Managing Complex Adaptive Networks

Paper for the International Conference on Intellectual Capital and
Knowledge Management, Cape Town Oct 2007.

Roy Williams

This paper is both theoretical and practical, and addresses current issues of social networking and global cultures. It is practical, in that it sets itself the task of creating a framework for designing and managing self-organising spaces for the generation, exchange, networking, and management of knowledge. It is theoretical in that it goes about this task by examining the theories of ecological ‘affordances’, actor-network theory, and complexity theory. Social networking, supported by social software, is key to the ‘micro-global’ social ecology within which self-organising behaviour takes place. It is paradoxical in that managing ecologies, and particularly managing self-organising networks, is somewhat of an oxymoron.

Keywords: complex adaptive networks, actor-networks, complex adaptive systems,

0. Introduction
Self organising behaviour is a property of Complex Adaptive Systems, or ecologies that evolve and display new, emergent properties, self-organising behaviour of their components, and are based on a reasonably stable infrastructure, the satisfaction of the most basic needs, and flexible, frequent, and open communication and interaction. Complex Adaptive Systems may be based on a few, simple rules, but can yield complex and unpredictable outcomes. The ‘Hole in the Wall’ project (Mitra 2003) is an interesting case in point, in the design of spaces for complex adaptive networks: touch screen computers were put in holes in walls in places where unschooled children congregated, and they were given no instructions on how to use the computers, or what to do with them, with startling results that they achieved by exploiting quite surprising affordances (e.g. using animated multi-media programmes on anatomy to teach themselves English, almost from scratch).

Complex systems may also display de-volution, in addition to e-volution, i.e. returning to some previous characteristics and affordances, if the full ensemble of capabilities is no longer essential. Our lack of the ability to produce our own vitamin C is a case in point, and the fairly recent habit of embarking on extended ocean journeys highlighted the cost of that particular ‘devolution’ rather painfully (Deacon 2004). The insights of devolution highlight the fact that ‘evolution’, or the emergence of new forms of life to adapt to new changes in the ecology does not inexorably lead to an increase in complexity. Complex adaptive systems are essentially adaptive, rather than more and more complex: traffic roundabouts (or ‘circles’), as alternatives to traffic lights (or ‘robots’) are a case in point. If the world is becoming more and more complicated, the emergence of global digital networks might be the event horizon that provides both the appearance of seemingly insurmountable complexity, and the necessary infrastructure for the complex adaptive networks that can assuage that problem.

Cetina goes so far as to say that institutional ‘lightness’ is one of four key characteristics in the case of the recently emerged complex global microstructures. And by ‘lightness’ she means “the mechanisms and structures involved suggest a reversal [a ‘devolution’] of the historical trend towards formal, rationalised (bureaucratic organisational) structures … and appear to facilitate a certain non-Weberian effectiveness [which] relies to a far greater extent than hitherto on the systematic and reflexive use of systems of amplification and augmentation [which] seek to exploit the potential for disproportionalities between input and output or effort and effect” (2005:215-216)

This paper will examine uncertainty and complexity in current social and systems theories, along with the affordances of social networking and social software, to see whether it is possible to integrate them into a theory of complex adaptive networks, which we can use to achieve the rather ambitious goal of designing and managing such networks, including networks of knowledge and learning. In the process we might usefully shift the emphasis from the rather modernist and dated ‘ICT’: ‘Information and Communication Technologies’, to ‘ICS’: ‘Interactive and Communicative Spaces’, or even ‘Interactive Complex Spaces’.

The term affordance will be used in Gibson’s ecological sense (1979) i.e. an affordance is the product of the interaction between the (social) individual and the environment: each interaction potentially alters the competence, knowledge and identity of the individual, as well as the micro-ecology. Learning, seen as the acquisition of knowledge, or the ‘capability to take effective action’ (St Onge 2003) can also be defined in terms of affordances, viz: Learning is the process of exploring, creating and benchmarking new affordances. Siemens’s diagram (Figure 1) shows the interplay between affordances, innovation and change (2006).


Figure 1: Change Pressures, Spaces and Affordances

1. Actor-networks
Actor-Network Theory (or ‘ANT’), provides a possible framework for dealing with uncertainty and complexity, within its ‘actor-networks’. This might help us to explore complex adaptive networks.

