Research

Publication Date: 22 December 2010 ID Number: G00209754

From Hierarchy to Panarchy: Hybrid Thinking's Resilient

Network of Renewal
Nicholas Gall

As enterprises become increasingly hyperconnected and as the business environment

becomes increasingly volatile and uncertain, the inadequacy of IT-centric, statically
layered models used by mainstream enterprise architecture becomes ever more obvious
and increasingly dangerous. Network-centric and biologically based dynamic models are
needed in order to design solutions, enterprises and industries that are more resilient in
the face of transformative change, especially unforeseen transformations. Such models
include "panarchy" from ecological science, "hyperconnected networks" from network
science and "relationship coordination" from organizational science. These emerging
models will be integrated using hybrid thinking to complement enterprise architecture
with a new paradigm of renewal known as "panarchitecture."

Key Findings
Periodic disruptions in many of today's solutions, enterprises, and industries are
recurring with greater frequency and greater disruption in an increasingly
hyperconnected and volatile world. The days in which people could cross their fingers
and hope that things would hold together until they had moved on are rapidly waning.

Panarchitecture is a process whose goal is to improve enterprise resilience in volatile,

uncertain, complex and ambiguous environments.

The emerging reality is that more and more domains — whether business domains or IT
domains — look like, and more importantly, behave like, highly dynamic complex
adaptive networks.

When you connect two differently modeled networks together, you need a third network
model to understand the resilience of the resulting network of networks.

The essence of the paradigm shift from enterprise architecture to panarchitecture is the
shift from managing the initial design and building of a robust system to managing the
successive designs and continual renewal of a resilient system.

Recommendations
For hyperconnected enterprises in a volatile world, the primary goal can no longer be to
simply protect and sustain the status quo for as long as possible. The primary goal must

© 2010 Gartner, Inc. and/or its affiliates. All rights reserved. Gartner is a registered trademark of Gartner, Inc. or its
affiliates. This publication may not be reproduced or distributed in any form without Gartner's prior written permission. The
information contained in this publication has been obtained from sources believed to be reliable. Gartner disclaims all
warranties as to the accuracy, completeness or adequacy of such information and shall have no liability for errors,
omissions or inadequacies in such information. This publication consists of the opinions of Gartner's research organization
and should not be construed as statements of fact. The opinions expressed herein are subject to change without notice.
Although Gartner research may include a discussion of related legal issues, Gartner does not provide legal advice or
services and its research should not be construed or used as such. Gartner is a public company, and its shareholders may
include firms and funds that have financial interests in entities covered in Gartner research. Gartner's Board of Directors
may include senior managers of these firms or funds. Gartner research is produced independently by its research
organization without input or influence from these firms, funds or their managers. For further information on the
independence and integrity of Gartner research, see "Guiding Principles on Independence and Objectivity" on its website,
http://www.gartner.com/technology/about/ombudsman/omb_guide2.jsp
 shift to understanding cycles of minor and major disruption and then "panarchitecting"
how best to detect, respond to and, ultimately, renew them.

Design dynamic hyperconnected networks, not simplistic static stacks.

Fail forward fast: Shift thinking from designing out failures upfront to designing in ways
to detect and respond to failures out back.

Panarchitecture teams should create human-centered transformative experiences for

evolving away from brittle, engineered hierarchies through perennial renewal.

Panarchitects must continually remind themselves that the essential elements of

resilience are people and their relationships, especially their respect for one another, not
technology.

Publication Date: 22 December 2010/ID Number: G00209754 Page 2 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
TABLE OF CONTENTS

Analysis ....................................................................................................................................... 4
1.0 Introduction ................................................................................................................ 4
2.0 Panarchitecture: Designing Resilience ....................................................................... 5
3.0 Network-Centric Models ............................................................................................. 6
3.1 Models of Interdependent Cycles From Ecological Science............................ 7
3.1.1 The S-Curve of Sustaining Innovation: Mind the Gap! ..................... 7
3.1.2 Modeling Complex Adaptive Systems as Adaptive Cycles of
Renewal ................................................................................................. 8
3.1.3 Focusing Panarchitecture on Pandemonium: The Inverse S-Curve
of Renewal ........................................................................................... 11
3.1.4 Panarchy — Resilient Network of Renewal ................................... 11
3.1.5 Examples of Panarchy in Natural and Human-Created Systems ... 13
3.2 Models of Interdependent Networks From Network Science ......................... 14
3.3 Models of Interdependent People From Organizational Science................... 16
4.0 Practices .................................................................................................................. 17
4.1 Pay Down Debt as You Go by Dynamically Refactoring Dependencies ........ 18
4.2 Plan for Design Succession ......................................................................... 19
4.3 Focus Relational Coordination on Transitions and Transformations ............. 20
5.0 Outcomes and Examples of Emerging Panarchitecture ............................................ 21
6.0 Definition and Conclusion ......................................................................................... 23
Recommended Reading ............................................................................................................. 24

LIST OF TABLES

Table 1. The Great Recession: Cascading Revolts Across Financial Cycles ............................... 12

LIST OF FIGURES

Figure 1. The S-Curve .................................................................................................................. 8

Figure 2. The Adaptive Cycle of Renewal ..................................................................................... 9
Figure 3. The Adaptive Cycle of Renewal Phases ....................................................................... 10
Figure 4. A Panarchy.................................................................................................................. 12
Figure 5. Examples of Panarchies .............................................................................................. 13
Figure 6. Network Versus Layer Models ..................................................................................... 15
Figure 7. Panarchitecture Applied to Buildings ............................................................................ 23

Publication Date: 22 December 2010/ID Number: G00209754 Page 3 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
ANALYSIS
This document introduces a new line of research for a paradigm called "panarchitecture" — an
emerging hybrid-thinking-based paradigm for designing enterprises as biodynamically
coordinated complex adaptive systems. Gartner considers this new line of research to be cutting
edge. Thus, it is intended only for leaders who are seeking what is emerging, not what is
established. It is itself an example of hybrid thinking: combining three disciplines by focusing them
on human-centered experiences to complement enterprise architecture. As with any new line of
research, but especially hybrid thinking research, it raises more questions than it answers. It is by
hybrid thinkers for hybrid thinkers — those who can embrace the ambiguous and incomplete, and
clarify it and extend it while applying it.

1.0 Introduction
"May you live in interesting times." — proverbial Chinese curse
For an increasing number of hyperconnected enterprises and industries, the early 21st century is
becoming increasingly volatile, uncertain, complex and ambiguous (VUCA). 1 From WikiLeaks to
oil leaks, from death of old media to death of privacy and from economic collapse to political
turmoil, it's shaping up to be a bumpy ride. In such a VUCA world, the need for increased
resilience becomes a strategic priority for hyperconnected enterprises. Such enterprises, and the
markets in which they operate, are becoming so dynamically complex and globally
interdependent that they behave more like fragile biological ecosystems than robust engineered
technological stacks. In such a world, small disturbances can easily trigger cascading
disturbances (for example, network outages, supply chain disruptions and economic collapses).
To create resilient solutions, enterprises and even industries in such a hyperconnected and
VUCA world, Gartner believes a new paradigm will be needed to complement simplistic models
based on a static hierarchy of layers. Such a paradigm must embrace and adapt models from
ecological science, network science and organizational science to create a new paradigm for
resilience. This paradigm will:

See volatility as not only a wicked problem, 2 but also a wicked opportunity for
transformative renewal

