sense-making in a complex and complicated world

8/2/2019 Sense-Making in a Complex and Complicated World

1/22

The new dynamics of

strategy: Sense-makingin a complex and

complicated world

by C. F. Kurtz

D. J. Snowden

In this paper, we challenge the universality ofthree basic assumptions prevalent inorganizational decision support and strategy:assumptions of order, of rational choice, and ofintent. We describe the Cynefin framework, asense-making device we have developed tohelp people make sense of the complexitiesmade visible by the relaxation of theseassumptions. The Cynefin framework isderived from several years of action researchinto the use of narrative and complexitytheory in organizational knowledge exchange,decision-making, strategy, and policy-making.The framework is explained, its conceptualunderpinnings are outlined, and its use ingroup sense-making and discourse isdescribed. Finally, the consequences ofrelaxing the three basic assumptions, usingthe Cynefin framework as a mechanism, areconsidered.

Over the past several years, our group has been con-

ducting a program of disruptive action research us-ing the methods of narrative and complexity theoryto address critical business issues. 1 Action researchhas been defined as grounding theory in contextualexploration, emphasizing participation, and embrac-ing change.

We started work in the areas of knowledge manage-ment, cultural change andcommunity dynamics,thenexpanded into product development, market cre-ation and branding, and in recent years have beenworking increasingly in the area of national and or-ganizational strategy. Some of this work has been

directly funded by the U.S. government throughDARPA (Defense Advanced Research ProjectAgency) as well as by other government agencies (inparticular in Singapore) which are interested in newapproaches to supporting policy-making. The cen-tral element of our approach is the Cynefin frame-work for sense-making. In this paper we describe theframework and its conceptual basis, and we detailsome of itsuses for sense-makingto support decision-making in varied dynamical contexts.

Conceptual approach. We begin by questioning the

universality of three basic assumptions that pervadethe practice and to a lesser degree the theory of de-cision-making and policy formulation in organiza-tions. These are:

The assumption of order: that there are underlyingrelationships between cause and effect in human in-teractions and markets, which are capable of discov-ery and empirical verification. In consequence, it ispossible to produce prescriptive and predictive mod-els and design interventions that allow us to achievegoals. This implies that an understanding of thecausal links in past behavior allows us to define best

practice for future behavior. It also implies thatthere must be a right or ideal way of doing things.

The assumption of rational choice: that faced with achoice between one or more alternatives, human ac-

Copyright 2003 by International Business Machines Corpora-tion. Copying in printed form for private use is permitted with-outpayment of royaltyprovidedthat (1)each reproduction is donewithout alteration and (2) the Journal reference and IBM copy-right notice are included on the first page. The title and abstract,but no other portions, of this paper may be copied or distributedroyalty free without further permission by computer-based andother information-service systems. Permission to republish anyother portion of this paper must be obtained from the Editor.

KURTZ AND SNOWDEN 0018-8670/03/$5.00 2003 IBM IBM SYSTEMS JOURNAL, VOL 42, NO 3, 2003462


2/22

tors will make a rational decision based only on

minimizing pain or maximizing pleasure; and, in con-sequence, their individual andcollective behavior canbe managed by manipulation of pain or pleasure out-comes and through education to make those con-sequences evident.

The assumption of intentional capability: that the ac-quisition of capability indicates an intention to usethat capability, and that actions from competitors,populations,nation states, communities, or whatevercollective identity is under consideration are the re-sult of intentional behavior. In effect, we assume thatevery blinkweseeisa wink, and act accordingly.We accept that we do things by accident, but assumethat others do things deliberately.

Thispaper contendsthatalthough these assumptionsare true within some contexts, they are not univer-sally true. We also believe that in decision-makingat both policy-making and operational levels, we areincreasingly coming to deal with situations wherethese assumptions are not true, but the tools andtechniques which are commonly available assumethat they are.

Order and chaos in antiquity. The human distinc-tion between order and chaos goes back to an abun-

dant presence in mythology, in which order arisesout of (and thus requires) and then vanquishes (andthus destroys) the mysterious forces of chaos. Forexample, in the Enuma Elish, 2 the Babylonian epicof creation, the world began under the reign ofTiamat, the mother of all things. In Tiamats world,none bore a name, and no destinieswere ordained.After several generations, Tiamats god-children ap-pointed a champion to seize control. Marduk notonly defeated his ancestor, but split her up like aflat fish into two halves that became heaven andearth. He then proceeded to order the universe infiner and finer detail:

He [Marduk] made the stations for the great gods;The stars, their images, as the stars of the Zodiac,

he fixed.He ordained the year and into sections he divided

it;For the twelve months he fixed three stars. . . .He founded the station of Nibir [the planet Ju-

piter] to determine their bounds;That none might err or go astray . . .

Note the words fixed, ordained, divided, de-termined, err, and astray. Control (in the first

four terms) and an absolute knowledge of right and

wrong (in the last two) are the salient points of Mar-duks new world. Of course Tiamat was never en-tirely vanquished; forces of chaos appear in all tra-ditions in the form of tricksters andmalcontentssuchas Bacchus, Loki, Coyote, the Monkey King, Anansi,and Hermes. The forces of order and chaos dancedwith each other throughout ancient times.

Science, order, and epiphenomena. Aristotle de-fined four types of cause: the material (what you aremade of,your muscles andorgans), the efficient (howyou came to be, the fact that your parents gave birthto you), the formal (your type, your species), and thefinal (yourfunction, your life itself, your place in theuniverse). He believed that to understand an eventor entity, one had to consider all of these factors inthe particular, the mysterious as well as the ordered.As others have pointed out, 3 the focus of Westernthinking post Kant on efficient cause only is to thedetriment of knowledge.

Since the birth of enlightenment science, the distinc-tion between order and chaos has held a prominencethat has profoundly influenced conceptual and prac-tical thinking. Kant separated things that we canknow empirically from things that are the provinceof God, and thereby helped to section off all but ef-

ficient causes to epiphenomena that could be safelyignored. This concept of ordered science triggereda massive growth in human knowledge andextendedover many disciplines. For example, sociology grewout of philosophy partly in an attempt to create ascienceof society thatcould duplicate the advancesbeing made in physics and biology through system-atic observation and causal explanation. It was ar-gued, mainly by Comte, that it was theoretically pos-sible to discover laws similarto those of physics whichcould explain thebehaviorof peoplein societies. Thegrowth of technology and the dominance of engi-neering-based approaches arising from the need for

automation and scalability reenforced the desire forand the assumption of order. In popular literature,the belief that all things can be known (in a New-tonian sense) persisted well into this past century.Asimovs classic science fiction Foundation Trilogy 4

builds on the character of Hari Seldon, the founderof psychohistory, whose mathematics permits theprediction of human behavior andsocial changecen-turies into the future.

The development of management science, from stop-watch-carrying Taylorists5 to business process reengi-neering, was rooted in the belief that systems were

IBM SYSTEMS JOURNAL, VOL 42, NO 3, 2003 KURTZ AND SNOWDEN 463


3/22


4/22

ulations of human behavior able to encompass

multiple dynamic individual and collective identitiesacting simultaneously and representing all aspectsof perception, decision-making, and action.

Humans are not limited to acting in accordance with

predetermined rules. We are able to impose struc-ture on our interactions (or disrupt it) as a result ofcollective agreement or individual acts of free will.

We are capable of shifting a system from complex-ity to order and maintaining it there in such a waythat it becomes predictable. As a result, questionsof intentionality play a large role in human patternsof complexity.3 Itisdifficult to simulate true free willand complex intentionality (for example, retrospec-tive elaboration, duplicity, groupthink, rumor, self-deception, manipulation, surprise, confusion, inter-

nal conflict, stress, changes in the meanings ofpreviously unambiguous messages, the deliberatecreation of ambiguity, inadvertent disclosure, cha-risma, cults, and pathologies) within a rule-basedsimulation. Social simulations have addressed issuessuch as cooperation, reputation, gossip, lying, andtrust,26 but always within an artificial frameworkwhich allows only limited numbers of options andconsiders limited numbers of phenomena operatingat once. It is interesting that searching the Internetfor simulation and most of the terms listed abovebrings up instances of simulations with which a userinteracts to explore patterns. 27 This may represent

a gap between agent-based simulation and humanbehavior similar to that found by the strong arti-ficial intelligence (AI) school of the 1970s whosegoal was to reproduce human intelligence. 28 We donot mean to say that there is no value to simulationof human behavior, but we do think we should notexpect it to succeed any time soon in predicting whatpeople will do in any particular circumstance.