Actor Network Theory, particularly as described by Latour (2005), insists that a range of ‘uncertainties’ are crucial to a sound epistemology of knowledge, for both the social and the natural sciences. Law (1992: 379) summarises the key tenets of Actor-Network Theory, or ‘ANT’: “social relations, including power and organisation …[are] network effects … networks are materially heterogeneous … the task of sociology is to characterise the ways in which materials join together to generate themselves and reproduce institutional and social patterns” (ibid). Law goes on to say that ANT is “a theory of agency, a theory of knowledge and a theory of machines” [and organisations] (op. cit. 389), and that the core of the actor-network approach is “a concern with how actors and organisations mobile, juxtapose and hold together the bits and pieces out of which they are composed … and how they manage, as a result, to conceal for a time [the way in which] a heterogeneous set of bits and pieces, each with its own inclination [can be turned] into something that passes … as an actor (op.cit. 386). In short, Law is concerned with the way in which the agency of animate and inanimate things is constantly reconfigured, or as Latour would say ‘re-assembled’: texts, machines, organisations, people and so on.

Latour (2005: 22) characterises ANT somewhat differently, although the upshot of his emphasis on ‘uncertainties’ is broadly at one with Law: both of them focus on the way in which the (micro-) interactions between “heterogeneous set of bits and pieces” (and nested layers of these!) continue to constitute what passes (rather deceptively they would say) for stable personal and institutional agency and process. Latour specifies five major uncertainties. These concern the nature of:

1. Groups: there exist many contradictory ways for individuals to be given identities.
2. Actions: in each course of action a great variety of actors seem to barge in and displace the original goals.
3. Objects: the type of agencies participating in interaction seems to remain wide open.
4. Facts: the links of natural sciences with the rest of society seems to be the source of continuous disputes
5. Studies: done under the name of the science of the social as it is never clear in which precise sense social sciences can be said to be empirical.

ANT’s point about networks is much the same as Foucault’s point about the ‘capillaries of power’ which are the ‘seat’ of Foucault’s ‘circulation of power’ in which, precisely, a series of physical and semiotic materials join together to generate themselves and reproduce institutional and social patterns’, just as Law says of actor-networks (ibid). Foucault’s concept of discourse, which can be paraphrased as: “a system of artefacts and texts which orders bodies, animate and inanimate, within a discourse community” (Williams 2005) puts more emphasis on relations of power and social organisation, and on the communities that maintain (or contest) them, all of which seem to be a bit perversely excluded in ANT, particularly by Latour (2005), who insists on keeping the researcher’s nose very close to the events.

Foucauldian ‘discourse’ is, of course, similarly ‘materially heterogeneous’ to Actor-networks: it includes Panopticon towers and psychiatric straight jackets, as well as laws, edicts and classifications, so there’s nothing new there. What ANT does provide is a slightly obsessive micro-empiricism, and a rigorous scepticism (both in the best traditions of science), which we can use to closely observe and to account for the way in which actors in the unfolding social bump into things and bump into each other, as the social is constantly reconfigured. If the social is based on layers of uncertainties (and therefore similar to Snowden’s ‘complexity’ see later on), then the social is indeed constantly reconfigured, or as Latour would say, re-assembled (2005).

ANT also makes a specific link between the person and their environment. Law writes that the term actor-network is crucial, because “an actor is also, always, a network” (op. cit. 384). This is similar to the point that Gibson makes about affordances, which are products of the interaction between the individual and their environment (ibid); and in fact Gibson goes further, and formulates perception itself in terms of interaction and affordances (Withagen & Michaels 2005). Such ecological affordances are established and exploited in much the same way as the steps unfold that an actor takes as an actor-network. This is the obverse of the Foucauldian perspective, in which an individual would act within several discourses, endorsing some and contesting others, simultaneously. ANT’s actor-networks seem to move from one part of an actor-network to another, but do not seem to have the facility or the affordances to move within more than one network at the same time, in a similar way to the way in which one can say that someone acts within several discourses at the same time.

This ANT conception of the actor is a bit restrictive, and it seems at times to be a little reductionist, a little too materialist. There is little obvious space in an ANT account for the cognitive and imaginative multi-tasking that a person might do, as they proceed to create an actor-network, while they are satisfying (or contesting) the demands of more than one ‘discourse’, unless of course the actors inhabit several n-dimensional networks simultaneously. Latour’s five uncertainties do have space for the uncertainty of the account that is created by the researcher, but it seems a little short on space for the multiple possible accounts that figure in the mind of the actor, as ‘actor-network’. In fact this ‘dualism’ between the actions and the minds of the actors is probably precisely the kind of thing that Latour wants to avoid: it’s methodologically possible to circumvent this duality, but it could lead to a rather convoluted approach.

Actors in ANT are, crucially, needed to account for themselves, but there seems to be a gap in the account, in that there doesn’t seem to be space for the actors to account to themselves, which they regularly do: sometimes in contradictory ways, and sometimes in meta-cognitive, or reflective ways. This is a problem, both for an adequate account of the actor/s-network/s, and for an adequate account for learning and knowledge creation and management.

