimately with human interests and conflicts, are intrinsically more ac-
tive in promoting ideological disputes than are natural scientists, and this
Social Sciences Won't Become Iligh-Consensus, Rapid-Discovery 175

difference will remain no matter what happens to the conditions determin-

ing the rapidity of the research front. At most, one can expect that social
sciences could become hybrids: part in the rapid-discovery, high-consensus
mode, other parts much as they are today.
Perhaps the more likely scenario is that the social sciences, except
perhaps in isolated pockets, will not develop streams of research technolo-
gies, and will never acquire the rapid-discovery mode. On the balance sheet
of pluses and minuses, this may not be altogether a bad thing. What we
forego is social prestige and material support, collective pride in what our
field has accomplished, and the excitement of intellectual adventure that
goes along with rapid discovery. What we preserve is the independence
that goes with having a great variety of methods; for some scholars, the
pleasures of pursuing humanistic scholarship, and the freedom to teach and
theorize over the whole corpus of our disciplinary history rather than ob-
serving a consensus that tracks past movement of the research forefront.
There are social advantages of living in an intellectual community with low
consensus and slow movement. Oddly, despite the self-images of many so-
ciologists today, this makes our field a conservative and traditionalistic one
-in terms of its own history -whatever political position sociologists may
hold in regard to the larger society. Precisely what we do not like about
the rapid-discovery sciences is their incessant progressivism, their trampling
on all other values in their march to new discoveries.
Ultimately, these value questions are moot. Whether the social sci-
ences become rapid-discovery sciences or not depends upon whether they
acquire streams of research technologies. And that contingency does not
depend primarily upon whether we believe it ought to happen.

REFERENCES

Clayman, Stephen E. Collinis, Ranidall

1993 "Booing: The anatomy of a disaffilia- 1975 Conflict Sociology: Toward an Ex-
tive response." American Sociological planatory Science. New York: Aca-
Review 58:110-130. demic Press.
Cole, Stephen 1987 "A mnicro-macrotheory of creativity in
1983 "The hierarchy of the sciences." Ameri- intellectual careers: The case of Ger-
can Journal of Sociology 89:111-39. man idealist philosophy." Sociological
Cole, S., J. R. Cole, and J. Dietrich Theory 5:47-69.
1978 "Measuring the cognitive state of a 1989 "Toward a theory of intellectual
scientific discipline." In Y. Elkana et change: The social causes of philoso-
al. (eds.), Toward a Metric of Science. phies." Science, Technology and Hu-
New York: Wiley. man Values 14:107-140.
Collins, Hlarry M. 1992 "Can sociology create an artificial in-
1974 "The TEA set: Tacit knowledge and telligence?" In Randall Collins, Socio-
scientific networks." Science Studies logical Insight, 2nd edition. New York:
4:165-186. Oxford University Press.
176 Collins

1993 "Maturation of the state-centered the- Leydesdorff, Loet and Olga Amsterdamska
ory of revolution and ideology." Socio- 1990 "Dimensions of citation analysis." Sci-
logical Theory 11(March):117-128. ence, Technology and Human Values
Cozzens, Susan E. 15:305-335.
1989 "What do citations count? The rheto- Lipset, Seymour Martin
ric-first model." Scientometrics 1960 Political Man. New York: Doubleday.
15:437-447. Moore, Barrington, Jr.
Dictionary of Scientific Biography 1966 Social Origins of Dictatorship and De-
1981 New York: Scribner's. mocracy. Boston: Beacon Press.
Etzioni, Amitai Morell, Virginiia
1975 A Comparative Analysis of Complex 1993 "Anthropology: Nature-culture battle-
Organizations. New York: Free Press. ground." Science 261(24 Septem-
Frede, Michael ber):1798-1802.
1987 Essays in Ancient Philosophy. Cam- Mtilkay, Michael
bridge: Cambridge University Press. 1985 The Word and the World. London:
Griffi'th, Belver C. Allen and Unwin.
1988 "Derek Price's puzzles: Numerical Needhani, Jose)h
metaphors for the operation of sci- 1959 Science and Civilization in China, Vol.
ence." Science, Technology and Hu- 3. Cambridge: Cambridge University
man Values 13:351-360. Press.
Goldstone, Jack A. Netigebatier, 0.
1991 Revolution and Rebellion in the Early 1957 The Exact Sciences in Antiquity. New
Modern World. Berkeley: University York: Dover.
of California Press.
Paige, JefTery
Gregory, Stanford 1975 Agrarian Revolution. New York: Free
1983 "A quantitative analysis of temporal
Press.
symmetry in microsocial relations."
Plrice, Derek J. de Solla
American Sociological Review 48:129-
135. 1986 Little Science, Big Science, and Be-
yond. (1961*). New York: Columbia
Gregory, Stanford, Stephen Webster, and
Gang Iluang University Press.
1993 "Voice pitch and amplitude conver- Roseinberg, Alexander
gence as a metric of quality in dyadic 1992 Economics: Mathematical Politics or
interviews." Language and Conimuni- Science of Diminishing Returns? Chi-
cation 13:195-217. cago: University of Chicago Press.
Grimshaw, Allen D., ed. Sacks, Ilarvey, Emanuel A. Schegloff, and
1990 Conflict Talk. New York: Cambridge Gail JefTerson
University Press. 1974 "A simplest systematics for the organi-
Hargens, Lowell and Warren 0. Ilagstroiii zation of turn-taking in conversation."
1982 "Scientific consensus and academic Language 50:696-735.
status attainment patterns." Sociology Sitapiii, Steveii and Simon Schaffer
of Education 40:24-38. 1985 Lcvialhan and the Air-Pump. Hobbes,
Ho Peng Yoke Boyle, and the Experimental Life.
1985 Li, Qi, and Shu. An Introduction to Princeton, NJ: Princeton University
Science and Civilization in China. Pr-ess.
Hong Kong: Hong Kong University Siviii, Natlhan
Press. 1969 Cosmos and Computation in Early
Jones, Alexander Chinese Mathematical Astronomy.
1991 "The adaptation of Babylonian metlh- Leiden: Brill.
ods in Greek numerical astronomy." Skocpol, Theda
Isis 82:441-453. 1979 States and Social Revolutions. New
Latour, Bruno York: Cambridge University Press.
1987 Science in Action. Cambridge, MA: Ttirner, Stephen P. and Jonathan R. Turner
Harvard University Press. 1990 The Impossible Science. An Institu-
1988 The Pasteurization of France. Cam- tional Analysis of American Sociology.
bridge, MA: Harvard University Press. Ncwbury Park, CA.: Sage.
Social Sciences Won't Become IligIi-Coiiseinstis, Rapid-Discovery 177

Wagner, David G. and Joseph Berger Whitley, Richard

1984 "Do sociological theories grow?" 1984 The Intellectual and Social Organiza-
American Journal of Sociology 90:697- tion of the Sciences. Oxford: Claren-
728. don Press.
Welch, Holmes and Anna Seidel Willer, David
1979 Facets of Taoism. Essays in Chinese 1987 Theory and the Experimental Investi-
Religion. New Haven, CT: Yale Uni- gation of Social Structures. New York:
versity Press. Gordon & Breach.