SnowdenComplityPDFcopy.doc © D J Snowden 2005 1 of 11
Created 31-Oct-04 2-Mar-05
Multi-ontology sense making
a new simplicity in decision making
David J Snowden
Director, Cynefin Centre for Organizational Complexity
Management Today, Yearbook 2005, Vol 20 No
Editor Richard Havenga email@example.com
Bus. Tel: +27 (0)11 326-2625
EDITORIAL NOTE: INSET FOLLOWING TEXT IN SIDE BAR OR BOX on PAGE ONE
Imagine organising a birthday party for a group of young children. Would you agree a set of
learning objectives with their parents in advance of the party? Would those objectives be
aligned with the mission statement for education in the society to which you belong? Would
you create a project plan for the party with clear milestones associated with empirical
measures of achievement? Would you start the party with a motivational video so that the
children did not waste time in play not aligned with the learning objectives? Would you use
PowerPoint to demonstrate to the children that their pocket money is linked to achievement of
the empirical measures at each milestone? Would you conduct an after action review at the
end of the party, update your best practice database and revise standard operation
procedures for party management?
No, instead like most parents you would create barriers to prevent certain types of behaviour,
you would use attractors (party games, a football, a videotape) to encourage the formation of
beneficial largely self organising identities; you would disrupt negative patterns early, to
prevent the party becoming chaotic, or necessitating the draconian imposition of authority.
At the end of the party you would know whether it had been a success, but you could
notdefine (in other than the most general terms) what that success would look like in
From The Cynefin Manifesto www.cynefin.net
The purpose of this article is to introduce a new simplicity into acts of decision-making and
intervention design in organizations. That may seem ironic given the title, with its use of the
terms “ontology” and “sense-making” which may be unfamiliar to readers; but new ideas
often need new or at least unfamiliar language and I make no apology for that, although some
readers may wish to skip the remainder of this introduction which may only be relevant to
academics wishing to situate my language. New language aside, the basic principles that
underlie this paper are very easy to understand and are illustrated by the inset example of the
children’s party. Multi-ontology sense making is about understanding when to use both
methods of management outlined in the story, both the structured and ordered approach based
on planned outcomes and the un-ordered, emergent approach focused on starting conditions
expressed as barriers, attractors and identities.
is derived from the Greek word for being and it is the branch of metaphysics which
concerns itself with the nature of things. In this article I am using it to identity different types
of system, and will later discuss two contrasting types of ontology (order and unorder) each of
which requires a different approach to both diagnosis and intervention. In practice we need to
consider three physical and five human ontologies. The three physical ontologies are order,
complexity and chaos; in human systems order divides into visible and hidden forms and we
add a fifth state of disorder. These are more fully described elsewhere (Kurtz & Snowden
SnowdenComplityPDFcopy.doc © D J Snowden 2005 2 of 11
Created 31-Oct-04 2-Mar-05
2003). For this article I will combine complex and chaotic into a single category of unorder
and ignore disorder.
Sense-making is most commonly associated with the Weick (1995) and Dervin (1998) and is
starting to gain more attention in management circles. I am closer to Dervin than Weick, and
in the context of this paper I am talking about sense making as the way that humans choose
between multiple possible explanations of sensory and other input as they seek to conform the
phenomenological with the real in order to act in such a way as to determine or respond to the
world around them. Multi-ontology sense making is thus a means to achieve a requisite level
of diversity in both the ways we interpret the world and the way we act in it. Requisite
diversity means ensuring the acceptance of a sufficient level of divergence to enable the
sensing of weak signals (terrorist threat or market opportunity) and avoidance of the all too
common pattern entrainment of past success, while maintaining a sufficient focus to enable
decisive and appropriate action. Above all it is about ensuring cognitive effectiveness in
information processing and this gaining cognitive edge, or advantage.