Eventually the concept of intelligent augmenta-tion 29 30 grew in popularity and could be said tobridge the gap between strong AI and reality. It ispossiblethat a middle ground between the beliefthat

all human behavior can be simulated and the belief

that the very effort is questionable will be found inthe use of simulation not to explain or imitate butto support human decision-making.

Humans are not limited to acting on local patterns.

People have a high capacity for awareness of large-scale patterns because of their ability to communi-cate abstract concepts through language, and, morerecently, because of the social and technological in-frastructure that enables them to respond immedi-ately to events half a world away. This means thatto simulate human interaction, all scales of aware-ness must be considered simultaneously rather thanchoosing one circle of influence for each agent. Thereis also the matter of simulating theinteraction (con-flict, reinforcement) between local and globalawareness. Many of the emergent patterns we seein nature depend critically on the limited (that is,local) ability of activators to diffuse through a vis-cous medium.13 We have not yetseen addressed howthese issues cause complex patterns in human so-cieties to differ from complex patterns in systems oflocally aware agents.

We call our practice of keeping the human contextforemost in our considerations phenomenological ormore commonly contextual complexity. It means

mainlythatwhenwe use agent-based simulation(andwe do, in certain circumstances), we use it as a toolfor the exploration of possibility and generation ofideas, not as a tool for recommending courses of ac-tion.

Order and un-order. To avoid much repetition of thelonger termsdirectedorder and emergent order,we call emergent order un-order. Un-order is notthe lack of order, but a different kind of order, onenot often considered but just as legitimate in its ownway. Here we deliberately use the prefix un- notin its standard sense as opposite of but in the less

common sense of conveying a paradox, connotingtwo things that are different but in anothersense thesame. Bram Stoker used this meaning to great ef-fect in 1897 with the word undead, which meansneither dead nor alive but something similar to bothand different from both. According to R. D. Cure-ton, e. e. cummings also used the prefix this way inhis poetry. Says Cureton, 31 In normal usage, beingand existing are stative concepts. They are not ac-tions which a person must consciously perform, en-gage in, create. Words such as unbe andunexist, how-ever, force the reader to see the dynamic nature ofhuman existence. . . . Thus by our use of the term

We are capable of shifting

a system from complexity

to order and maintaining it therein such a way that

it becomes predictable.



5/22

un-order, we challenge the assumption that any

order not directed or designed is invalid or unim-portant.

Keep the baby, lose the bathwater. Let us sound aquick warning about running into the trap of believ-ing that everything is complex. Some recent popu-lar books on complexity in businessand managementhave been full of breathy enthusiasm for the edgeof chaos and would have businesses maintainthem-selves as far from equilibrium as possible, regard-less of context or purpose. We think this is throwingout the baby with the bathwater. We cannot simplygo from sayingthings are ordered to saying thingsare un-ordered and leave it at that; things are bothordered and un-ordered at once, because in realityorder and un-order intertwine and interact. Kostof32

puts it well in his description of cities: . . . the twoprimary versions of urban arrangement, the plannedand the organic, often exist side by side. . . . Mosthistoric towns, and virtually all those of metropol-itan size, are puzzles of premeditated and sponta-neous segments, variously interlocked or juxta-posed. . . . In other words, it is useful to artificiallyseparate order and un-order so that we can under-stand the different dynamics involved, but we shouldnot expect to find one without the other in real life.In many organizations, for example, formal com-

mand structures and informal trust networks sup-port (while simultaneously competing with) eachother. 3334 The joke that the only thing worse thanan inefficient bureaucracy is an efficient bureaucracyhas some ground in reality.

Methods for un-ordered space

Ordered-systems thinking assumes that through thestudy of physical conditions, we can derive or dis-cover general rules or hypotheses that can be em-pirically verified and that create a body of reliableknowledge, which can then be developed and ex-

panded. As we have mentioned, this assumptiondoesnot hold in the domain of un-order.

In practice, all decision makers know this: howevermuch they might like thingsto be ordered, they knowthat there are also circumstances in which culturalfactors, inspired leadership, gut feel, and othercomplex factors are dominant. All of these are pat-terns, which arise through the interaction of variousentities through space and time. In the space of un-order the seeds of such patterns can be perceived,and new ways of thinking can emerge. In fact, learn-ing to recognize and appreciate the domain of un-

order is liberating, because we can stop applying

methods designed for orderand instead focus on le-gitimate methods that work well in un-ordered sit-uations. TomStewart35 references the case of a groupof marines taken to the New York Mercantile Ex-change in 1995 to be taught and to play with sim-ulators of the trading environment. Naturally thetraders won each time. But when the traders visitedthe Marine Corps base in Quantico and playedwar games against the marines, they won yet again.What they realized is that the traders were skilledat spotting patterns and intervening to structurethose patterns in their favor. The Marines, on theother hand, like most business school graduates, hadbeen trained to collect and analyze data and thenmake rational decisions. In a dynamic andconstantlychanging environment, it is possible to pattern un-order but not to assume order.

In another case, a group of West Point graduateswere asked to manage the playtime of a kindergar-tenasa final year assignment. The cruel thing is thatthey were given time to prepare. They planned; theyrationally identified objectives; they determinedbackup and response plans. They then tried to or-der childrens play based on rational design prin-ciples, and, in consequence, achieved chaos. Theythen observed what teachers do. Experienced teach-ers allow a degree of freedom at the start of the ses-sion, then intervene to stabilize desirable patternsand destabilize undesirable ones; and, when they arevery clever, they seed the space so that the patternsthey want are more likely to emerge.

In the ordered domain we focus onefficiency becausethe nature of systems is such that they are amenableto reductionist approaches to problem solving; thewhole is the sum of the parts, and we achieve op-timization of the system by optimizationof the parts.In the domain of un-order, the whole is never thesum of the parts; every intervention is also a diag-nostic, and every diagnostic an intervention; any actchanges thenature of thesystem. As a result, we haveto allow a degree of sub-optimal behavior of eachof the components if the whole is to be optimized.

Pattern entrainment

Humans use patterns to order the world and makesense of things in complex situations. Give a childa pile of blocks, and he or she will build patterns outof them. Give an adult a daily commute, and he orshe will build patterns within it. Patterns are some-

KURTZ AND SNOWDEN IBM SYSTEMS JOURNAL, VOL 42, NO 3, 2003466


6/22

thing we actively, not passively, create, as Mary Dou-

glas36

so well pointed out:

. . . whatever we perceive is organized intopatternsfor which we the perceivers are largely responsi-ble. . . . As perceivers we select from all the stim-uli falling on our senses only those which interestus, and our interests are governed by a pattern-making tendency, sometimes called a schema. Ina chaos of shifting impressions, each of us con-structs a stable world in which objects have rec-ognizable shapes, are located in depth and havepermanence. . . . As time goes on and experiencebuilds up, we make greater investment in our sys-

tems of labels. So a conservative bias is built in.It gives us confidence.

Visually, we hold in sharp focus at any one instanta mere tenth of a percent of our visual range, so eventhe process of seeing is one of putting together manydisparate observations. 37 We fill in the gaps to cre-ate an experience-based pattern on which we act.This aspect of human decision-making is a greatsource of power, but it also brings limitation.

A television advertisement for a liberal broadsheetnewspaper in the United Kingdom illustrates this

well. The advertisement is set in a terraced inner citystreet. It is dusk; litter blows down the street; over-all, a threatening environment.

Scene One: The camera points down the street andpicks up a skinhead who comes around the cor-ner. A police car stops beside the skinhead, whoimmediately runs towards the camera. Pattern en-trainment of decision-making: what do we as-sume? The skinheadmust be running from the po-lice.

Scene Two: The camera now changes its perspec-

tive; it is now behind the skinhead, and we see thathe is running towards a well dressed man who isclutching a briefcase which must be full ofmoney, as the man conforms to the stereotype ofa rent collector. We can also see that the man isterrified in the face of the rapidly approachingskinhead. Pattern entrainment of decision mak-ing: what do we assume? The man is going to bemugged.

Scene Three: The camera changes perspective forthe final time; we are now looking down on thestreet from above, and we see the skinhead grab

the man and pull him into the portico of a build-

ing just before a crate of building material wouldhave fallen on his head and killed him.

The unspoken message of a brilliant advertisementis, See things from a different perspective; read thenewspaper.