Recognize that resilience is not a static property, but a biodynamic one, which ebbs and
flows as networks interact, disrupt and reorganize in interdependent cycles of sustaining
growth and transformative renewal3
This paradigm for strengthening resilience by embracing renewal is what Gartner calls
"panarchitecture."4 It is intended to complement the conventional enterprise architecture
paradigm, not replace it. This complementary thinking is already under way in a number of
industries — such as pharmaceuticals, financial services and industrial supply chains — where
leading-edge thinkers are recognizing that their organizational dynamics increasingly resemble
5
those of complex adaptive systems in the natural world.
Embracing the ecologically inspired concept of sustainable and resilient design is foundational to
Gartner's concept of hybrid thinking. In "Introducing Hybrid Thinking for Transformation,
Innovation and Strategy,"6 Gartner introduced the view of hybrid thinking outcomes as balances
of three sets of concerns: meaningfulness, feasibility and sustainability.
Panarchitecture focuses on the sustainability aspect of that triad — specifically, on how to design
resilience to achieve sustainability in a hyperconnected and VUCA world — that is, a world that
increasingly behaves like an ecosystem of dynamically interacting biological networks. The goal

Publication Date: 22 December 2010/ID Number: G00209754 Page 4 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
of panarchitecture is to shift the balance toward good transformations and away from bad ones.
To help hybrid thinking architects embrace this new paradigm for taking on the wicked problems
(and wicked opportunities for transformative renewal), this research first introduces the goal of
ecological resilience and then highlights the three most important aspects of panarchitecture for
achieving that goal: models, practices and outcomes. This research concludes with the "official"
definition of panarchitecture.

2.0 Panarchitecture: Designing Resilience

"If our designs are failing due to the constant rain of changing requirements, it is our designs
that are at fault. We must somehow find a way to make our designs resilient to such
changes and protect them from rotting." — Robert C. Martin
In the face of wicked problems, focusing one's primary efforts on designing out failures upfront
(that is, during the design phase) is a nonstarter, especially wicked problems characterized by
interactions and interdependencies that subtly and slowly change over many years. Such
problems simply cannot be decomposed into fully analyzable and cleanly separated layers (as
years of architectural thinking have urged us to do).
The only way forward is to extend and shift the architectural focus from robust design, which
designs out failures upfront, to resilient design, which designs in better abilities to cope with
cascading failures out back (that is, throughout the operational phase). A simplistic but illustrative
example for business leaders is the difference between designing an indestructible automobile
tire and designing a better spare tire (better in the sense of costing less, taking up less space,
being easier to change and so on).
This shift from robust design to resilient design will require moving beyond the traditional
engineering notion of resilience as merely bouncing back from a disruption to the status quo (for
example, the tire absorbs a nasty bump). We must embrace the emerging ecological notion of
resilience as renewing forward (for example, we can use the spare tire as a "new" tire when the
current one fails).
A more sophisticated example more suited to IT leaders is the difference between designing an
indestructible router and designing a better rerouting algorithm for low-cost routers (better in the
sense of more rapidly and more optimally reconfiguring the network's routing topology in the face
of router failures).
The traditional engineering notion of resilience is focused on the ability to return to normal. It
focuses on maintaining a system in a stable, steady state for as long possible by recovering from
disturbances and returning to normal as quickly and efficiently as possible. This is also a
traditional business view of resilience. 7
The emerging ecological notion of resilience, on the other hand, is defined by ecological theorists
Lance Gunderson and C.S. Holling as "The degree of disruption required to transform a system
from its current balance to a reorganized new balance." (Emphasis added.) It hypothesizes that
what keeps ecosystems sustainable is not just the ability to return to normal after a disruption (for
example, a tree bending in a strong wind), but the ability to renew themselves to create a novel
new normal out of the often broken or damaged pieces left behind after a disruption — for
example, the growth of new ecosystems after a catastrophic event such as a forest fire.
To continue the tire analogy, traditional engineering (or at least traditional business) would see
the spare tire as redundant waste to be eventually optimized out as normal tires become ever
more robust. (Traditional network engineering would see many routers running at low capacity as
suboptimal redundancy.) The traditional ecologist would see such redundancy as a rich source of

Publication Date: 22 December 2010/ID Number: G00209754 Page 5 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
potential novelty and a resource for renewal. For example, a spare tire could potentially help a
stranger whose less-robust tire went flat.
Both types of resilience — recovery resilience (aka robustness) and renewal resilience — are
important, and for simpler organizations in more stable business environments, a primary focus
on recovery resilience is typically sufficient. However, for increasingly hyperconnected enterprises
in an increasingly VUCA world, a primary focus on recovery resilience is not only insufficient, it is
dangerous. For such organizations, the primary goal can no longer be to simply protect and
sustain the status quo for as long as possible.8
The periodic disruption and even destruction of current business models — and the subsequent
need for reinvention and renewal — are not only natural and inevitable but also rapidly
accelerating. Accordingly, for hyperconnected enterprises in the throes of a VUCA world, the
primary goal must shift to renewal resilience: understanding cycles of minor and major disruption
and then panarchitecting how best to detect, respond to and ultimately renew from them.
To create such resilience, panarchitects must begin to explore ways of creating processes and
structures that better coordinate less traumatic and more creative transitions through periodic
cycles of disruption and renewal. By doing so, the business can reduce the negative impacts and
increase the positive outcomes of disruptions in an increasingly VUCA world.
Given such a wicked problem, panarchitecture applies hybrid thinking to incorporate models and
practices from other disciplines. It combines them with enterprise architectural models and agile
methodology from software development. It evolves renewal resilience by creating better, more
human-centered transformative experiences.

3.0 Network-Centric Models

"All models are wrong, but some are useful." — George E.P. Box
One of the most pervasive, yet increasingly pernicious, models in all realms of architecture, and
especially enterprise architecture, is the model of the static hierarchy of layers. It is popular
because it helps decompose complexity into simpler layers. It is pernicious only when it is used to
model systems that are in reality far more complex than the model can adequately represent —
that is, complex dynamic networks. In such circumstances, a static hierarchy of layers
dangerously misrepresents the real degree of interdependency among layers and the dynamic
tempo of such dependencies. This leads to architectures that are exceptionally brittle in the face
of unforeseen circumstances and unpredictable timing.
Yet, despite these limitations, the eternal dream for many enterprise architects is to transform
their current-state spaghetti diagrams into future-state diagrams containing only a few clean and
simple layers. Architects need to stop dreaming and embrace the complex beauty of the
spaghetti!
The emerging reality is that more and more domains — business domains or IT domains — look
like, and more importantly, behave like, highly dynamic complex adaptive networks. For several
years, our research on middle-out architecture has promoted the principle: "Architect the lines,
not the boxes." Panarchitecture continues this trend with the principle: "Design dynamic,
hyperconnected networks, not simplistic, static stacks."
Panarchitecture represents solutions, enterprises, industries and so forth as biodynamically
coordinated hyperconnected networks by unifying network-centric models from three
complementary disciplines:
1. The panarchy model from ecological science

Publication Date: 22 December 2010/ID Number: G00209754 Page 6 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
 2. The coupled network model from network science
9
3. The relational coordination model from organizational science
The integration of all three disciplines will provide a richer understanding of the myriad dynamic
interdependencies among hyperconnected enterprises and how to cope with them.