The ideas and concepts may be novel and even threatening to a generation of managers, civil
servants and academics who have been trained in what I will later define as single-ontology
sense making. The dominant ideology of management inherits from Taylor (1911) a view of
the organization based on the necessity and the probity of order. In this world things are
deemed to be known or knowable through proper investigation and relationships between
cause and effect once discovered repeat. It is the world of the mechanical metaphors of
Taylor and most management theorists who came afterwards; it is the Newtonian universe of
predictable relationships between cause and effect which can be calculated; the world of the
five year plan and the explicit performance target; of hypothesis and empirical proof through
observation and explanation of events in retrospect. This paper challenges that particular
weltanschauung not by denial, but through bounding and limiting its applicability.
The fad cycle in m anagement theory and practice
The fad cycle in m anagement theory and practice
Single ontology sense making itself in a by now all too familiar pattern and can be
summarized and to a degree satirized as follows:
An Academic group studies a range of organizations to identify causal linkages
between things those organizations do and results that they achieve or fail to achieve,
from which they derive hypothesis that forms a definition of best practice. A popular
management book then follows and a new “fad” is born.
Consultants and IT providers
produce industrial strength recipes based on the new
idea, ideally involving a consultancy process, followed by a technology
implementation and some for of organizational change or cultural alignment with the
programme to orientate employees to the new goals
Managers go through a process based on the recipe to determine a desired end state
defined in terms of economic performance, behavior characteristics etc. They then
determine the current state and identity a series of process steps to achieve that goal
and roll out the programme ingpromised substantial improvements as a result to their
stake holders many of whom in the “employee” category are already suffering from
substantial initiative fatigue. Some years after the fad has run its course in industry
and the limitations the consultants find a lucrative secondary market in applying
“industrial best practice” to government clients.
It is not my intention to argue against management fads per se; indeed the different
perspectives and novelty that they introduce can be valuable even though in the main they are
based on an out of date understanding of science in the context of management and
organizations. Equally I am not denying that substantial benefits have been achieved over the
SnowdenComplityPDFcopy.doc © D J Snowden 2005 3 of 11
Created 31-Oct-04 2-Mar-05
years through these new methods and ideas, although the benefits are often over claimed or
not sustained beyond their initial novelty value impact.
That said, it is my contention that the vast majority of these methods have been simplistic in
their conception and execution; in particular when they claim universality of application. I
am using simplistic here in a negative sense in contrast to simple and I will argue later that
simplicity can lead to complex solutions, while being simplistic leads to over-complication.
To illustrate this, one of the dominant fads of the last fifty years, namely Business Process
Engineering (Hammer & Champy 1993) emerged, as do many methods from manufacturing,
shifting from a horizontal focus on produce in contrast with the previous organization by
functional silos. As such it worked well, but then it was over extended beyond its valid
ontological boundary (I will define this term later) to more human less mechanical aspects of
the service sector at which point it started to fail. A similar point can be made in respect of
the Learning Organization (Senge 1990), Emotional Intelligence, Knowledge Management
and many others. Multi-ontology sense making argues that different approaches are
legitimate, but within boundaries and that methods and tools that work in one ontology, do
not work in another. It is thus behoven on management to know which ontological domain
they are operating in, and what transitions between domains they wish to achieve.
So, what are the boundaries that exist that legitimize or invalidate methods? To demonstrate
this I intend to usea categorization model “The landscape of management” which is designed
to position the various types of management theory that have evolved over the last century
and discuss some of the implications that arise from that model.
This paper will not cover the sense making framework “The Cynefin Model” that seeks to
provide a mechanism for managers to determine the boundaries between ontologies, and the
dynamics of cross boundary movement between ontologies. Readers interested in that model
should look to two other papers: Kurtz & Snowden 2003 and Snowden 2004 both of which
can be obtained from www.cynefin.net.
The landscape of management
The landscape of management
The two by two matrix set out here contrasts the nature of systems (ontology) with the nature
of the way we know things (epistemology) and accordingly the way we act; I contend that
knowledge and action are intimately intertwined (Snowden 2002). The matrix was originally
produced from a EU Study on knowledge management and was used to demonstrate that the
strategic advantage for Europe
(and I would content for Africa
and Asia) lies not in imitating
the USA, but in utilizing its
multi-culturalism as a
competitive advantage through
exploitation of social
complexity in which it currently
has a intellectual lead. I have
more fully described the matrix
and some its implications
elsewhere (Snowden 2003) and
(Stanbridge & Snowden 2004)
and the full report can be
obtained from www.cynefin.net
and is recommended to readers.