Most of the time if you are standing in a run-downinner city area at dusk carrying a briefcase full ofmoney, and a skinhead suddenly runs towards you,it is not a good idea to stand there and say Ah, Imay be about to be rescued from a crate of fallingbuilding material; you should run for your life. Theissue in decision-making is to know when to run andwhen to stand still. A choice must be made betweenallowing the entrained patterns of past experienceto facilitate fast and effective pattern application andgaining a new perspective because the old patternsmay no longer apply. We will address these differ-ences in the following section.

The Cynefin framework

The name Cynefin is a Welsh word whose literaltranslation into English as habitat or place fails todo it justice. It is more properly understood as theplace of our multiple affiliations, the sense that we

all, individually and collectively, have many roots,cultural, religious, geographic, tribal, and so forth.We can never be fully aware of the nature of thoseaffiliations, but they profoundly influence what weare. The name seeks to remind us that all humaninteractions are strongly influenced and frequentlydetermined by the patterns of our multiple experi-ences, both through the direct influence of personalexperience and through collective experience ex-pressed as stories.

The Cynefin framework originated in the practiceof knowledge management as a means of distinguish-

ing between formal and informal communities, andas a means of talking about the interaction of bothwith structured processes and uncertain conditions.It has now outgrown its application in knowledgemanagement, having been in use by our group forseveral years in consultancy and action research inknowledge management, strategy, management,training, cultural change, policy-making, product de-velopment, market creation, and branding. We arenow beginning to apply it to the areas of leadership,customer relationship management,and supplychainmanagement, with other topics to come. It has alsobeen used by third parties. 38



7/22

We consider Cynefin a sense-making framework,which means that its value is not so much in logicalarguments or empirical verifications as in its effecton the sense-making and decision-making capabil-

ities of those who use it. We have found that it givesdecision makers powerful new constructs that theycan use to make sense of a wide range of unspec-ified problems. It also helps people to break out ofold ways of thinking and to considerintractableprob-lems in new ways. The framework is particularlyuse-ful in collective sense-making, in that it is designedto allow shared understandings to emerge throughthe multiple discourses of the decision-making group.

We make a strong distinction here between sense-making frameworks and categorization frameworks.In a categorization framework, four quadrants are

often presented in a two-by-two matrix (for exam-ples, pick up any management textbook or analystreport). Typically, it is clear (though often unstated)that the most desirable situation is to be found inthe upper right-hand quadrant, so the real value ofsuch a framework is to figure out how to get to theupper right. In contrast, none of the domains we willdescribe here is more desirable than any other; thereare no implied value axes. Instead, the frameworkis used primarily to consider the dynamics of situ-ations, decisions, perspectives, conflicts, and changesin order to come to a consensus for decision-mak-ing under uncertainty.

As can be seen in Figure 1, the Cynefin framework

has five domains, four of which are named, and afifth central area, which is the domain of disorder.The right-hand domains are those of order, and theleft-hand domains those of un-order.

Ordered domain: Known causes and effects. Here,cause and effect relationships are generally linear,empirical in nature, and not open to dispute. Re-peatability allows for predictive models to be cre-ated, and the objectivity is such that any reasonableperson would accept the constraints of best practice.This is thedomain of process reengineering,in whichknowledge is captured and embedded in structuredprocesses to ensure consistency. The focus is on ef-ficiency. Single-point forecasting, field manuals, andoperational procedures are legitimate and effectivepractices in this domain. Our decision model hereis to sense incoming data, categorize that data, andthen respond in accordance with predeterminedpractice. Structured techniques are not only desir-able but mandatory in this space.

Ordered domain: Knowable causes and effects.While stable cause and effect relationships exist inthis domain, they may not be fully known, or theymay be known only by a limited group of people. Ingeneral, relationships are separated over time and

space in chains that are difficult to fully understand.Everything in this domain is capable of movementto the known domain. The only issue is whether wecan afford the time and resources to move from theknowable to the known; in general, we cannot andinstead rely on expert opinion, which in turn createsa key dependency on trust between expert advisorand decision maker. This is the domain of systemsthinking, the learning organization, and the adap-tive enterprise, all of which are too often confusedwith complexity theory.18 In the knowable domain,experiment, expert opinion, fact-finding, and scenar-io-planning are appropriate. This is the domain of

methodology, which seeks to identify cause-effectrelationships through the study of properties whichappear to be associated with qualities. For systemsin which the patterns are relatively stable, this is bothlegitimate and desirable.

Our decision model here is to sense incoming data,analyze that data, and then respond in accordancewith expert advice or interpretation of that analysis.Structured techniques are desirable, but assumptionsmust be open to examination and challenge. This isthe domain in which entrained patterns are at theirmost dangerous, as a simple error in an assumption

Figure 1 Cynefin domains

COMPLEX

Cause and effect are onlycoherent in retrospectand do not repeat

Pattern management

Perspective filters

Complex adaptive systems

Probe-Sense-Respond

CHAOS

No cause and effectrelationships perceivable

Stability-focusedintervention

Enactment tools

Crisis management

Act-Sense-Respond

KNOWN

Cause and effect relationsrepeatable, perceivableand predictable

Legitimate best practice

Standard operatingprocedures

Process reengineering

Sense-Categorize-Respond

KNOWABLE

Cause and effectseparated over timeand space

Analytical/Reductionist

Scenario planning

Systems thinking

Sense-Analyze-Respond



8/22


9/22

ences near the center of the space. As a result, in-dividuals compete to interpret the central space onthe basis of their preference for action. Those mostcomfortable with stable order seek to create or en-

force rules; experts seek to conduct research and ac-cumulate data; politicians seek to increase the num-ber and range of their contacts; and finally, thedictators, eager to take advantage of a chaotic sit-uation, seek absolute control. The stronger the im-portance of the issue, the more people seem to pullit towards the domain where they feel most empow-eredby theirindividual capabilitiesand perspectives.We have found that the reduction in size of the do-main of disorder as a consensual act of collabora-tion among decision makers is a significant steptoward the achievement of consensus as to the na-ture of the situation and the most appropriate re-

sponse.

The apple-orange problem

People are often confused by the apple-orange na-ture of the four Cynefin domains: they say, Whynot known, knowable, somewhat knowable and un-knowable? or, Why not simple, complicated, com-plex and chaotic? The distinction is intentional. TheCynefin framework is a phenomenological frame-work, meaning that what we care most about is howpeople perceive and make sense of situations in or-der to make decisions; perception and sense-mak-

ing are fundamentally different in order versus un-

order. The framework actually has two largedomains, each with two smaller domains inside. Inthe right-side domain of order, the most importantboundary for sense-making is that between what wecan use immediately (what is known) and what weneed to spend time and energy finding out about(what is knowable). In the left-side domain of un-order, distinctions of knowability are less importantthan distinctions of interaction; that is, distinctionsbetween what we can pattern (what is complex) andwhat we need to stabilize in order for patterns toemerge (what is chaotic). Thus we often draw theframework with the vertical boundaries strong andthe horizontal boundaries weak, denoting their rel-ative importance in sense-making.

The Cynefin framework is based on three ontolog-ical states (namely, order, complexity and chaos) anda variety of epistemological options in all three ofthose states. We arecurrently engaged in furthercon-ceptual and experimental work to more strongly de-velop the separation of ontological from epistemo-logical aspects of the framework in order to root theframework in a variety of scientific disciplines whilemaintaining the essential interweaving of ontologyand epistemology, which appears to be an essentialaspect of human sense-making in practice.

Connection strengths of Cynefin domains

Another way to look at the Cynefin framework is inthe types of component connections that are mostprevalent in each domain (Figure 2). On the side oforder, connections between a central director andits constituents are strong, often in the form of struc-tures that restrict behavior in some wayfor exam-ple, procedures, forms, blueprints, expectations, orpheromones. On the side of un-order, central con-nections are weak, and attempts at control throughstructure often fail from lack of grasp or visibility.

In the complex and knowable domains, connectionsamong constituent components are strong, and sta-ble group patterns can emerge and resist changethrough repeated interaction, as with chemical mes-sages, acquaintanceship, mutual goals and experi-ences. The known and chaotic domains share thecharacteristic that connections among constituentcomponents are weak, andemergent patterns do notform on their own.

In any of these domains, a reasonable strategy cap-italizes on the stability afforded by strong connec-tions without allowing them to harden so much that

Figure 2 Connection strength of Cynefin domains

COMPLEX

WEAK CENTRAL,STRONGDISTRIBUTED

CHAOS

WEAK CENTRAL,WEAKDISTRIBUTED

KNOWABLE

STRONG CENTRAL,STRONGDISTRIBUTED

KNOWN

STRONG CENTRAL,WEAKDISTRIBUTED



10/22

they destroy flexibility and also capitalizes on the

freedom and renewal afforded by weak connectionswithout allowing them to permanently remove use-ful patterns.