3.1 Models of Interdependent Cycles From Ecological Science

One of the richest and most actively applied dynamic models in use today comes from the
domain of ecological science. Recent work in the field has demonstrated the central importance
of understanding and managing an ecosystem as a complex network of interdependent cycles of
widely varying dynamic scales. For example, a forest ecosystem is best viewed as a complex
network of interdependent species, such as trees, bushes, animals, insects and fungus. Such
species range in spatial scale from the microscopic to the macroscopic (for example, kilometer-
wide stands of trees) and in dynamic scale from hours (for example, the life span of a bacterium)
to centuries (for example, the life span of a redwood). Ecologists refer to such a network as a
panarchy. The prefix "pan" connotes both "across" (for example, across many scales and across
many disciplines) and the Greek god Pan, whose character connotes creativity and disruption.
Arguably the most-critical dimension of a complex adaptive system, which is virtually absent from
most enterprise architecture frameworks, is the dimension of time (that is, the temporal or
dynamic dimension).10 While many enterprise architecture frameworks incorporate simplistic life
cycle concepts for various technology components (for example, introduction, mainstream, legacy
and retirement), none of them come close to adequately dealing with the temporal dimension,
especially the dynamic behavior of complex adaptive systems. As dynamic hyperconnected
enterprises increasingly behave more like complex adaptive systems and less like merely
complicated technology stacks, enterprises should embrace the ecological concept of panarchy.
The representation of a panarchy differs from a traditional hierarchy in two fundamental ways.
First, a panarchy is a network, each node of which is not a static thing, but a cyclic process,
called an adaptive cycle of renewal. As the next section discusses, an adaptive cycle of renewal
extends the traditionally linear S-curve into a nonlinear cycle by complementing it with an inverse
curve known as the back loop.
Second, the network relationships among adaptive cycles of renewal (that is, the lines connecting
such cycles) do not represent traditional command and control relationships, but rather the
ecological concepts of revolt and remembrance. As discussed in the following section, the revolt
relationship is responsible for upwardly cascading disruptions from small and fast cycles to large
and slow ones. The remembrance relationship is responsible for downwardly cascading
reorganizing resources.

3.1.1 The S-Curve of Sustaining Innovation: Mind the Gap!

Up until now, the well-understood S-curve of sustaining innovation, made famous by Clayton
Christensen's "The Innovator's Dilemma," has been the dominant model for managing technology
and business cycles. The model highlights, among other things, that over time, complexity slows
the rate of sustaining innovation of a technology to the point that it becomes an unsustainable
legacy. When that happens, an inflection point is reached where a new innovation will trigger a
new curve to succeed the previous one, as market needs change (see Figure 1).

Publication Date: 22 December 2010/ID Number: G00209754 Page 7 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
Figure 1. The S-Curve

Source: Gartner (December 2010)

Traditional advice for managing across such cycles has been focused on strategies for exploiting
one's current curve as rapidly as possible for as long as possible, and then optimally timing the
jump to the next one. Such an approach sounds great in theory. However, in practice, many
enterprises find the transitions between S-curves to be extremely unpredictable, disruptive and, in
many cases, fatal.
This is because the S-curve model lacks any detail about what happens in between these S-
curves — and how to navigate the transition from one to the other. Such transitions are literally a
visual and explanatory gap in the model (between the top of a lower S-curve and the bottom of a
higher S-curve). It's as if the S-curve model was saying, "then magic happens" or "here be
dragons."
An answer to the question of what happens in these gaps, and why, was published by Gunderson
and Holling eight years ago. It has been slowly diffusing across other disciplines ever since.

3.1.2 Modeling Complex Adaptive Systems as Adaptive Cycles of Renewal

Examples of complex adaptive systems include natural ecosystems and disease epidemics in the
natural world, as well as electrical grids and global financial markets in the human-created one.
These systems feature a complex web of interdependencies among diverse, autonomous agents.
These are not static systems with long stretches of preplanned stability; instead, they change and
evolve over time through sequential and complementary cycles of growth, destruction and
renewal.
In their groundbreaking 2002 book, "Panarchy: Understanding Transformations in Human and
Natural Systems," Gunderson and Holling proposed a model called "the adaptive cycle of

Publication Date: 22 December 2010/ID Number: G00209754 Page 8 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
renewal," which they derived from the comparative study of the dynamics of biological and
11
human-created ecosystems (see Figure 2).

Figure 2. The Adaptive Cycle of Renewal

Source: "Panarchy: Understanding Transformations in Human and Natural Systems," by Lance Gunderson and
C.S. Holling

According to the model, all complex adaptive systems go through four perennial phases,
organized into two loops:
1. Front Loop (in green):
Exploitation: After a brief period of finding its bearings, so to speak, a newly
reorganized system rapidly matures through a sustained period of innovative
exploitation enabled by its new structure. This increases both its connectedness (the x-
axis) and its accumulated value (the y-axis of capital). However, such increasing
interdependencies lower its recovery resilience (z-axis).
Conservation: The system enters this phase as its ability to sustain innovation begins
to flatten out. It eventually reaches a critical threshold where its resilience and ability to
generate additional value approaches zero.

Publication Date: 22 December 2010/ID Number: G00209754 Page 9 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
 2. Back Loop (in red-orange):
Release: Some disruption (see the discussion of "revolt" that follows) causes the
system to fragment to varying degrees. This phase is the phase of creative
destruction.12
Reorganization: Some of the fragments of the system that came apart are reorganized,
sometimes in new ways with new or different parts, into a new, immature system.
Gunderson and Holling describe the separate objectives of the front and back loops, as follows:
"The first maximizes production and accumulation; the second maximizes invention and
reassortment … and the success in achieving one tends to set the stage for its opposite." They
add that adaptive cycles of renewal — whether in natural or human-created systems — "embrace
the opposites of growth and stability on the one hand, and change and variety on the other."
Examples of how these goals and phases appear in different types of adaptive cycles of renewal
are shown in Figure 3.

Figure 3. The Adaptive Cycle of Renewal Phases

Adapted from Gunderson and Holling

Source: Gartner (December 2010)

How does this relate to hyperconnected enterprises achieving resilience through renewal? Simply
put, only systems that have this dual-loop nature can ever truly evolve. Companies that don't
have a business goal or requirement for evolvability can stick with traditional planning around S-
curves. They can focus on preserving and refining the current way of doing things as long as
possible. However, if an enterprise seeks to truly evolve, it will need to constantly generate
novelty rather than just refine "old novelties" — so it will need to incorporate both loops of the
adaptive cycle of renewal. Such enterprises will actively seek, model and adapt to patterns of

Publication Date: 22 December 2010/ID Number: G00209754 Page 10 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
change — continually generating new business models based on new insights and new patterns,
13
rather than simply refining old models and old patterns of behavior.
Another planning implication of such adaptive cycles of renewal is that a big design upfront
(BDUF) approach — in which a system is carefully planned with the goal that it will remain static
and require few modifications in the future — is typically unsustainable in a hyperconnected
VUCA world. Instead, an approach that allows for things to be planned in a more flexible and
dynamic fashion is more likely to be successful.

3.1.3 Focusing Panarchitecture on Pandemonium: The Inverse S-Curve of

Renewal
"A crisis is a terrible thing to waste." — Paul Romer
The primary focus of panarchitecture is on the causes of and responses to transitioning to the
back loop:
1. What will cause our enterprise to lose resilience?
2. What might trigger our enterprise to enter the release phase?
3. How can we better coordinate to reduce the destructive aspect of the release phase or
at least better cope with such destruction?
4. How do we more effectively coordinate to increase the creative aspect of the reorganize
phase to generate a new or better enterprise?
As can be seen from the examples in Figure 3, the back loop is often a period of pandemonium
— wild and noisy disorder and confusion. The role of the panarchitect is to design ways of
increasing the creative aspects of back-loop pandemonium while decreasing its destructive
aspects. Hence, the title "panarchitect" is perfectly suited to the role of coping with pandemonium!