The vertical dimension of the matrix contrasts two types of system, namely order and un-
order. In the earlier story of the childrens’ party the first approach, namely that of objectives,
planning and best practice is in effect an illustration of the type of approach that is typically
SnowdenComplityPDFcopy.doc © D J Snowden 2005 4 of 11
Created 31-Oct-04 2-Mar-05
adopted in an ordered system and it can be legitimate. Where there are clearly identified (or
identifiable) relationships between cause and effect, which once discovered will enable us
control the future, then the system is ordered. It can be structured on the basis of a desired
outcome with structured stages between where I am “now” and where I want to be “then”.
This is contrasted with un-order in which the relationships between cause and effect do not
repeat, except by accident and in which the number of agents interacting with other agents is
too great to permit predictable outcome based models, although we can (as is the case with
the party) control starting conditions and monitor for emergence. “Un” is used here in the
sense that mBran Stoker uses it of Dracula: the un-dead are neither dead not alive, they are
something different that we do not fully understand or comprehend.
At its simplest the difference between management in order and un-order can be summarized
as follows. Ordered systems are those in which a desired output can be determined in
advance and achieved through the application of planning based on a foundation of good data
capture and analysis. In un-ordered systems no output be determined in advance, in other
than the most general terms but we can manage the starting conditions and may achieve
unexpected and more desirable goals that we could have imagined in advance, or (and this is
commonly the case especially in the case of teenage parties) we can just be more successful
in avoiding failure.
While the vertical dimension represents two distinct states, the horizontal dimension is more
of a continuum between the low ambiguity of rules that can easily be made explicit and the
more ambiguous use of heuristics or rules of thumb which provide guiding principles but have
high levels of ambiguity. I sometimes illustrate this difference by comparing a complex US
government manual on procurement (anyone who has contracting under US Government
rules can take you to highly complicated web sites which prescribe all possible circumstances
on the basis that anything which is not explicitly permitted is not allowed) with a mission or
value statement for an organization which states broad principles that set expectations, can be
comprehending quickly and are easily memorable; as a result of which they can be applied
without reference to the rules.
Having established the dimensions, we can use the model to look at the current situation in
respect of management theory by taking each of the quadrants in turn, and in doing so look at
the limits on their applicability.
Ordered ontology, rule based epistemology
In effect the last century of management theory and practice, from Taylor’s Scientific
management (1911) to its logical extreme Business Process reengineering most commonly
attributed to Hammer and Champy (1993). A strong mechanical metaphor characterizes these
approaches. The focus is on efficiency, stripping away all superfluous functions in order to
ensure repeatability and consistency. The most recent manifestation is Six-Sigma
in GE which continues the focus on efficiency with a strongly quantitative approach to
measurement although with some cult like overtones in its imagery: black belts etc.
It is a common characteristic of engineering approaches that they start in manufacturing
processes where they gain their initial success and then extend to other less structured aspects
of an organization at which point problems start to emerge. One of the reasons for this is the
important difference between a focus on efficiency and one on effectiveness. The engineering
process takes place in a specific context and once achieved, shifts in that context require the
engineering design process to be repeated to some degree before efficiency can be achieved
again. Radical shifts in context may make the entire approach redundant or lead to
catastrophic failure. In the context of a manufacturing plan or a stable industrial sector this is
not a problem, or if it is a problem it is shared by all of our competitors. We have to make
major investments in process to achieve efficiency and that investment is always a sunk cost.
Manufacturing plant, payment systems in a bank and the like are all closed systems that can
SnowdenComplityPDFcopy.doc © D J Snowden 2005 5 of 11
Created 31-Oct-04 2-Mar-05
be structured and standardized without any major issue. We can in effect define best practice.
However when we apply the same techniques to systems with higher levels of ambiguity, for
example customer interactions, sales processes and the like we encounter more difficulties.