How the Cynefin framework is used

In our engagements with clients, we use the Cynefinframework at several different levels, depending onthe context and purpose of the project. Several com-plementary exercises and programs help people tointernalize and make use of the framework. Somegroups consider only thefive domains and what sortsof situations or problems can be found there; someconsider distinctions between extreme and equivo-cal conditions within domains; some consider mul-tiple perspectives and how they can be resolved orused to advantage; some talk about boundary tran-sitions, boundary sensing, and boundary manage-ment; some talk about dynamics ranging over thewhole space of the framework. For example, groupsmight use theCynefin framework to gain new insightson a contentious issue, plan interventions to movea situation from one domain to another, considerhow they should approach or manage different for-mal and informal communities, or differentiate theirstrategies for knowledge retention based on multi-ple contexts of knowledge exchange.

Contextualization. This is a critical exercise that en-ters into nearly every Cynefin project. Contextual-ization is also a good example of how the Cynefinframework concentrates on collective sense-makingas a consequence of discourse. We will describe atypical contextualization session here. The sessionbegins with the collection of many items throughstructured brainstorming. The items might be com-munities, products, actions, motivations, forces,events, points of view, beliefs, traditions, rituals,books, metaphors, anecdotes, myths, and so on: theyare any items that are important to the sense-mak-

ing process. The items are related to one theme orissue of concern, which should be broad but not in-finitely so. Whatever sorts of items are chosen (andmultiple types are permissible), they should be di-verse and concrete in nature: diverse to allow mul-tiple perspectives to emerge, and concrete to moveaway from existing entrained abstract beliefs. We as-sure diversity by giving different groups of peopledifferent directions, by giving directions that are de-liberately ambiguous and so can be taken in diverseways, and by changinggroup compositions frequentlyso that people do not fall into entrained thinking.To keep items concrete, we rely heavily on narrative

methods. These provide a rich context that allows

patterns of experience rather than opinion or beliefto emerge.

We use an array of different methods to help peopleprepare a contextually meaningful field of items forsense-making, including:

Narrative database. We might prepare in advance anarrative database of relevant stories surroundingthe issue, drawn from oral histories, collected an-ecdotes, published reports, historical documents, andthe like. 40 Workshop participants might be asked toreview the material in advance and answer somesense-making questions about it. In the first part ofthe workshop, we may ask people to talk about theirthoughts on the material and the narratives whichthey found most compelling. From these discussionsa list of such items as situations, actors, events, andforces might be generated.

Convergence methods. Rather than pulling items di-rectly from anecdotes discussed, we might go throughone of several convergence methods in order to in-tegrate much disparate material andachieve a stron-ger set of sense-making items. For example, we mightask people to construct composite fables from an-ecdotes by using one of a variety of fable templates,

working from either the narrative database or theirown experiences as source material. 41 This exerciseis especially useful when we need to bring peopleinto the realm offiction so that they can more freelyexpress their true opinions on sensitive subjects. Itis also a useful integrator of diverse sources. In aroom of 50 people constructing fables in smallgroups, several hundred anecdotes may be consid-ered as material for creation of each groups fable.Items are then drawn from aspects of the integratedfables (characters, events, situations) for sense-mak-ing.

Alternative history. Another useful method (with orwithout preparatory material) is the use of alterna-tive histories. In this method, we ask people to de-scribe the history of an organization, society, or event,always working backwards from the present (to anystarting time they think is appropriate). We ask themto determine turning points (moments when smallevents caused large changes), write them simply onhexagonal sticky notes, andarray them on a wall. (Weprefer hexagonal notes because people tend to clus-ter them in beehive shapes, while people tend to cat-egorize square notes.) When the factual history iscomplete, we then ask people to choose two extreme



11/22


12/22

At this point, the characteristics of each domain and

boundaryas they pertain to the context at handcan be considered using the sense-making items inplace.

Use of the contextualized framework. After theCynefin framework has been created, we may moveon to other exercises in which the items in specificareas (for example, near the boundaries or at theextremes) are considered in more detail or con-versely, in which the entire contextualized frameworkis used to describe the changing dynamics in a his-torical case or a contemporary situation, possiblyfrom different points of view.

We should point out here that though we draw theframework as four simple areas in two-dimensionalspace (because it is quick to draw and grasp), thisis only a quick reference for discourse, and that sim-ple drawing expands into something more multidi-mensional in use. For example, we might ask peopleto consider different aspects of or perspectives ona single situation that might be located in differentareas or moving in different directions. In this sense,all sense-making exercises transcend the simple rep-resentationof the frameworkat some point. We con-sider the framework to be a jewel of contempla-

tion that has many facets, like a geographicalinformation system, in which many layers of in-formation are overlaid so that they can be inte-grated or separated at will; but the ultimate re-ality is that every representation (including theentire framework itself) is created for a purpose.

The value of the completed contextualized frame-work lies in two main benefits. First, nearly every con-textualization exercise we have seen has ended withexpressions of surprise from those participating.They often see, for the first time, patterns that over-turn their entrained beliefs about the issue they are

considering and about their purpose, goals, and iden-tity. For example, one group completed their Cyne-fin framework and reviewed it. They had done al-ternative histories to derive their sense-making items,and they had been asked to provide at least one ac-cident in each fictional time line. On their Cynefinframework, theonly items in the chaos domain werethose (color-coded) accidents. One participantlooked at them and said, We are being complacent,arentwe? Theparticipant meant that they hadonlybeen willing to consider wholly chaotic situationswhen they had been forced to add accidents to thelist.

This increased awareness (reflected in many other

such stories) is the highest achievement of the com-pletion of the contextualization exercise: that thegroup should accomplish Descriptive Self-Aware-ness, or a greater understanding of their own biasesand potentials. This is also our goal in helping peo-ple go through the process, because it is our placeto enable clients to achieve self-awareness ratherthan to provide expert advice, which has a muchlower value in practice.

The second benefit of the contextualized Cynefinframework is that it provides a new shared languagewith which the members of the decision-makinggroup can discuss situations, perspectives and pos-sible actions. This new language is unique to the con-cerns of thegroup andabstract enough to covermanyparticulars,but resonant with meaningso thatit tendsto be brought up spontaneously when issues are dis-cussed. It can be used to talk about interpretationsof currentconditionsbased on gathered data, to eval-uate strategic interventions, and to constructivelymanage conflict and bring about consensus, withoutremoving conflict. Also, multiple groups who havecreatedtheir own contextualized frameworks can usethem as artifacts for negotiation of common mean-ing. For example, if one group placed theinfluenceof the Internet on globalization in complex space

and another group placed it in knowable space, bothgroups would learn something about their respec-tive perspectives. We have been helping groups touse such frameworks to guide discussion among dif-ferent government branches, for example.

We previously contrasted Cynefin as a sense-mak-ing framework with categorization frameworks. Wedo sometimes use the contextualized Cynefin frame-work for categorization within a particular context,meaning that theterms used are not generic but spe-cific. Categorization in context has some excellentuses: for example, for training in standard operat-

ing procedures, for aligning perspectives and objec-tives among groups (for example, strategic and op-erational), or for initiation of new people into thegroup.

Wider implications

As we mentioned previously, the contextualizationexercise is just one of several elements involved inusing theCynefin framework to support peoplemak-ing decisions and crafting strategies. As part of thework we are doing for public-sector and private-sec-tor bodies, including the government, we are design-



13/22


14/22

When we are using the Cynefin framework as con-

textualized by a group to their context and purpose, we ask them to consider the boundaries betweenCynefin spacesby using metaphorical types of bound-ary. We explain that each boundarymight have a dif-ferent form in one direction or the other, and thisis where we must break away from the two-dimen-sional drawing somewhat. In addition, one bound-ary might have different forms for different people,whose perceptions or circumstances make their expe-rience of the boundary different. Generally, we findthree basiclevels of sophistication in theuse of Cyne-fin boundaries for sense-making. First, one consid-ers an awareness of crossing the boundary, so thatone can respond quickly to new conditions after onehas arrived on the other side. Second, one considersan awareness ofapproaching the boundary, so thatone can sense when change is incipient and respondbefore the boundary is crossed (perhaps to cross itpurposefully, perhaps to avoid it). Third, one con-siders managing the boundary and the perceptionssurrounding it, so that one can, for example, put adeep-chasm boundary in place for ones adversarywhile maintaining a shallow-river boundary for onesown use.