3.1.4 Panarchy — Resilient Network of Renewal

As rich as the model of the adaptive cycle of renewal is, the behavior of complex adaptive
systems can't be modeled based on a single instance of the cycle. Gunderson and Holling's
breakthrough was to model complex adaptive systems as hyperconnected networks of many
interacting cycles of diverse speeds and sizes, all influencing one another — that is, a panarchy.
As this network of cycles interacts at different scales of space and time — from small and fast to
large and slow — all of them have cross-scale "ripple effects" on one another (see Figure 4). For
example, when a disruptive event triggers the release phase of a micro cycle system, this can, in
turn, cause an interaction with a meso cycle system that causes the latter to itself enter its
release phase — a phenomenon known as revolt. Such revolts can cause subsequent revolts to
create a cascade of failures.

Publication Date: 22 December 2010/ID Number: G00209754 Page 11 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
Figure 4. A Panarchy

Source: Gartner (December 2010)

An example in the financial realm was the recent cascade of failures that almost brought down
the global financial system, aka the Great Recession (see Table 1).

Table 1. The Great Recession: Cascading Revolts Across Financial Cycles

Scale System Release Phase Revolt

Micro Homeowners Default on home loans Housing prices plummet
Meso Banks Default on reserve requirements Credit freezes and spending
declines
Macro National economies Recession Layoffs, bankruptcies
Mega Global economy Collapse War, civil unrest
Source: Gartner (December 2010)

Conversely, a resilient system safely within the conservation phase of its cycle may provide
ample resources or assistance to a smaller/faster system to enable the latter to enter the
reorganize phase more quickly or easily — a phenomenon known as "remembrance." To
continue the Great Recession example, the U.S. federal government, a system with a bigger and
slower cycle than the U.S. economy, intervened (aka remembrance). It provided resources out of
its reserves such as loans, guarantees, and other assistance to banks and businesses to soften
and speed up their journey through the back loop into a renewed structure. If the government had
itself been on the threshold of release (for example, all reserves depleted), the economic collapse

Publication Date: 22 December 2010/ID Number: G00209754 Page 12 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
could have cascaded into a governmental collapse. Gunderson and Holling's breakthrough was
to model complex adaptive systems as hyperconnected networks of many interacting cycles of
diverse speeds and sizes, all influencing one another — that is, a panarchy.14

3.1.5 Examples of Panarchy in Natural and Human-Created Systems

Figure 5 provides some additional examples of adaptive cycles of renewal at micro, meso and
macro levels. Note that, in addition to varying in size, these cycles vary in speed as well — the
larger, macro cycles are always slower, while the smaller, micro ones are faster.

Figure 5. Examples of Panarchies

Adapted from Gunderson and Holling

Source: Gartner (December 2010)

In the natural world, forest ecology provides a good example of adaptive cycles interacting in a
panarchy. Brush fires are relatively quick, frequent micro cycles that may be contained, or may
have a major impact on the larger cycle of underbrush building up in the forest over time. The
cycle of underbrush buildups are slower and larger meso cycles. At a macro level — and over
much longer periods of time — a major fire that could wipe out the entire forest ecosystem could
be triggered by a brush-destroying fire at the meso level.

Publication Date: 22 December 2010/ID Number: G00209754 Page 13 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
One of the key insights of panarchy is that you cannot understand or manage a complex adaptive
system by focusing on only one scale. Instead, since every scale is connected to other scales
changing at different rates, you must focus on at least three:

Conventional view: The scale under analysis (aka meso scale)

Outside-in view: The scale above (aka macro scale)

Bottom-up view: The scale below (aka micro scale)

For panarchitects facing wicked problems, the implications are clear: To achieve renewal
resilience for their enterprises, they can no longer only understand the dynamic behaviors of their
respective enterprises. Now, they must factor in the dynamics of the larger, macro environment
the enterprise is part of, as well as the dynamics of the smaller loops that compose their
enterprise. This notion of three-level analysis ties in directly into the principles associated with the
middle-out architecture style15 — specifically the outside-in principle: You need to design the
larger ecosystem in which your enterprise operates, because network effects from that (macro)
ecosystem will have a profound impact on your (meso) enterprise.

3.2 Models of Interdependent Networks From Network Science

As discussed in the previous section on panarchy, a hyperconnected network model is a richer
and more accurate representation of the interdependencies among the subsystems that compose
a system than a simplistic layered stack. Recent discoveries in network science make such richer
models even more critical to understanding the behavior of complex adaptive systems and how to
improve their resilience.
Until recently, network scientists assumed that the connections between two or more coupled
networks had little impact on the resilience of the individual networks involved. The resilience of a
network is measured by how many nodes can fail (that is, nodes enter the release phase) before
the network itself fragments into isolated partitions (that is, the aggregate node revolts cause the
network itself to enter its release phase). The assumption was that, under normal circumstances,
coupling two resilient networks via a series of connections between them would result in an
equally resilient network of networks.
If this were generally the case, then vaguely representing two coupled networks as two separate
layers (or even clouds) might be a perfectly appropriate representation for virtually all purposes.
For example, contrast the simplistic model of the electrical power layer and Internet
communications layer in Figure 6(b) to the richer model of the actual interdependencies within
and between the two Italian networks in Figure 6(a). Until recently, the more-detailed network
diagram was primarily of interest only to network engineers responsible for the day-to-day
operations of the respective networks.
Unfortunately, for those accustomed to the simplistic layered model, network scientists
discovered earlier in 2010 that, under an increasingly common set of conditions, coupling two
resilient networks through a series of connections, such as those in Figure 6(a), results in a
fragile network of networks that is prone to catastrophic fragmentation.

Publication Date: 22 December 2010/ID Number: G00209754 Page 14 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
Figure 6. Network Versus Layer Models

Source: (a) "Catastrophic Cascade of Failures in Interdependent Networks," Buldyrev et al., Nature, Vol. 464, 15
April 2010, doi:10.1038/nature08932; (b) Gartner (December 2010)

In the paper from which Figure 6(a) comes, the scientists highlight a real-world example of their
research: the blackout of 28 September 2003, which affected much of Italy. The reason the
blackout was so extensive was that nodes in the power grid failing caused nodes in the Internet to
fail as well — as expected. But because these Internet nodes carried, among other traffic, vital
automated control traffic for the power grid, the failure of the Internet nodes cascaded the
blackout to yet more power nodes, which in turn cascaded to yet more Internet nodes — which
was completely unexpected.16
What these network scientists found in their own domain was exactly what Gunderson and
Holling had predicted with their panarchy model in their domain of ecological science: The model
of the connections between two heterogeneous networks (that is, the connections that enable the
revolt and remembrance relationships to exist) is fundamentally different from the models of the
connections within either of the connected networks and can have a major impact on the
resilience of both of them.
For example, in the case of the Italian blackout, the model of the connections between the power
grid and the Internet is completely different from the model of the power grid connections and the
model of the Internet connections. In other words, when you connect two differently modeled
networks together, you need a third network model to understand the resilience of the resulting
network of networks.
As we saw in the section on panarchy, such coupled networks are not an aberrant or exceptional
phenomenon. The world is increasingly composed of such interdependent business and
infrastructure networks:

Communications networks

Power networks
Transportation networks

Publication Date: 22 December 2010/ID Number: G00209754 Page 15 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
 Supply chain networks
Financial networks