Some of these arise from the fact that significant aspects of what we know cannot be
measured or made explicit: we always know more than we can say, we will always say more
than we can write down. Others arise from the impossibility of anticipating all possible
situations and shifting context. In these cases we need a different focus, one of effectiveness
in which we leave in place a degree of inefficiency to ensure that the system has adaptive
capacity and can therefore rapidly evolve to meet the new circumstances. Examples would
include apprentice schemes of knowledge transfer, maintaining mavericks or misfits, allowing
people to take training in subjects with no apparent relevance to their current jobs and
providing more delegated authority.
Boisot (1998) makes the valid point that organizations who invest heavily in knowledge
creation tend to assume that the same knowledge will require similar costs for their
competitors and uthis focus a massive effort on protection through patents etc. He calls these
N-Learning organizations in contrast with S-Learning cultures who see value arising from the
exploitation of knowledge, not its possession and thus tend to share and collaborate even with
competitors. The open source movement is a good illustration of the latter. Boisot goes on to
demonstrate through several examples the way in which N-Learning cultures fail to adapt to
changing circumstances; these include IBM’s failure to see the change to micro computers
until it was almost too late, and the failure to understand the operating systems market to the
point where they lost it to Microsoft. There are many other examples in other large
companies who have tended to adopt engineering processes and built large bureaucracies and
There is nothing wrong with an engineering approach; there are many things that need high
degrees of order and control. However taken to excess, and it has nearly always been so
taken, it sacrifices human effectiveness, innovation and curiosity on the altar of mechanical
Ordered ontology, heuristic based epistemology
Towards the end of the last century we saw some rebellion against the mechanic metaphors of
scientific management and its successors. Tom Peters in various speeches and books, Senge
(1990) with Learning Organization, Nonaka with various co-authors in books and articles
covering Knowledge Management represent the more popular examples. Systems thinking
challenges the apparent simplicity of process based approaches and their associated
mechanical metaphor arguing for both non-linear relationships between cause and effect and
the greater ambiguity of human systems. We see the birth here of approaches based on
articulating mission statements, establishing value systems and idealized behavior all of
which would then be mandated for employees. Senge argues that employees should sacrifice
their individual objects and goals to gain from the assumption of a common identity in the
organization to which they belong; Nonaka recognizes the social nature of human knowledge
transfer and the need to separate tacit knowledge sharing from the process of making tacit
knowledge explicit; Peters emphasizes motivation and leadership. The ambiguity of human
systems is recognized, but the basic concept of central control or planning remains at the
heart. Leaders set objectives, they (to use Senge’s metaphor) are the designers of the ship.
Competences and behaviors it is argued can be taught and learnt and alignment of the
individual with the collective is thus possible.
One of the easiest ways to identify a systems thinking approach is to look for the process
models – those with lots of boxes, arrows and feedback loops are generally characteristic of
systems thinking. By accepting that the world is more complicated than implied by process
reengineering and the introduction of feed back loops, concepts such as double loop learning
and discourse analysis systems thinking humanized the heirs of Taylor to a degree. However
SnowdenComplityPDFcopy.doc © D J Snowden 2005 6 of 11
Created 31-Oct-04 2-Mar-05
the basic assumptions of order pertain. Systems are configured based on end point objectives;
humans are seen as assets or capabilities that can be aligned with those objectives.
Reductionism still stands; think of the balanced score card, another popular manifestation of
systems thinking philosophy in which the range of activities of an organization are reduced to
a set of interconnected measurable items in which the whole is assumed to be the sum of the
parts. The strength of systems thinking is its recognition that human systems are messy, they
frequently need focus and alignment; its weakness is that it assumes that the design of that
focus and alignment is a top dsown objective based process.