One thing we have foundin using boundaries as partof the Cynefin framework is that different people,

with different training and personalities, seem to ben-efit from different uses of boundaries. People whoare used to classifying items into categories benefitfrom removing boundaries, as takes place in the stan-dard contextualization exercise described earlier.However, people who are used to thinking in a morefluidwayabout gradients rather thanboundariesseem to benefit more by constructing boundariesthan by removing them.

For example, when the workshop is large enough orwhen we feel that it is required, we often hold a par-allel session of contextualization in which people are

asked to distribute their sense-making items alonga line, ranging from the most tractable items to themost intractable items. After the line has been com-pleted (and there are many negotiations to placeitems in relation to each other), we ask people tofind places along the line at which they feel that theunderlying dynamic has shifted. In other words, weask them to create boundaries along gradients. Wethen pull the line into a rainbow curve and place iton the Cynefin framework, with the most tractableitems in the known domain and the most intractableitems in chaotic space. This produces an alternatecontextualization, with the same ultimate effect of

creating the Cynefin framework anew, but with a

stronger emphasis on the negotiation of whereboundaries are found. We sometimes ask people tonegotiate boundaries as though they were represen-tatives of the different domains, coming up with amutual agreement on what the boundary means andwhere it is placed.

The boundary issue is one on which we continue topursue a strong research agenda (although of coursethe entire framework is a work in progress), espe-cially with regardto its use for strategic decision sup-port.

Cynefin dynamicsWhen people use the Cynefin framework, the waythey think about moving between domains is as im-portant as the way they think about the domain theyare in, because a move across boundaries requiresa shift to a different model of understanding and in-terpretation as well as a different leadership style.Understanding the differences among the differentmovements in the framework increases the sophis-tication of the response of a decision-making groupto rapid change. We describe here some of the pat-terns of movement which we use to help groups con-sider historical, contemporary, and future change.

In general, one of thefunctions of theCynefin frame-work is to increase awareness of the upper domainsof the framework and their potential to create sus-tainable change; several of these movements are de-signed to make the upper domains more accessible.

Movement at the known-chaos boundary. Thisboundary is the strongest of the four, in which a per-fectly working machine operates inches away froma devastating fire. For that reason, this boundary isthe most dangerousand the most powerful iftreated with respect.

Asymmetric collapse (Item 1 in Figure 4) is move-ment from the known to the chaotic, disastrously.We have seen a tendency for organizations to os-cillate between the domains of the known and thechaotic, avoiding the upper domains. Organizationssettle into stable symmetric relationships in knownspace and fail to recognize that the dynamics of theenvironment have changed until it is too late. Thelonger the period of stability and the more stable thesystem, the more likely it is for asymmetric threatsor other factors to precipitatea move into chaos. Thedecision makers in the system dont see things thatfall outside the pattern of their expectation, and they



15/22


16/22

plexity as exploration versus exploitation. 17 Explo-

ration is an opening up of possibilities by reducingor removing central control without a total disrup-tion of connections. In organizations, explorationtakes many forms, but trust is key in this movement.One is, in effect, taking a risk by allowing constit-uent connections to form and strengthen at the ex-pense of central control, and that requires not onlygood planning and awareness of the shadow side

of the organization, but also careful (but unobtru-sive) monitoring of the situation. In most organiza-tions there is a strong and often untapped resourceto be found in exploratory moves such as this. Forexample, informal communities, which may rangefrom public to secret in their profile, provide a richand fertile source of knowledge and learning that istoo large and complex to be formally managed. One

study of actual practice in IBM Global Services 34 in-dicated some 50 60 official knowledge areas, com-plemented by many tens of thousands of private ar-eas. By providing spaces in which members of theorganization could naturally share with people theytrust, a fertile source of learning was created.

Just-in-time (JIT) transfer (Item 5 in Figure 4) ismovement from the complex to the knowable, se-lectively. This movement is often called exploitationin the complexity literature, and it involves the se-lective choice of stable patterns in complex space forordered representation. In the IBM example just

cited, the 50 60official communities are able to drawon information coming out of the informal trust-based communities as long as they refrain from dis-turbing their delicate balance. Knowledge can bemoved into the formal space on a just-in-time basis:knowledge is made available when it is needed. Tech-niques to achieve this include subject-matter flag-ging and privacy-ensured searching of content. Theissues here are those which gave rise to JIT techniquesin manufacturing some decades ago: organizationsrealizedthat the cost of maintaining stock on the fac-tory floor was out of all proportion to the benefits,with high levels of wastage over and above stock hold-

ing costs. In consequence, stock holding shifted back

to the suppliers, entering the factory just in time.43

Movement at the complex-chaotic boundary. Thisboundary, like the known-knowable boundary, isfluid and in fact difficult to delineate. In nature, sys-tems move back and forth across this boundary of-ten. In that sense, traffic across this boundary mir-rors that across the known-knowable boundary: oneis an engine for technological and scientific order,and the other is an engine for organic order. In thesocial sphere, we can use the engine of complexityto enable emerging patterns that prove useful.

Swarming (Item 6 in Figure 4) is movement from thechaotic to the complex, to the knowable; first, in anemergent manner and then, selectively. Draconianimposition of order is most appropriate in symmet-ric conditions and partial remediations, but underasymmetric conditions, or when whole-system inter-ventions are required, we need to move from chaosto the complex, not to the known. The boundary be-tween chaos and order is a chasm difficult to cross,but a vertical transition across the more permeableboundary betweenchaos andcomplexityis inherentlymore manageable. A transition from the chaotic tothe complex is a matter of creating multiple attrac-

tors, or swarming points, around which un-order caninstantiate itself, whereas a transition from the cha-otic to the known requires a single strong attractor.For example, if one were trying to evacuate a pan-icked crowd in a theater on fire, it would make moresense to shout out the blinking orange lights areabove the exit doors,which is a complex swarming-point triggerthat relies on local knowledge only, thanto shout out come towards the back of the theatre,an ordered trigger that relies on global knowledgewhich may be unavailable.

After we have achieved the shift from chaos to thecomplex, then we have the possibilities of many pat-terns forming around the new attractors; those wefind desirable we stabilize through a transfer to theexploitable domain of the knowable; those that areundesirable are destroyed. We have found in sev-eral recent engagements that the contrast ofswarm-ing with imposition provides a new language for ex-ecutives and appears to provide new perspectives oncrisis management. We will be studying actual be-havior and creating more subjects in this field overthe next year as our use of dynamic movementswithin the Cynefin framework develops.

Informal communities,

which may range from public

to secret in their profile, providea rich and fertile source

of knowledge and learning.



17/22

Divergence-convergence (Item 7 in Figure 4) is amovement from thecomplexto the chaotic andback,repeatedly. The active disruption of a complex sys-tem to precipitateits move to chaosis less of a change

than moving it to either of the ordered domains, andthis is easier to manage across a permeable bound-ary. In knowledge management, for example, infor-mal communities that occupy the complex domainare more resilientwhen askedto undergo radical dis-ruption in an innovation program than the expertcommunities of the knowable domain.39 Smallstart-up companies handle disruption better thanlarge bureaucratic ones, but even within large bu-reaucratic organizations, there are small groups thatcan act in the role of start-ups, and they can increasethe adaptability of the organization.

Our complexity-inspired workshop techniques, as ex-plained above, make intensive use of the boundarybetween the complex and the chaotic, in effect cy-cling between the dynamics of the two states as asort of pattern generator to create a rich variety ofpatterns among which to chooseto stabilize andto disruptin order to facilitate sense-making.

Visiting chaos

There are some good reasons to move deliberatelyfrom order to chaos. There are times when it is nec-essary to break rigid structures in precipitation of a

natural collapse (as one approaches the boundary),

so that the transition can be managed more care-fully; and there are times when a strong disruptionis the only mechanism that will break up a strongbut unhealthy stability. The last three movementtypes we will consider use the chaotic space for tem-porary disruption of all connections (possibly withina restricted context) as a stimulant to new growth.