Accounting networks

Social networks
Legal networks
Globally complex organizations, such as multinational enterprises, national governments and
global network industries are increasingly composed of such coupled networks.
It's only going to get worse with the gradual migration to cloud computing. Cloud computing
represents a huge transformation from merely complicated technology stacks (for example, three-
tier platforms) to a hyperconnected world of application clouds, platform clouds, infrastructure
clouds, and so on. The assumption of most enterprises beginning their migration is that
connecting two resilient heterogeneous clouds together will typically result in an equally resilient
cloud of clouds. As the combined models from ecological science and network science
demonstrate, such enterprises couldn't be more wrong.
However, in virtually no cases of such hyperconnected networks are there detailed models of the
connections coupling the multiple networks together. In our ignorance, we use simplistic and
wildly inaccurate analogies based on homogeneous analogs like Lego blocks or fluffy clouds to
represent such diverse hyperconnected systems. Hybrid thinking architects facing wicked
problems must soon trade in their toy analogs and trade up to more realistic biological ones — for
example, comparing the connections between diverse networks to connective tissue.17

3.3 Models of Interdependent People From Organizational Science

Models that help us understand how people interact and coordinate are the most important, yet
most difficult, models required for transformative change. Unlike the IT components of an
enterprise architecture, people have emotions, preferences, desires, and so forth, which make
their behavior far more uncertain and difficult to model.
If dynamic hyperconnected systems, and the connections between them, are better understood
as networks, then so are dynamic groups of people, such as teams, departments, and so forth.
The explosion of interest in social networks should have made that blindingly obvious. What is not
at all obvious is how to better coordinate such social networks for improved performance and
resilience.
While the general models from ecological and network science discussed previously can be used
to model social groups at a high level, the current generality of such models makes them less
immediately useful for organizing high-performance resilient teams, and high-performance
resilient teams are crucial to the successful application of panarchitecture. It is especially such
teams that need a model for improving team performance and resilience. One promising
approach is "relational coordination," a recent theory in organizational science pioneered by Jody
Hoffer Gittell:
"Relational coordination is defined as ― a mutually reinforcing process of interaction
between communication and relationships carried out for the purpose of task integration.
The theory of relational coordination differs from ... other theories by proposing three specific
dimensions of relationships that are needed for effective coordination. While many of the
more recent theories emphasize the importance of shared knowledge or shared
understandings, the theory of relational coordination argues that shared knowledge or
shared understandings are necessary but not sufficient. If effective coordination is to occur,

Publication Date: 22 December 2010/ID Number: G00209754 Page 16 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
 participants must also be connected by relationships of shared goals and mutual respect.
Together these three relational dimensions form the basis for coordinated collective action."
(Emphasis added.)18
Many IT organizations (including IT-centric enterprise architecture groups) highlight almost
exclusively the importance of shared knowledge, since information sharing and knowledge
management are their comfort zones. Shared knowledge alone can be sufficient for adequate
performance in front-loop processes. But as Gittell's research states, shared goals and mutual
respect are essential for "coordinating work that is highly interdependent, uncertain and time-
constrained" — that is, back-loop work.
Improved relational coordination, especially improvements in shared goals and mutual respect,
can have a profound impact, not only on team performance, but also on company performance.
The models that form the basis of relational coordination were derived from actual observation
and measurement of high-performing teams at Southwest Airlines in an attempt to understand
how the "Southwest Airlines Way" helped it consistently outperform other airlines.
One of the key insights from the study of Southwest is that a major factor affecting the widely
varying performance of gate departure teams across airlines was the degree of mutual respect
among team members. Teams from other airlines often showed little respect for one another, as
demonstrated by the derogatory names used by team members to refer to members from
different occupations: agent trash, ramp rats, pillow fluffers and so forth. Such airlines had
significantly poorer on-time departure performance than airlines such as Southwest, which
19
focused on creating mutual respect across different occupations.
To connect this to the recent insights from network science, consider that members of a gate
departure team are also members of diverse occupational networks, with very different levels of
status, such as pilot, flight attendant, cabin cleaner, baggage handler, mechanic and gate agent.
Therefore, all the gate departure teams for a particular airline and all the diverse occupational
networks represented on the team constitute a hyperconnected occupational network.
According to the network science research on coupled networks (aka hyperconnected networks),
the relationships between such occupational networks will be different from the relationships
within those occupations. The difference in mutual respect between team members from different
occupations versus team members within the same occupation is a perfect example of this
phenomenon.
Therefore, the work on relational coordination translates abstract network theory into an important
business insight. Employees aren't just members of your corporate network; they are also
members of their occupational (often professional) networks. Just because some employees may
identify with, respect and share common goals with fellow employees within their occupation (for
example, all accountants in your accounting department understand and respect one another),
don't assume that those same employees respect fellow employees from different occupations.
For example, to what extent do bean counters identify with, respect and share common goals
with used-car salesmen? Such interdisciplinary identity, respect and shared goals must be
created by the business itself, especially for teams operating in the back loop.

4.0 Practices
"There is nothing so practical as a good theory." — Kurt Lewin
While good theoretical models are an essential foundation, without equally good practices for
applying such models, they are useless. For panarchitecture to be of practical use, the three
models discussed previously must be unified by a set of practices that apply them. In this section,
we discuss three important sets of panarchitecture practices for applying the above models.

Publication Date: 22 December 2010/ID Number: G00209754 Page 17 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
4.1 Pay Down Debt as You Go by Dynamically Refactoring Dependencies
Throughout this research, we have been highlighting the importance of understanding the
interdependencies among complex systems. This should come as no surprise because the
essence of architecting is to design the key dependencies that define a system. Until recently,
most architectural practice (for example, building architecture, software architecture and
enterprise architecture) focused only on static dependencies and only at design time.
The classic model of such an approach is the waterfall model of system creation. All the
dependencies are designed at the beginning, and the dependencies are assumed to remain static
throughout the life of the system, because the design is supposedly frozen when the design
phase is complete. The notion of changing the design — that is, changing the dependencies,
throughout the life of the system — would be a contradiction of the very definition of the waterfall
model.
In reality, systems designed using the waterfall model do have design changes during their useful
lives. However, such design changes are usually considered to be due to unintentional errors that
violate the design rules or, at best, intentional hacks (quick and clever changes) that are deemed
necessary as a work-around for limitations in the original design. In the software world, such
changes to a system are often referred to as "bit rot," "software rot" or, sometimes, "design rot."
Such rot is what causes an application's value to flatten and its resilience to decrease as it
ascends the front loop (that is, the S-curve). One can think of the value (called capital in the
panarchy model) of an application as its profitability to the enterprise. An application's profit, like
all profit, is equal to revenue minus cost. In the simple case of a revenue-generating application
— for example, an airline reservation system — the revenue of the application is roughly the
revenue of the business service it supports. The cost of the system is the cost of running it and,
more importantly, from a resilience perspective, changing it.
Therefore, the decline in profit for most applications is not due to their revenue declining over
time. In fact, for successful businesses, an application's revenue usually goes up, driving the S-
curve up. The flattening of profit is caused by the cost going up. The most important factor in
causing the cost to go up is the increasing design rot caused by all the poorly made changes and
uncertain dependencies they caused.
To mix metaphors, one can treat making poorly designed changes as borrowing and treat the
resulting rot (that is, those uncertain dependencies that have crept into the design of the
application) as debt. Ward Cunningham (the father of the wiki) created this metaphor of rot as
debt.
If rot is debt, then what is the interest payment on that debt? It is the added cost of running and
changing the application caused by such rot. In other words, every quick and dirty change
(borrowing) causes the rot (debt) to go up, and this, in turn, causes the cost of running and
changing the application (interest payment) to go up.
To continue the metaphor, unless the debt is paid down, the application will eventually become
unprofitable and go bankrupt — that is, the cost to run and change the application (interest
payment) will eventually be greater than the revenue the application produces. Usually, the debt
is liquidated (by retiring the application) before it goes completely bankrupt.
The only way to prevent (or more accurately postpone) such liquidation is to pay down debt as
you go. Despite the obvious advantages of such a techno-fiscally prudent approach, most
organizations fail to do so. Gartner research estimates that the current amount of combined debt
across Fortune 2000 businesses and large public sector agencies is $500 billion, and that the