Like Process Engineering, Systems Thinking is strongly linked to computer based automation
and modeling. The speed of computers allows the complication of systems thinking models
to be calculated on a consistent basis with associated reporting and control mechanisms. For
complicated aspects of an organization it is very powerful, allowing models to be constructed
to enable an understanding of the inter-relationships between people, process and technology
(a three fold focus mantra that typifies thinking in this domain). When the number of people,
the complication and context changes associated with process and the capabilities of
technology exceed a threshold level the system shifts from being complicated to being
complex, from order to un-order in which an output cannot be defined in advance and in
which the sheer number of relationships means that order emerges from the interaction of the
various agents over time, and the nature of that order is unique to each emergence. At this
point we shift to the unordered quadrants.
Unordered ontology, rule based epistemology
“A new awareness of the ancient counterpart to order began over a century ago with Poincaré
and several others, and has surged in recent decades (e.g., Nicolis and Prigogine 1989, Lorenz
1993, Holland 1998, Kaufmann 2000). In fact there is a fascinating kind of order in which no
director or designer is in control but which emerges through the interaction of many entities.
Emergent order has been found in many natural phenomena: bird flocking behaviour can be
simulated on a computer through three simple rules (e.g., Reynolds 1987); termites produce
elegant nests through the operation of simple behaviours triggered by chemical traces (e.g.,
Camazine et al. 2001); each snowflake is a unique pattern arising from the interactions of
water particles during freezing (e.g., Ball 1999). The patterns that form are not controlled by a
directing intelligence; they are self-organizing. The new science of complexity spawned by
these findings is interdisciplinary, touching fields from mathematics to evolution to
economics to meteorology to telecommunications. In the domain of emergent order, a goal
“to predict (and thereby control) the behavior of systems not yet studied (but similar to those
that have been studied) under conditions not yet extant and in time periods not yet
experienced” (Arrow et al 2000) is difficult if not impossible to achieve – but other goals are
Awareness of emergent order has as yet had comparatively little influence on mainstream
theory and practice in management and strategy (for a good introduction see Axelrod and
Cohen 1999) however there are a growing number of examples. Computer based simulations
based on agent models have been used to handle complex issues such as traffic management
and package routing for airlines. A growing use is in economical modeling and clustering.
The procedure here is to look at a population of human agents and identify the rules on which
they make decisions, then produce a computer model in which individual agents make
decisions based on those rules and order emerges as a result.
Note a key difference here with ordered systems, unorder is bottom up; although
mathematical complexity shares the concept of rules with business process re-engineering
along with the associated heavy reliance on computing power, the rules apply to agents from
which behavior emerges, it is not possible to create rules top down for that behavior, but the
rules apply at a lower level of agent behavior.
SnowdenComplityPDFcopy.doc © D J Snowden 2005 7 of 11
Created 31-Oct-04 2-Mar-05
We also see one of the most interesting aspects of a complex system, namely its simplicity; a
few rules give rise to complex forms of order. This leads to a contrast of simple-complex
with simplistic-complicated which while not universally true is a useful way of looking at the
Of course not all systems are unordered, and applying an unordered approach based on agent
simulation would be dangerous if we used it for something like payment systems in a web
based trading system or for the regulatory processes of the pharmaceutical industry.
Accordingly we should resist the Universalist claims of some complexity practitioners as
much as we resist those of engineering and systems thinkers. There is however another
limitation to mathematical complexity namely the fact that other than in a limited set of
circumstances human beings are not the same entities as ants, birds or crystals. I am
sometimes amazed that this point has to be made, but have come to the conclusion that for
many economists and sociologists they would like humans to be ants as then their
mathematical models would work. The differences between human systems and ants, is
similar to the differences between human systems and the mechanical metaphors of process
which gave rise to Systems Thinking; which leads us logically to our final quadrant.
Unordered ontology, heuristic based epistemology
Social complexity shares with mathematical complexity the concept of unorder and
emergence, but also shares with systems thinking the belief that human systems are different;
these differences are summarized in the next section. Social complexity is linked in some
cases to postmodernism (Cilliers 1998) and has some strong advocates in the field such as
Stacy (2001) and Juarrero (1999). It is the main focus of the Cynefin Centre which I founded
(www.cynefin.net) and offers interesting possibilities for the government and industry alike.