Entrainment breaking (Item 8 in Figure 5) is move-ment from the knowable to the chaotic to the com-plex, periodically. In entrainment breaking, we movefrom the knowable to chaos and thus stimulate thecreation of new complex systems as the system re-bounds into the complex domain. This is a commonapproachto disrupt the entrained thinking of expertswho, in our experience, tend to be the most conserv-ative when it comes to radical new thinking. Themove to complex space is not radical enough to dis-rupt those patterns; we need to challenge at a morebasic level the current assumptions of order. By us-ing the complex space as a staging post, we createa more fertile space of interactions from which wecan select stabilization points for the movement tothe knowable. A knowledge management exampleis the creation of formal communities by clusteringand swarming informal activities from existingtrusted relationships. 39 In strategy, this method can

be used to create andvalidate newsources and struc-tures for decision-making.

Liberation (Item 9 of Figure 5) is movement fromthe known to the complex to the knowable, period-ically. Organizations tend to assume that they candesign the nature of new systems. For example, anorganization that needs new expertise in an areamight commission a university to carry out a study,recruit specialist staff, or identify individuals withinthe organization and assign them new responsibil-ities. This is a successful and effective strategy whenthe conditions are suitable for ordered approaches.

However, if thesituation is uncertain, it is more use-ful to shift the problem from the domain of theknown to the complex. Organizations need to in-crease both internal and external levels of contactto thepointwhere newpatternscan emerge.Boisot44

makes the point that companies need to use bothhoarding strategies, in which they place defensivebarriers around what they know and focus on exploi-tation, and sharing strategies, where knowledge isshared within and outside the organization with theintent of increasing the volume of opportunities, withthe strategic advantage shifting to speed of exploi-tation of knowledge.

Figure 5 Cynefin dynamics using chaos

8 ENTRAINMENT

BREAKING

9 LIBERATION

10 IMMUNIZATION



18/22

Foran organization, sharing strategies involve a pro-

cess of letting go, of creating freedom within heu-ristic boundaries to allownewpatterns andnew lead-ership to emerge. One of the techniques we use inthis areais Social Network Stimulation,39which aimsto stimulate the interactions of agents (orrather iden-tities) within systems to allow the emergence of newcoalitions, alliances, and leadership. To use a met-aphor, we cast seeds (ideas, deliberately ambiguousgoals), which are cheap, across a broad landscapeand see where growth occurs. As soon as growth isevident, we respond quickly to shift the newly emer-gent idea or leaders or coalition into the knowable.We have called this movement liberation because itbreaks the entrainment of bureaucracy but like allletting go, it is difficult. This is one of the most threat-ening of transitions to entrenched managers, but oneof the most important.

Immunization (Item 10 of Figure 5) is movementfrom the known to the chaotic, temporarily. Immu-nization in chaos is a smaller visit to chaotic spacethat shakes up the way things are enough to causereflection but not enough to destabilize the entiresystem. Immunization serves two purposes. First, itinures people to the devastating force of chaos sothat they will be better prepared to face those forcesin the future. A perfect example: it is said that the

great director Buster Keaton was able to craft hisdeath-defying stunts (such as a house falling aroundhim, a rescue from a drenching waterfall, amazingpratfalls, and so on)because as a toddler he wasliftedout of bed by a tornado and set down unhurt in thestreet.45 Second, immunization brings new perspec-tives, which cause radical disruptions in stable pat-terns of thought and lead to new complex patterns.Examples of such events are scattered throughoutliterature, in the accident that changes a politicianscareer, or the chance encounter that causes a lonelywomans lifeto fill up with new meaning, or in manyother kinds of radical departures that make every-

thing on which one had relied seem meaningless andrestricting.

Metaphors are particularly useful agents of immu-nizationbecause theyallow conversation about pain-ful things, enable disruptive and lateral thinking, pre- vent entrainment of attitudes, and clear out thecobwebs of stagnant ways. One technique we use forthis purpose is called the Grendel game (materialavailable from the authors, not yet published), whichcombines anthropological study, complexity theory,and managed war games to create an exciting andinnovative learning event. Here, following a study

of the organization, using anthropological techniques

developed and proven within earlier research intoanecdote collection, a fictional planet is populatedby aliens selected to reflect the current culture andnew scenarios. This is done with a leading scientist,who in his spare time designs consistent alien en-vironmentsfor use by sciencefiction and fantasy writ-ers. Members of the organization then seek to col-onize the planet in a managed war game. They facetheir own organization in a metaphorical setting thatallows more profound and meaningful learning.Newness is simulated without threat, and the par-ticipants are habituated to perspective shift and un-certainty.

Background movements

In any consideration of deliberate change, one mustconsider what is already going on. The forces of thepast tend to cause clockwise drift in the Cynefinspace: people living together and sharing mutualneeds lead to the emergence of ideas; convenienceleads to stabilization and ordering of the ideas; tra-dition solidifies the ideas into ritual; and sometimes,either lack of maintenance or the buildup of biasesleads to breakdown. The forces of the future pushdynamics to the counter-clockwise: the death of peo-ple and obsolescence of roles cause what is known

to be forgotten andrequire seeking; newgenerationsfilledwithcuriosity begin new explorations that ques-tion the validity of established patterns; the energyof youth breaks the rules and brings radical shifts inpower and perspective; and sometimes impositionof order is the result. In a sense, these two forces arealways pulling society in both directions at once, andthis is reflected in organizations as well. The oldguard is forgotten at the same time that its beliefsaffect newcomersin ways they cannot see. An aware-ness of these dynamics in the organization must pre-cede any deliberate attempt to affect it by deliber-ate change.

Use of Cynefin dynamics in practice

Our use of the boundary transitions described hererelies on narrative becauseboundaries are essentiallyabout change andnarrative is about change. We mayask people to consider a situation in the past andwhat movements took place in it from different per-spectives, or we may ask people to envision fictionalnarratives about the past, present or future in whichselected movements form the backbone of the story.(Remember this is all done not in the abstract, butusing the contextualized Cynefin framework that



19/22

makes sense in the context of concern.) These nar-

ratives of change are used in the sense-making pro-cess, in which they may be quickly created and dis-carded, or they may be saved and used to generatea shared language about change in much the sameway as cycles of folk tales long ago. They are not,however, allowed to stabilize into expectations; theymust remain fluid to be useful.

A critical distinction between this type of narrativegeneration and that of scenario planning 46 47 is thatthe source of Cynefin-based narratives is not a setof expected ranges of expected variables, but a con-sideration of dynamics in which the variables andranges are sometimes unknown and perhaps evenunspecified.This creates a more diverse, flexible, andchangeable set of narratives that should be truly sur-prising. For example, a scenario-planning exercisedone by a police force might result in scenarios ex-ploring the space from high to low crime, corrup-tion, andterrorism. These scenarios areuseful in con-sidering a future in which those axes are the mostimportant but what about the situation where theyare overridden by a previously irrelevant factor?Nar-ratives generated using a contextualized Cynefinframework explore spaces in which the dynamic sit-uation arises from any source. Optimally, a widerange of diverse possibilities (chaos from a nuclear

accident, a coup, an epidemic, an alien invasion) isbest, even though some scenarios might be incon-ceivable or even nonsensical. Conceivability is notthe point: preparation for the unexpected is.

Note that we do not believe these methods shouldsupplant scenario-planning in all contexts, becausethat method works well in knowable space, whereits results are productive. We do believe, however,that these are the methods of choice in complex sit-uations where a wider range of possibility needs tobe explored. In fact, as we increase the number ofmethods available for the un-ordered domain, meth-

ods that work well in the ordered domain get evenstronger because they are no longer used in situa-tions in which they have limited applicability.

Relationship to other frameworks

We do not pretend that all the basic ideas inherentin the Cynefin framework are new or unique. Theycan in fact be found floating around history for thou-sands of years. The distinction between order andun-order (and their interactions) is ancient, as wementioned, as well as being taken up by recent au-thors.48 The chaotic-complex distinction has been

much debated in recent years, with some saying com-

plexity exists at the edge of chaos,10

some sayingthat the two phenomena have separate origins andcannot be placed together, 19,49 and some even say-ing that the distinction is artificial and arbitrary.50

The distinction between known and knowable iswidespread andgoes back to ancient philosophy. We

do claim originality for the development of the ideasbehind the framework in its full form (as we havedescribed it, and in some other aspects outside thescope of this paper) and for the methods we use tomake the framework useful in practice (though ofcourse they have their relationships with other ac-tion research and sense-making methods).