Publication Date: 22 December 2010/ID Number: G00209754 Page 18 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
 20
debt is likely to reach $1 trillion by 2015. Clearly, this is no longer a sustainable business model
for application management.
Fortunately, leading-edge businesses are already adopting a "pay down technical debt as you go"
approach to application management. Most of these are Web and software as a service (SaaS)
businesses whose value proposition is based in part on the continuous enhancement of their
customer-facing software-based services with minimal disruptions. To accomplish this, such
organizations establish a disciplined approach to continually paying down technical debt by
continually refactoring their applications to reduce the uncertainty and complexity of the
dependencies in their designs.
The practice of refactoring is fundamental to all agile methodologies. It is essentially a
commitment to regularly clean up code and design by reducing or restructuring dependencies,
even though such clean up is not strictly necessary to keep the application functioning. While
most traditional enterprises have not been able to create such an agile culture, many Web-centric
enterprises have.
For example, since 2007, salesforce.com has employed agile methodologies to deliver major
releases roughly ever quarter. A major factor in its success has been an explicit and far-reaching
dependency management discipline on the part of IT and the business. One of the most
innovative aspects of salesforce.com's dependency management is its Virtual Architecture Team
(VAT).
The VAT is a virtual team in that it is composed entirely of developers from every one of the 30+
Scrum teams at salesforce.com. The VAT is responsible for maintaining and evolving
salesforce.com's SaaS software architecture. It is the VAT that collectively decides on a regular
basis what needs to be cleaned up and refactored. Like most organizations, many product teams
and product owners are reluctant to carry out the requested cleanup work.
To overcome this reluctance, salesforce.com asks business units to commit 20% of their time
every release to implementing the cleanup and refactoring recommended by the VAT. Although
Google's 20% innovation time is famous, salesforce.com's 20% cleanup time is far more
innovative. A good analogy would be a forest products company dedicating 20% of its resources
to periodically maintaining the firebreaks and cleaning up the underbrush throughout the forest,
rather than allowing debt to accumulate over decades until it reached a critical threshold and
caused a catastrophic fire that consumed the entire forest.
Note that salesforce.com does not make such a major commitment to paying down debt merely
to reduce the cost of running its SaaS platform. Run costs alone would be insufficient to justify
such a commitment. The primary reason it regularly refactors its software architecture is to
21
ensure it can maintain its tempo of delivering one release per quarter to its customers.
This sustainably rapid pace of innovation is driving what is effectively a panarchitectural process.
By dynamically renewing its architecture, salesforce.com keeps its interest payments low by
continually paying off its debt.

4.2 Plan for Design Succession

One of the weaknesses of agile methodologies is their insistence on not putting potentially useful
capabilities into the design upfront. This agile principle is captured by the phrase, "You ain't
gonna need it" (YAGNI).
However, even the staunchest "agilists" admit, sometimes grudgingly, that sometimes
stakeholders will eventually need a capability that is not accommodated by the simpler initial
design. For example, a system may initially only need to handle only one transaction per contract,

Publication Date: 22 December 2010/ID Number: G00209754 Page 19 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
but later need to handle multiple transactions per contract — for example, multiple construction
project bids on a single contract.
A leading agilist, Kent Beck, recently introduced the concept of design succession into his
thinking to deal with such cases. The idea is not to design for multiple transactions upfront, but
rather to design with the notion of ecological succession in mind. In ecological succession, a
series of ecosystems succeed one another in a predictable way — for example, from grasslands
to shrubs, softwood forests and hardwood forests. Each preceding ecosystem prepares the way
for its successor:
"Applied to software design, succession refers to creating a design in stages. The first stage
is not where you expect to end up, but it provides value. You have to pay the price to
progress from stage to stage, but if jumping to the final stage is prohibitively expensive in
time or money, or if you don't know enough to design the 'final' stage, or if ... the customers
don't know enough to specify the system that would use the final stage, then succession
provides a disciplined, reasonably efficient way forward." 22
This emergence of panarchitectural thinking among agilists is no surprise. Many of the concepts
underlying agile methodology have a basis in natural systems. The price to be paid for
progressing from architectural stage to architectural stage is of course the kind of design
refactoring discussed above.
A concrete example of such design succession is the Amazon.com site design. The information
architecture has evolved through successive designs of the tab navigation for the site, as the
diversity of product categories has increased over the years. At first, a simple tab design worked
fine. Then, the number of tabs became unwieldy, and specific tabs were, therefore, eliminated.
Finally, a hybrid tab structure was adopted. 23 It is highly unlikely that Amazon.com's UX designers
could have designed the current tab structure right from the start. Amazon.com had to grow into
it.

4.3 Focus Relational Coordination on Transitions and Transformations

Since mainstream enterprise architecture teams typically focus only on front-loop architectural
issues, they can perform adequately using conventional models of group behavior to guide their
organizational interactions. Panarchitecture teams, however, have no such luxury because they
not only design ways to better cope with back loops before they occur, but they also must actively
participate in (and often lead) potentially pandemonious back-loop initiatives as they erupt.
Crisis management teams, hospital teams, P&G's Flow to the Work team and even airline
departure teams all work in predominantly back-loop environments: Such teams seek to transition
and transform a disorganized breakdown phase into a reorganization phase that provides the
basis for renewal. In their own unique ways, they are all panarchitectural teams. Such teams are
most in need of relational coordination's deep emphasis on shared understanding, shared goals,
and mutual respect to enhance their performance and resilience.
A good example of an IT team that leads such back-loop initiatives is P&G's Flow to the Work
team:
Filippo Passerini, P&G's CIO and head of its Global Business Services (GBS, which
includes IT), created a "design shop" culture (that is, intense, multidisciplinary, creative and
time-boxed work) by reinventing the structure and approach of GBS. He outsourced all the
routine aspects of business and IT operations to a network of strategic partners and
transformed what remained into a flow-to-the-work organization. Passerini once explained:
"I want P&G's [GBS] to be delivering an ever-stronger foundation for P&G business growth:
driving business transformation and working as the 'go to' organization for all wicked

Publication Date: 22 December 2010/ID Number: G00209754 Page 20 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
 problems. This means designing ourselves for simplicity so that, together with our partners,
we can 'flow to the work' and respond to emerging priorities."
P&G used its flow-to-the-work teams to successfully integrate its Gillette acquisition in 15
months instead of the usual three to four years. 24
Speeding up a major acquisition is a great example of a planned transition into a back loop, since
typically, the acquired company is broken into pieces (release) and then recombined with pieces
of the acquiring company (reorganization). A team permanently dedicated to taking on wicked
problems in the back loop is a perfect example of a panarchitecture team.
Ultimately, resilience is fractal: Resilience of industry depends on resilience of enterprises, which
depends on resilience of teams, which depends on resilience of people. Hybrid thinkers are more
resilient people. Hybrid teams are more resilient teams, and so on. Accordingly, organizational
resilience is ultimately a personal issue, a team issue and a cultural issue — that is, a human
issue. Panarchitects must continually remind themselves that the essential elements of resilience
are people and their relationships, not technology — for example, a disaster recovery site issue.