The relevance social complexity is illustrated by the metaphor of the children’s party with
which this article stated and which aptly summarizes the differences between and ordered and
unordered approach. The first approach to managing a children’s party is based on the
assumption of order, the second is based on unorder. The argument is not that one or other
approach is absolutely right or wrong, but that both are right (and wrong) in context.
This awareness of context is not common in Management science and consultancy practice
which is dominated by approaches based on an assumption that the systems being researched
and managed are essentially ordered in nature. They are thus susceptible to methods based on
best practice and the creation of structured top down approaches. In ordered systems we can
create repeatability and scalability with consistency. Failure is a failure of design or
implementation not a result of the nature of system itself.
The importance of learning how to manage in unordered environments is easily understood by
looking at the dilemma facing governments around the world. On the one hand they face
increasing requirements for the provision of public services, but on the other they have static
or declining levels of resource. Managing unorder through the manipulation of boundaries,
attractors and identity offers a potential path to the resolution of that dilemma; managing
unorder on the basis of methods and tools appropriate for ordered systems requires ment
ofdeploying major resources and the likelihood of making things worse. The same
dilemma & opportunity exists not-for-profit and commercial organizations alike.
Unique aspects of human systems
Unique aspects of human systems
Different schools of thought identify different distinguishing features of human systems. The
following summary has been developed from various sources over the years in the context of
creating explainable and comprehendible reasons for management audiences engaged in the
early stages of applying thinking from social complexity
SnowdenComplityPDFcopy.doc © D J Snowden 2005 8 of 11
Created 31-Oct-04 2-Mar-05
Humans make decisions based on patterns
This builds on naturalistic decision theory in particular the experimental and observational
work of Gary Klein (91944) now validated by neuro-science, that the basis of human decision
is a first fit pattern matching with past experience or extrapolated possible experience.
Humans see the world both visually and conceptually as a series of spot observations and they
fill in the gaps from previous experience, either personal or narrative in nature. Interviewed
they will rationalize the decision in whatever is acceptable to the society to which they
belong: “a tree spirit spoke to me” and “I made a rational decision having considered all the
available facts” have the same relationship to reality.
Accordingly in other than a constrained set of circumstances there are no rules to model.
Humans create and maintain multiple identities
An individual can be distinguished by their roles; clans or context. We both create and
maintain multiple often parallel identities shifting between and amongst them as needed
without so much as a second thought. As a male individual I can be father, brother, son or
husband, I can switch between work based identities or home based ones. My employees if
distanced from me may never associate my person with the role I occupy. I am a member of
many clans, from sporting clubs, cohort groups, participants in a senior executive programme:
there are many examples. Context is of particular interest here, working as a crew in a bush
fire by identity is very strongly associated with the role and common threat and I can sustain
it for a period of time while I am “on watch”; however such a contextual identity and the
behaviors associated with it cannot be transferred outside of the context.
Accordingly in other than a constrained set of circumstances there are no clear agents to be
Humans ascribe intentionality and cause where none necessarily exist
There is a natural tendency to ascribe intentionality to behavior in others, whilst assuming that
the same others will appreciate that some action on our part was accidental. Equally if a
particular accidental or serendipitous set of actions on our part lead to beneficial results we
have a natural tendency to ascribe them to intentional behavior and come to believe that
because there were good results, those results arose from meritorious action on our part. In
doing so we are seeking to identify causality for current events. This is a natural tendency in
a community entrained in its patter of thinking by the enlightenment. Deacon
established that the concept of co-evolution of the brain and language removes the need for a
“universal grammar” as an explanation of language and a similar application of Ockham’s
razor can remove much of the supposed causality in both government and industry. One of
the key insights of social complexity is that some things just “are” by virtue of multiple
interactions over time and the concept of a single explanation, ascription of blame or for that
matter credit are not necessary.
Humans have learnt how to structure their social interactions to create order
For the purpose of this article we will avoid the potentially troublesome concept of free will
and instead focus on the ability of humans through social structures and less tangible things
such as myth, ritual and taboo to create stability and predictability in their systems.