Courtneys 51 framework, in which he distinguishesbetween four states of increasing residual uncer-

tainty (uncertainty that cannot be reduced by anal-ysis) questions universal assumptions about the useof known-space methods and tools. He does seemto believe, however, that the level of uncertainty canbe definedthat is, he has no domain of disorder(which is essentially a state of uncertainty about un-certainty). This, we suspect, may cause people togravitate to the domain they find most plausible, aswe have seen happen. Also, his toolkits of choicefor each level of uncertainty seem to cover only theknown and knowable spaces in the Cynefin frame-work, with a slight nod to narrative methods in com-plex space. In effect, he does not break out of the

righthand side of the Cynefin framework. He alsodoes not seem to consider the possibility that a sin-gle circumstance may contain competing aspects andperspectives with different degrees of uncertainty orthat such differences can be used to strategic advan-tage. Certainly there is much to be gained from allattempts to diversify responses to differing contexts,however.

Returning to assumptions

To complete our circle, we should address the threeassumptions we identified at the start of this paper.

As we increase the number

of methods available for the

un-ordered domain, methods that

work well in the ordered domain

get even stronger.



20/22

As we said at the beginning, these assumptions are

valid in some circumstances, but not in others. Letus examine each one again and consider how we canexpand these assumptions to a universally applica-ble set. We will also briefly speculate on emergingapplications of the Cynefin framework to manage-ment science and practice.

The assumption of order. Relaxing this assumptionis the basis of the Cynefin framework, which pro-poses instead two types of order, each with distinc-tions inside, and a recognition that uncertainty mayexist in distinguishing these types (the domain of dis-order). The assumption of order holds for orderedspace, the known and knowable, so we need new as-sumptions for the domains of un-order and disor-der. In complex space, we cansafelyassume that pat-terns will form, unpredictable in their details butusually recognizable in their basic forms, and thatwith practice we can learn to detect these formingpatterns, stabilize or disruptthemdepending on theirdesirability, and seed desirable patterns by creatingattraction points. In chaotic space, we can assumethat all connections have been broken, that possi-bility reigns, that old patterns have been disrupted,and that the outcome is not predictable. In the spaceof disorder, we know something very valuablethatwe do not know. We need to gain more understand-

ing (in every way possible) so that we can find pat-terns and react to them.

In management, relaxing assumptions of ordermeans recognizing that not all effective solutions areefficient solutions. It does not mean that trust hasto be given blindly or that complex processes can-not be affected; it only means that when the meansmatch the context, less energy need be expended forthe same result. In a very real sense, managers havesuccessful models available in the domains of un-or-der in the way they manage their children; they useboundaries and interventions to encourage desirable

behavior but do not attempt to control it throughgoal-based direction. The use of these methods isopening some high-potential developments in man-aging organizations following mergers and in creat-ing the conditions for continuous learning and in-novation.

The assumption of rational choice. Relaxing this as-sumption means that context and perspective be-come as important as rationality. This is an impor-tant reason that the Cynefin framework is not aboutobjective reality but about perception and under-standing; it helps us to think about theways in which

different people might be perceiving the same sit-

uation. For example, there is an old folk tale fromIndia 52 in which a wise man decides that in order toescape an impossible royal demand, he will fake in-sanity in the kings court. He is operating in com-plex space because he is using cultural shorthandsto provoke predictable reactions but is gambling thathis ruse will seed the pattern he wants to create. Heknows that from the perspective of his audience, whoare operating in the space where things are boundby tradition and thus known, he appears to be act-ingchaotically, because they canconceive of no otherreason for him to act this way in front of the king(who wouldsurely behead himif he was faking). Thusby proving that he cannot be faking, he pulls off thefake. Understanding not only that there are differ-ent perspectives on an event or situation, but thatthis understanding can be used to one s advantage,is the strategic benefit of relaxing this assumption.Narrative techniques are particularly suited to in-creasing ones exposure to many perspectives on asituation.

In management, there is much to be gained by un-derstanding that entrained patterns determine re-actions. This realization has major implications fororganizational change and for branding and market-ing. Our own work on narrative as a patterning de-

vice is gaining presence in this and other areas. Spec-ulating, one of the most significant possibleapplications of this understanding is a move awayfrom incentive-based targets and formal budgetingprocesses both of which, we contend, produce asmuch negative as positive behavior. It is a truism tosay that any explicit system will always be open togaming. Paradox and dialectical reasoning are keytools for managers in the un-ordered domains.

The assumption of intentional capability. The effectof relaxing this assumption is also one of consider-ing context, but more of context on action than on

perspective. It means asking not, What did they havein mind when they did that? but, What does itmean that thishappened? John F. Kennedy was pre-sented with such a dilemma when he received twoletters in rapid succession from the Soviets duringthe Cuban missile crisis, one conciliatory and onehard-line. 53 Which letter was a wink and which wasa blink? It turned out, of course, that both letterswere winks but of different actors (Khruschev andthe Politburo, respectively). The different authors ofthe letters probably saw their actions from the com-plex space of the Cynefin framework because theyunderstood the internal politics that produced them.



21/22

To Kennedy, however, receipt of the two letters

plunged him into chaotic space, where he could con-ceive of no sensible intention to send such contra-dictory lettersto confuse, perhaps, or to delaywhilethe missiles were being prepared. It was only afterscrutiny by linguists that the Kennedy administra-tion was able to understand that the intentions be-hindthe two letterscame fromdifferent sources. Thismoved his understanding into complex space wherehe could respond. It is only possible to consider al-ternative explanations for actions when one relaxesthe assumption that all actions are deliberate.

In management and in strategy, the issue of assump-

tions about intentions can have pronounced effects.For example, one may assume that the rejection ofa new initiative by employees is intentional, when infact it is an accident of emergent patterns of inter-action. Treating the rejection as intentional maycre-ate exactly the reaction that the managers wished toavoid in the first place. The same is true of over-re-action to accidental competitor behavior that is pre-sumed to be intentional.

Conclusion

This paper outlines a new approach to strategy, both

in policy formation and in operational decision-mak-ing. While it is new, this approach also recognizesas critical the value of what has been done to date.As such, it starts to break the fad cycle that has be-devilled management science for several decades. Itrecognizes the progression of human knowledge, inthat something which has provided value is not ren-dered valueless by new thinking, but is bounded bynew insight and legitimized within boundaries, andthereby, made more, rather than less, effective. Thisapproach is, we think, unique in that it recognizesthe value and interaction of order and un-order. Assuch, it also allows us to make a critical distinctionbetween efficiency and effectiveness. Human groupsneed to be effective; machines and structured hu-man interactions (such as manufacturing processesor the application of rules of engagement) need tobe efficient.

Cited references

1. D. Snowden, The Paradox of Story: Simplicity and Com-plexity in Strategy, Scenario and Strategy Planning 1, No. 5,16 20 (November 1999).

2. L. W. King and P. Tice, Enuma Elish, The Seven Tablets ofCreation; The Babylonian and Assyrian Legends Concerningthe Creation of the World and of Mankind, Book Tree Pub-

lishers (1999). Available on-line at http://www.sacredtexts.com.

3. A. Juarrero, Dynamics in Action: Intentional Behavior as aComplex System. MIT Press, Cambridge, MA (1999).

4. I. Asimov, The Foundation Trilogy: Foundation, Foundationand Empire, and Second Foundation, Ballantine Books, Inc.(1983).

5. F. W. Taylor,Principlesof Scientific Management, HarperandBrothers, NY (1919).

6. M. Hammer and J. Champy, Reengineering the Corporation,Nicholas Brealey Publishing, London (1993).

7. G. Nicolis and I. Prigogine,Exploring Complexity,W.H.Free-man and Co., NY (1989).

8. E. N. Lorenz, The Essence of Chaos, University of Washing-ton Press, Seattle, WA (1993).

9. J.H. Holland,Emergence:From Chaosto Order, PerseusPub-lishing, Boulder, CO (1998).

10. S. A. Kauffman, Investigations, Oxford University Press, Ox-ford (2000).11. C. W. Reynolds, Flocks, Herds, and Schools: A Distributed

Behavioral Model, Computer Graphics 21, No. 4 (SIG-GRAPH 87 Conference Proceedings), 2534 (1987).

12. S. Camazine, J. Deneubourg, H. R. Franks,J. Sneyd, G. Ther-aula, and E. Bonabeau, Self-Organization in Biological Sys-tems. Princeton University Press, Princeton, NJ (2001).

13. P. Ball, The Self-Made Tapestry: Pattern Formation in Nature,Oxford University Press, Oxford (1999).

14. J. Briggs andF. D.Peat, TurbulentMirror:An Illustrated Guideto Chaos Theory and the Science of Wholeness , Harper andRow Publishers, Inc., NY (1989).