5.0 Outcomes and Examples of Emerging Panarchitecture

"Consider all outcomes before taking a step, and spend your life on one leg." —Anonymous
According to Gartner's definition of enterprise architecture, the primary outcome of a successful
enterprise architecture process is the translation of business vision and strategy into effective
enterprise change. Implicit in this description is the notion that enterprise architecture is
fundamentally based on a structured approach of gap analysis and incremental gap closing
between the current state and an expected, and planned for, future state. Such an approach is
perfectly suited to the front loop of the adaptive cycle of renewal.
However, due to its focus on effective enterprise change, enterprise architecture is not designed
to deal with unexpected enterprise change — that is, the sudden, the unforeseen, the
serendipitous, and the disruptive (including catastrophic disruptions) — therefore, the need for
panarchitecture for hyperconnected enterprises in a VUCA world. The primary outcome of a
successful panarchitecture initiative is a solution, enterprise, or industry that is more resilient to
changes that are either difficult or impossible to predict.
When such change occurs, panarchitecture ensures that the back-loop phases of release and
reorganization are less disruptive and more productive. Panarchitecture provides a general
framework for coping with an open-ended range of unforeseeable events.
Panarchitecture can also be of use in the front loop as well. By applying practices such as
repaying debt as you go (that is, refactoring dependencies), performing controlled burns (that is,
small, planned and controlled disruptions) and improving relational coordination, panarchitecture
can help sustain a more robust and productive front loop for a longer period of time. 25
Increasingly, thinkers in various business disciplines that face hyperconnected wicked problems
in a VUCA world are beginning to extend beyond conventional engineering-based thinking to
panarchitectural thinking. The following are just a few examples:

Financial Markets: In 2009, Andrew Haldane, executive director of financial stability for
the Bank of England, authored a paper, "Rethinking the Financial Network." In it,
Haldane proposed that lessons from network disciplines such as ecology, epidemiology
and biology be applied to the financial sphere in the wake of the 2008 global slowdown
in financial markets to better understand the dynamics that caused it. This collapse, he
says, was a manifestation of:

Publication Date: 22 December 2010/ID Number: G00209754 Page 21 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
 "… the behavior under stress of a complex, adaptive network. Complex because these
networks were a cat's cradle of interconnections, financial and nonfinancial. Adaptive
because behavior in these networks was driven by interactions between optimizing, but
confused, agents."
A cat's cradle of financial networks driven by confused agents is a perfect description of the
kind of hyperconnected networks that lack resilience due to their coupling. His paper was
one of the inspirations for this new line of research.
The Pharmaceutical Industry: The new field of pharmacovigilence is using
panarchitecture principles that establish a new approach to tracking complex drug
interactions across different dynamic cycles in a highly complex network of patients.
Drug testers and regulators today face the wicked problem of trying to cope with not only
the increasing number of drugs being used simultaneously, but also the longer time
periods over which they are taken (for example, decades in many cases). How, in such
a VUCA world, can the FDA hope to prevent upfront (that is, before approval) all
possible adverse reactions from such long-term interactions? It can't.
This being the case, the FDA is recognizing that, since it can no longer test for all possible
adverse effects before approval, all drugs are now essentially in a kind of permanent beta (a
perfect term for panarchitectural thinking). Therefore, in exchange for lessening some of the
burden of the preapproval testing processes formerly imposed on drug companies, the FDA,
in return, is imposing more requirements for postapproval pharmacovigilence: keeping a
vigilant eye on drug interactions across the panarchy of patient life cycles and drug life
cycles. It is establishing the Sentinel Initiative — to detect and address adverse drug
interactions as quickly as possible while also monitoring for long-term effects over many
years. Pharmacovigilence represents fail forward fast thinking 26: a major shift in thinking
from designing out failures upfront to designing in ways to detect and respond to failures out
back.27

Supply Chain Planning: At the Center for Resilience at Ohio State University, a group
network of engineers, scientists and business scholars is researching ways to improve
the resilience of industrial supply chains and the environments in which they operate,
based on the panarchy model. This research center is exploring ways in which industrial
supply chains can emulate the attributes of dynamic, adaptive systems that can flourish
and grow in the face of uncertainty and change — and seeking to develop practical tools
for measuring and improving the resilience of supply chain processes. 28

Building Architecture: In "How Buildings Learn," Stewart Brand offers a framework for
how building architects can consider how change can be accommodated gracefully
within their structures, without building in dependencies that would cause shorter-cycle
changes to be disruptive to longer-cycle ones (see Figure 7). "Building architects,"
Brand said, "can mature from being artists of space to become artists of time."
In a related sense, the goal for many enterprise architects will be to mature from being
designers of components to becoming designers of coordination. His model of a building's
shearing layers is a good example of a panarchy of nested cycles. Each of the shearing
layers is potentially an adaptive cycle of renewal. A key question a panarchitect would ask
would be, "What are the various connective tissues between each pair of cycles, and what
29
can I do to improve their resilience?"

Publication Date: 22 December 2010/ID Number: G00209754 Page 22 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
Figure 7. Panarchitecture Applied to Buildings

Source: Stewart Brand's 6 S's from "How Buildings Learn"

6.0 Definition and Conclusion

"Justice will take us millions of intricate moves." — William Stafford, "Thinking For Berky"
With these models, practices and outcomes in mind, Gartner's "official" definition of
"panarchitecture"30 follows:
"Panarchitecture" is a process whose goal is to improve enterprise resilience in volatile,
uncertain, complex and ambiguous environments. It tackles this wicked problem by
coordinating relationships among multidisciplinary teams of hybrid thinkers. These teams
create human-centered transformative experiences for evolving away from brittle,
engineered hierarchies through perennial renewal. Over time, resilience increases as better-
managed, biodynamically coordinated hyperconnected networks operating holistically
across scales emerge.
Giving a definition of "panarchitecture" and explaining it in an introductory note is like defining
"democratic government" and explaining it in a note. This is only the first, vanishingly small step
of a great, centuries-long enterprise and experiment.
Periodic restructurings in many of today's solutions, enterprises and industries are recurring with
greater frequency and greater disruption in an increasingly VUCA world. The days in which we
could cross our fingers and hope that things would hold together until we had moved on are
rapidly waning. Once-in-a-generation or once-in-a-career disruptions are now occurring once a
decade.
Accordingly, hybrid thinking architects and other business leaders will soon need to embrace and
extend panarchitecture to gain a much deeper understanding of the ways in which
hyperconnected networks interact and how they can be made sufficiently resilient, despite such
interactions. It is because of such a need that Gartner is convinced that scientific research into
panarchy, network science and relational coordination holds valuable lessons for hybrid thinkers
tackling wicked problems of transformative renewal.
A critical enterprise goal must be to design for effective preparation for and response to these
inevitable reorganizations — to better enable the enterprise to exploit such events to refactor
things in a way that the enterprise, and its people, emerge stronger and more resilient. Just as
modern forest management no longer tries to stop all forest fires — recognizing that small ones

Publication Date: 22 December 2010/ID Number: G00209754 Page 23 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
are periodically needed to clear out the underbrush — hyperconnected enterprises must allow for
certain amounts of "creative destruction" in return for more resilience in the enterprise.
Architects in hyperconnected enterprises must understand that they cannot design all the risks
out of their processes, organizations and systems in advance. They must begin to focus more of
their efforts on preparing for how to coordinate resilient renewal, which may mean beginning to
focus less on attempting to stabilize, tweak and maintain their current architecture to sustain the
status quo for as long as possible. 31
The essence of the paradigm shift from enterprise architecture to panarchitecture is the shift from
managing the initial design and building of a robust system to managing the successive designs
and continual renewal of a resilient system.