Depending on where you live it is correct to drive on either the left or right hand side of the
road, we have advanced from the adaptive nature of bird flocking behavior (fly to the centre
of the flock, avoid collision, match speed) as a means of managing traffic to create a
predictable form of order that not only provided stability in our day to day lives, but also
allows planning for road design etc. This is linked to the human capacity to store knowledge
in the external environment, or “scaffolding” to use Clarke’s
(1997) term. Humans have thus
learnt how to move between order and unorder.
SnowdenComplityPDFcopy.doc © D J Snowden 2005 9 of 11
Created 31-Oct-04 2-Mar-05
Couple the above differences with the phenomenological aspects of human perceptions of
reality and we see that there are substantial and major differences between human and non-
human unorder. This led the Cynefin Centre to coin the term “contextual complexity” in
contrast with “participative complexity” from Stacy (2001) and colleagues. Contextual
complexity argues that humans have the ability to operate in all quadrants of the model and
the ability to move between them as a result of both accidental and deliberate action.
Multi-ontology sense making then reflects the need to adopt different diagnostic techniques,
different intervention devices and different forms of measurement depending on the
ontological state. This is contrasted with any single ontology form of sense making whether
based on order, complexity or chaos. Understanding this concept of ontological switches also
helps prevent the degeneration into “un-manageability” and fatalism which can occur when
people start to understand complexity based thinking.
The order and unorder distinction has many applications and these are summarized in the
Focus on rational individuals making choices
based on personal self interest
Focus on identities making decisions based
on patters arising from personal experience
and collective knowledge expressed in
Manage to achieve goals based on ideal
models and central planning
Manage starting conditions and monitor for
the emergence of pattern to sustain or disrupt
Simple – complex
Efficiency (focus on core capability,
outsource the rest)
Effectiveness (requisite diversity, allow
inefficiency for adaptability)
Resilience and adaptability
Reductionist measures: ROI, balanced score
Indivisible, emergent measures
Measure outcomes based on explicit goal
Measure the stability of barriers, the
attractiveness of attractors and the stability of
Dichotomy and the resolution of dilemmas as
an either or choice
Dialectic and the resolution of paradox to see
the world in a different way
Analysis and Expert interpretation
Stimulated emergence so that the patterns of
possibility become more visible.
Economic example – credit scoring
Economic example – micro lending
The above summarizes material which has already been explained, or implied in the text
above. Some examples are more enigmatic, such as those on measurement and are covered in
referenced articles. The final economic example deserves more explanation and also allows
as simple case to form the conclusion of this article and exemplar of an unordered
intervention. The case is drawn from Axelrod and Cohen’s (1999) introductory text on
SnowdenComplityPDFcopy.doc © D J Snowden 2005 10 of 11
Created 31-Oct-04 2-Mar-05
This is the Grameen Bank (Yunas 1999) which was created in Bangladesh to provide small
loans to poor people. The name Grameen comes from the Bangla word for village. This is a
market which the conventional banking system finds unattractive. Most commercial and
private loans are based on credit scoring, an ordered concept in which the characteristics of
good and bad debtors are identified and used as predictors and therefore controls for future
lending. This increases the cost of lending as the various processes have to be administered,
and small loans this become uneconomic. In the Grameen Bank everyone who took out a
loan was required to be a part of a self regulating borrowers’ group in which each member of
the group had to take responsibility for the debts of the others. This simple rule which costs
little to administer produced a 97 percentage repayment rate comparable with best
achievements of the large banks; there are now over two million clients of the Grameen bank
and the approach has proved both scalable and portable.
I find the Grameen Bank an inspiring case, and an illustration of the great benefits that
complex or unordered thinking can bring. Just as in the case of the children’s party,
managing the starting conditions not an idealized end state can produce lower cost more
effective solutions. Complex thinking is not a nice to have in modern management, it is a
fundamental necessity and in the history of management science is another “Taylor” bringing
a new science to bear for the first time. It is a new and exciting way of thinking about the
world that, properly understood, does not mean that we abandon any of the ways we currently
manage, but instead understand and apply the boundaries of their applicability. With that
change we enter a new simplicity in management decision making.