15. S. Johnson, Emergence: The Connected Lives of Ants, Brains,Cities, and Software, Scribner Book Co. (2001).

16. H. Arrow, J. E. McGrath, and J. L. Berdahl, Small Groups

As Complex Systems: Formation, Coordination, Development, and Adaptation, Sage Publications, Thousand Oaks, CA(2000).

17. R. Axelrod and M. D. Cohen, Harnessing Complexity: Orga-nizational Implications of a Scientific Frontier, Free Press,NY(1999).

18. R. D. Stacey, Complex Responsive Processes in Organizations:Learning and Knowledge Creation, Routledge, London (2001).

19. P. Cilliers, Complexity and Postmodernism, Routledge, Lon-don (1998).

20. D. Zohar,and I. Marshall, QuantumSelf: Human Nature andConsciousness Defined by theNew Physics, Quill/William Mor-row, NY (1991).

21. R. Lewin, Complexity: Life at the Edge of Chaos, Universityof Chicago Press, Chicago, IL (1999).

22. G. Weiss, Editor, Multiagent Systems: A Modern Approach to

Distributed Artificial Intelligence, MIT Press,Cambridge,MA(1999).

23. M. W. Macy, Social order in artificial worlds, Journal ofArtificialSocieties and Social Simulation1, No. 1 (1998). Avail-able at http://www.soc.surrey.ac.uk/JASSS/1/1/4.html

24. R. Axelrod, The Complexity of Cooperation, Princeton Uni-versity Press, Princeton, NJ (1997).

25. R. Dawkins, The Selfish Gene, Oxford University Press, Ox-ford (1989).

26. M. Prietula, K. Carley, and L. Gasser, Simulating Organiza-tions: Computational Models of Institutions and Groups, MITPress, Cambridge, MA (1998).

27. T. Schmalberger,Dangerous Liaisons: A Theoryof Threat Re- lationships in International Politics, Ph.D. dissertation, Insti-tut Universitaire de Hautes Etudes Internationales (1998).



22/22

Availableat http://www.usc.edu/dept/LAS/ir/cis/cews/html_pages/Dangerous_Liaisons

28. P. Baumgartner and S. Payr, Speaking Minds: Interviews withTwenty Eminent Cognitive Scientists, Princeton UniversityPress, Princeton, NJ (1995).

29. D. C. Engelbart, Augmenting Human Intellect: A ConceptualFramework, Summary Report AFOSR-3223 under ContractAF 49(638)-1024, SRI Project 3578 for Air Force Office ofScientific Research,Stanford Research Institute,MenloPark,CA (1962).

30. P. Skagestad, Thinking With Machines: Intelligence Aug-mentation,Evolutionary Epistemology, and Semiotic,Jour-

nal of Social and Evolutionary Systems 16, No. 1, 157180(1993).

31. R. D. Cureton, e.e. Cummings: A Study of the Poetic Useof Deviant Morphology, Poetics Today 1, No. 2, 213 44(1979).

32. S. Kostof, The City Shaped: Urban Patterns and Meaningsthrough History, Bulfinch Press, Boston, MA (1991).33. L. Bolmanand T. Deal,Reframing Organizations, Jossey-Bass

Publishers, San Francisco, CA (1997).34. D. Snowden, Liberating Knowledge, in Liberating Knowl-

edge CBI Business Guide, Caspian Publishing, pp. 9 19 (Oc-tober 1999).

35. T. A. Stewart, How to Think with Your Gut, Business 2.0(November 2001). Available at http://www.business2.com/archives/mag/0,1059,317,00.html.

36. M. Douglas, Purity and Danger, Routledge, London (1966).37. A.Clark,Being There:Putting Brain, Body andWorldTogether

Again, MIT Press, Cambridge, MA (1997).38. T.Sheaand D.Mavis,Editors,Managing the Unexpected: Four

Case-Study Reports, U.S. Medicine Institutefor HealthStud-ies,ExecutiveLeadership Program Report (2002). Available

at http://www.usminstitute.org/leader.html.39. D. Snowden, Complex Acts of Knowing: Paradox and De-

scriptive Self Awareness,Journal of Knowledge Management6, No. 2 (May 2002).

40. D. Snowden, The New Simplicity: Context, Narrative andContent,Journal of Knowledge Management5,No.10,1115(July August 2002).

41. D. Snowden, The Art and Science of Story or Are you sit-ting uncomfortably? Part 2: The Weft and the Warp of Pur-poseful Story in Business, Information Review 17, No. 4(2000).

42. H. Beyer andK. Holtzblatt, Contextual Design: Defining Cus-tomer-Centered Systems, Morgan Kaufmann Publishers, SanFrancisco, CA (1998).

43. D. Snowden, Just in Time Knowledge Management, KMReview 5, No. 5, 14 17 (November/December 2002), and 5,

No. 6, 24 27 (January/February 2003).44. M. Boisot, Knowledge Assets, Oxford University Press, Ox-

ford (1998).45. E. Keaton and J. Vance, Buster Keaton Remembered, H. N.

Abrams, Inc., NY (2001).46. P. Schwartz, The Art of the Long View: Planningfor the Future

in an Uncertain World, Doubleday, NY (1991).47. A. Kleiner, The Age of Heretics, Doubleday (1996).48. M. de Landa, A Thousand Years of Nonlinear History, Zone

Books, NY (1997).49. M. Baranger, Chaos, Complexity, and Entropy: a Physics Talk

for Non-Physicists, Wesleyan University Physics Dept. Col-loquium (2001). Available at http://www.necsi.org/projects/baranger/cce.html.

50. J. Horgan, The End of Science: Facing the Limits of Knowl-

edge in the Twilight of theScientificAge, BroadwayBooks, NY(1997).

51. H. Courtney,20/20 Hindsight, HarvardBusiness School Press,Cambridge, MA (2001).

52. A. K. Ramanujan, Folktales from India: A Selection of OralTales from Twenty-two Languages, Pantheon Books, NY(1991). (The referenced folk tale is The Boy Who SoldWis-dom, pp. 240 243.)

53. G. Allison and P. Zelikow, Essence of Decision: Explainingthe Cuban Missile Crisis, (2nd ed.), Pearson Education, Up-per Saddle River, NJ (1999).

Accepted for publication April 24, 2003

Cynthia F. KurtzIBM Global Services Lotus Development, 1 Rog-ers St., Cambridge, MA 02142 ([email protected]). Ms. Kurtz isPrincipal Researcherfor IBMs Cynefin Centre fororganizational

complexity. Shereceivedher B.S. degreein biology from ClarionUniversity of Pennsylvania in 1986 and her M.A. degree in ecol-ogyand evolutionfromthe State University of NewYorkat StonyBrook in 1990 (after completing five years of Ph.D.-level grad-uate work studying theevolution andcomplexity of socialbehav-ior). From 1992 to 1998 she co-owned a small software business,and is the co-creator of two internationally distributed and widelyreferencedenvironmental education simulatorson plantgrowth,soil science and evolutionary design. In 1999, she joined theKnowledge Socialization Group at the Thomas J. Watson Re-search Center, where she carried out pioneering work on nar-rativein organizations, includingthe creation of an XMLschemafor describing stories and storytelling contexts. She moved to theInstitute for Knowledge Management in 2001 to work on bothnarrative programs and complexity programs beforehelping foundthe Cynefin Centre in 2002. She is currently leading work on the

use of narrative and complexity-informed techniques for gainingmulticultural perspective in decision environments, and is mainlyresponsible for the development of narrative database technol-ogy.

David J. SnowdenIBM Global Services, 79 Staines Road West,Sunbury-on-Thames, TW16 7AN United Kingdom ([email protected]). Mr.Snowden is thedirectorof IBMs Cynefin Cen-tre shortly to be based in Cardiff University, Wales. He receivedhis B.A. degree in philosophy from the University of Lancasterin 1975 and anM.B.A.degree from Middlesex University in 1986.His early career was in the specification of decision support sys-tems for international companies, progressing to general man-agement and strategic marketing positions in Datascience priorto its acquisition by IBM in 1996. He was a director in IBM s In-stitute for Knowledge Management before founding the Cynefin

Centre in 2002. He is an adjunct professor of knowledge man-agement at the University of Canberra, an honorary fellow inknowledge management at the University of Warwick, and hasreceived several awards for his work in knowledge management,including one from the Academy of Management in 2001 for hisoriginal contributions to the field.

sense-making in a complex and complicated world

Documents