RECOMMENDED READING
"Embrace Hybrid Thinking to Drive Transformation, Innovation and Strategic Change
"Introducing Hybrid Thinking for Transformation, Innovation and Strategy"
"Eight Hybrid Thinking Principles for Enterprise Architects"
"Operational Tempo and Pace Layers: Go Fast When You Must; Be Thorough When You Can"
"Chief Enterprise Architects Must Become Hybrid Thinkers to Take on Wicked Problems"
"How Hybrid Thinking Can Complement Pattern-Based Strategy"
"How to Use Pace Layering to Develop a Modern Application Strategy

Evidence
1
The term "VUCA" originated in military vocabulary and came into more common use in the late
1990s. For more information on VUCA, see
http://en.wikipedia.org/wiki/Volatility,_uncertainty,_complexity_and_ambiguity.
2
Wicked problems are intractable problems that defy conventional approaches to understanding,
planning, design, implementation and execution. For more information on wicked problems, see
"Introducing Hybrid Thinking for Transformation, Innovation and Strategy."
3
In our panarchitecture research, we use the term biodynamic merely as shorthand for the
somewhat awkward phrase "dynamic biological" or the even longer "biologically dynamic."
Biodynamic conveys the important requirement that for models of complex adaptive systems to
be useful, they must be both dynamic (representing the temporal dimension) and biological
(based on some biological model or analogy). Our use of the term does not refer to the concept of
biodynamics as defined by Rudolf Steiner.
4
We introduce the neologism (some might say, jargon) panarchitecture for two reasons: (1) to
highlight the importance of the concept of ecological panarchy presented in this research; and (2)
to highlight and reinforce that the paradigm introduced in this research is not a replacement for
mainstream enterprise architecture; it is a complementary approach with complementary goals.
We realize that many people object to the proliferation of architectural jargon and, therefore, may
dislike the term "panarchitecture." If so, there are a number of perfectly suitable alternatives that
may work better in their context — for example, hybrid architecture, hyperconnected architecture,
rearchitecture, renewal architecture, resilient architecture and organic architecture.
5
See examples from the industries that follow in the Outcomes and Examples of Emerging
Panarchitecture section of this research.

Publication Date: 22 December 2010/ID Number: G00209754 Page 24 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
6
This also appears in the Gartner Special Report, "Embrace Hybrid Thinking to Drive
Transformation, Innovation and Strategic Change."
7
In fact, in 2002, Gartner defined a business as "an organization that rebounds, adjusts quickly
and resumes operations." For more information, see "The Blueprint for the Resilient Virtual
Organization."
8
"When the rate of change outside your organization exceeds that within your organization, the
end is near." — Jack Welch
9
Since this is an introductory document to a new line of research, we cannot fully explore the
evidence for each of the following models. Accordingly, the primary focus of this research will be
the panarchy model. The sections on coupled networks and relational coordination will be less in
depth, with just enough background to integrate all three models into enterprise architecture.
Future research will flesh out all three domains.
10
Throughout this research, we shall prefer the term "dynamic" over the term "temporal" for ease
of readability.
11
Gunderson and Holling simply refer to the cycle as the "adaptive cycle." We add the phrase "of
renewal" to make explicit the central role that renewal and reorganization play in the cycle.
12
"Creative destruction" is a term popularized by Joseph Schumpeter in his 1942 book, "The
Process of Creative Destruction": "The opening up of new markets and the organizational
development from the craft shop and factory to such concerns as U.S. Steel illustrate the process
of industrial mutation that incessantly revolutionizes the economic structure from within,
incessantly destroying the old one, incessantly creating a new one ... [The process] must be seen
in its role in the perennial gale of creative destruction; it cannot be understood on the hypothesis
that there is a perennial lull." — Joseph Schumpeter
13
For more information on how the common denominator of tempo and cycle can connect
panarchitecture and Pattern-Based Strategy, see "The Adaptive Organization: How Boyd's
Decision Cycle and Pattern-Based Strategy Drive Rapid Change." For more information on the
synergies between hybrid thinking and Pattern-Based Strategy, see "How Hybrid Thinking Can
Complement Pattern-Based Strategy."
14
The Resilience Alliance is a focal point for ongoing work in panarchy, the adaptive cycle of
renewal, and their application in diverse domains. For more information see, www.resalliance.org.
15
See "Understanding EA Approaches: Middle Out."
16
"Catastrophic Cascade of Failures in Interdependent Networks," Buldyrev et al., Nature, Vol.
464, 15 April 2010, doi:10.1038/nature08932.
17
For more information on governing such connective tissue, see "How to Use Pace Layering to
Develop a Modern Application Strategy."
18
For more information, see "Relational coordination: Coordinating work through relationships of
shared goals, shared knowledge and mutual respect" by Jody Hoffer Gittell, 2006. In Relational
Perspectives in Organizational Studies: A Research Companion. Edited by O. Kyriakidou and M.
Ozbilgin.
19
For more information, see "The Southwest Airlines Way" by Jody Hoffer Gittell.
20
For more information, see "Measure and Manage Your IT Debt."

Publication Date: 22 December 2010/ID Number: G00209754 Page 25 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
21
"We grow in direct proportion to the amount of chaos we can sustain and dissipate." — Ilya
Prigogine
22
For further discussion, see www.threeriversinstitute.org/FirstOneThenMany.html.
23
For further discussion, see http://andrewchenblog.com/2009/11/25/product-design-debt-versus-
technical-debt.
24
See "Introducing Hybrid Thinking for Transformation, Innovation and Strategy."
25
For other ways in which hybrid thinking can be applied to enterprise architecture in the front
loop, see "Eight Hybrid Thinking Principles for Enterprise Architects."
26
"All of old. Nothing else ever. Ever tried. Ever failed. No matter. Try again. Fail again. Fail
better." — Samuel Beckett
27
For more information, see www.fda.gov/Safety/FDAsSentinelInitiative/default.htm, as well as
the private-sector site, www.iguard.org.
28
For more information, see http://resilience.osu.edu.
29
For more information, see "Operational Tempo and Pace Layers: Go Fast When You Must; Be
Thorough When You Can."
30
We put "official" in quotes to indicate that the definition is not actually official in the sense of
unalterable or mandatory. Such a definition would contradict the very informal nature of
panarchitecture. It is better to think of the definition as being official in the sense of "officially the
first definition of the term."
31
"Some new principle of refreshment is required. The art of progress is to preserve order amid
change, and to preserve change amid order. Life refuses to be embalmed alive. The more
prolonged the halt in some unrelieved system of order, the greater the crash of the dead society."
— Alfred North Whitehead

This research is part of a set of related research pieces. See "Embrace Hybrid Thinking to Drive
Transformation, Innovation and Strategic Change" for an overview.

Publication Date: 22 December 2010/ID Number: G00209754 Page 26 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.
REGIONAL HEADQUARTERS

Corporate Headquarters
56 Top Gallant Road
Stamford, CT 06902-7700
U.S.A.
+1 203 964 0096

European Headquarters
Tamesis
The Glanty
Egham
Surrey, TW20 9AW
UNITED KINGDOM
+44 1784 431611

Asia/Pacific Headquarters
Gartner Australasia Pty. Ltd.
Level 9, 141 Walker Street
North Sydney
New South Wales 2060
AUSTRALIA
+61 2 9459 4600

Japan Headquarters
Gartner Japan Ltd.
Aobadai Hills, 6F
7-7, Aobadai, 4-chome
Meguro-ku, Tokyo 153-0042
JAPAN
+81 3 3481 3670

Latin America Headquarters

Gartner do Brazil
Av. das Nações Unidas, 12551
9° andar—World Trade Center
04578-903—São Paulo SP
BRAZIL
+55 11 3443 1509

Publication Date: 22 December 2010/ID Number: G00209754 Page 27 of 27

© 2010 Gartner, Inc. and/or its Affiliates. All Rights Reserved.