Arrow, H., McGrath, J. E., & Berdahl, J. L. (2000). Small Groups As Complex Systems:
Formation, Coordination, Development, and Adaptation. Sage Publications.
Axelrod, R. and Cohen, M.D. (1999) Harnessing Complexity: Organizational Implications of
a Scientific Frontier. Free Press.
Ball, P. (1999) The Self-Made Tapestry: Pattern Formation in Nature. Oxford University
Boisot, M. (1998) Knowledge Assets Oxford University Press
Camazine, S. et al. (2001) Self-Organization in Biological Systems. Princeton University
Cilliers, P (1998) Complexity and Postmodernism. Routledge.
Clark, A. (1997) Being There: Putting Brain, Body, and the World Together Again MIT
Deacon, Terrence (1997) The Symbolic Species Penguin
Dervin, B. “Sense Making theory and practice: an overview of user interests in knowledge
seeking and use” in Knowledge Management 2:2 MCB University Press December 1998
Hammer, M., and Champy, J. (1993) Reengineering the Corporation Nicholas Brealey
Holland, J.H. (1998). Emergence: From Chaos to Order. Perseus Books.
Juarrero, A (1999) Dynamics in Action: Intentional Behavior as a Complex System. MIT
Kauffman, S.A. (2000) Investigations. Oxford University Press
SnowdenComplityPDFcopy.doc © D J Snowden 2005 11 of 11
Created 31-Oct-04 2-Mar-05
Klein, G. (1994) Sources of power: how people make decisions” MIT
Kurtz, C & Snowden, D (2003) “The New Dynamics of Strategy: sense making in a complex-
complicated world” in IBM Systems Journal” Volume 42 Number 3 pp 462-483
Lorenz, E.N. (1993). The Essence of Chaos. University of Washington Press.
Nicolis, G. and Prigogine, I. (1989). Exploring Complexity. W.H. Freeman.
Reynolds, C. W. (1987) Flocks, Herds, and Schools: A Distributed Behavioral Model, in
Computer Graphics, 21(4) (SIGGRAPH '87 Conference Proceedings) pages 25-34.
Senge, P. (1990) The Fifth Discipline: The Art & Practice of the Learning Organization.
New York: Doubleday Currency
Snowden, D. (2002) “Complex Acts of Knowing: Paradox and Descriptive Self Awareness”
in the Journal of Knowledge Management – Vol. 6, No. 2, (May) pp. 100-111
Snowden, D. (2003) “Innovation as an object of knowledge management: Part 1 The
Landscape of Management” in Knowledge Management Research and Practice (2003) 00, 1-
Snowden, D (2004) “Strategy in the context of Uncertainty” in Patricia Coate (ed) The
Handbook of Business Strategy Emerald Group Publishing
Stacey, R.D. (2001) Complex Responsive Processes in Organizations: Learning and
Knowledge Creation. Routledge.
Stanbridge, P and Snowden D 2004 “The Landscape of Management
creating the context for understanding social complexity” to be published in Emergence,
Taylor, F.W. (1911) Principles of Scientific Management. Various publishers.
Weick, Karl E. Sensemaking in organizations Sage 1995
Yunas, Muhammad (1999) “The Grameen Bank” Scientific American 281 November 1999
Ontology is commonly misused in the IT profession as an elevated version of taxonomy and
is in fact closer to onomastics than it is to ontology
Consultants and IT Vendors are becoming increasingly interdependent and often identical. One can
trace the large growth of management consultancy to the advent of Business Process Reengineering as
a management philosophy and the development of enterprise wide software solutions such as SAP.
Indeed the financial model of the large consultancy firms is increasingly dependent on large scale
technology implementation with associated programmes for design, cultural change etc.
Six Sigma shares some aspects with Systems Thinking and is not solely confined to Business Process.
The quoted paragraphs that start this section are extracted from the previously cited Kurtz and
This section is largely extracted from the Stanbridge and Snowden article previously referenced and
published in Emergence – probably the international journal of social complexity