2. The Capitalisation of Knowledge
ANT as an epistemology for research is rigorously sceptical (particularly of critical sociology). It is invaluable in pointing out the requirements for empirical research, and in particular, the uncertainties that are involved in tracing the unfolding configurations of the social. But it does not give an adequate account of the process of commodification and the production of formal knowledge and its capitalisation in science and information systems, which are an increasing portion of the value embedded in productive and regulatory systems and artefacts, which include the policies and procedures of bureaucracies and liberal democracies.

The process of the capitalisation of knowledge, e.g. science, is based on the production of procedural information as a commodity. Commodification, in science, finance, and bureaucracy is the process which strips out subjects, context, agency (and therefore actor-networks) to yield information which can be exchanged, shared, traded, and, potentially, used by any actor/network, anywhere. This is a straightforward extrapolation of Marx’s semiotic analysis of financial capital – the basic semiotic parameters of commodification in other fields are substantially the same: strip out the subject, the context of the subject, and the context of use. The result is a set of commodified semiotic information, or meta-semiotics (Williams 2005), which can be used in any context. The social, it is true, is configured by and with actor-networks, but it is also, increasingly, configured within meta-semiotics: precisely the systems of procedural information and knowledge that Latour fights shy of, unless he would say that commodified forms of knowledge are just yet another set of actors/actants.

The commodification of information remains a major factor in economic and social development across the late 20th and early 21st Century. What is changing now is that a series of event horizons are emerging that provide for something quite different: rather than the modernist ‘embedding’ of value, or the ‘programming’ of information or knowledge into productive and regulatory systems and artefacts, the relationship is becoming a flexible, re-programmable one: the configuration of ANT’s ‘heterogeneous bits and pieces’ and their relationship to knowledge and procedural information, is changing radically.

The nature of the Information Age was, to paraphrase Toffler, to increase the component of knowledge/information that was embedded in systems and artefacts, but the nature of the Knowledge Age is to change the very nature of that relationship, from one of ‘embedding’ to ‘re-programming’, and from ‘re-programming’ to ‘networking’, arriving at what might be called the Age of Transparency, where transparency, and no longer knowledge, is power: the projection of power is now, more than ever before, a function of your siting (and thus your line of sight) within interactive and communicative networks. One’s affordances as an actor-network can change radically with changes in your access to, and your exploitation of, communication networks and the extent to which they are transparent to you.

3. From Embedding Knowledge to Networking
Let’s have a more detailed look at the changes in the process of embedding knowledge …

Embedding is a one-way process, and is typically ‘cast’, ‘once and for all’ in the monuments of Modernism: iron and steel ocean liners, steel-reinforced concrete bridges and highways, and steel and glass skyscrapers (9-11 notwithstanding). The affordances of Modernism are designed and built to remain (the same), even though some of the great ocean liners become, literally, anchored luxury hotels (but were they ever anything else?)

Programming increases the ratio of information or knowledge in the mix, and starts to disengage information systems from productive systems, into separate computers, communications technologies, and IT systems, which are often then re-embedded in the artefacts. Cars, and even toys, in the late 20th Century often had several ‘computers’ built into them.

Re-programming reconfigures the human-machine interface, and starts to allow the humans to shift (up) from just designing the information input, to being able to continue to manage, change, and even reconfigure that input while the system is in operation, in the light of strategic decisions that they make, independent of what was embedded or programmed into the productive and regulatory systems and artefacts in the first place. DELL computer assembly was an early, but rather limited example of this.

Networking, in complex adaptive systems, takes this one step further. In complex networked systems, the way the components interact with each other, even at a micro-level, results in constant shifts in the ecology, which has knock-on effects that impinge, even if slightly, on the global social ecology – in the sense that an ecology includes, by definition: material, semiotic, self-organising and self-reproducing elements: from the earliest material/semiotic of RNA and viruses onwards.

Of course, in an actor/network sense, all systems are networks, but that doesn’t help us much. More to the point is the fact that global digital networks add several orders of magnitude to networks which were previously more cumbersome, massively more expensive, much slower, less rich and versatile, and extremely inaccessible to most people. So we make a sharp distinction between complex adaptive systems and complex adaptive networks, or complexity prior to, and subsequent to, the event horizon of the global connectivity and the transparency of the Internet and social software. Cetina (2005: 214) adds more detail and richness to this model, in her analysis of micro-global networks. She writes that “global microstructures [e.g. financial markets, global terrorism] need not imply further expansions of institutional complexity. In fact they may become feasible only if they avoid complex institutional structures. Global financial markets for example … simply outrun the capacity of such structures … Global [micro-systems] do not exhibit institutional complexity but rather the asymmetries, unpredictabilities and playfulness of complex (and dispersed) interaction patterns”.

Snowden (2002:7) makes a similar distinction between the complex and the complicated: “an aircraft is a complicated system: all of its thousands of components are knowable, definable, and capable of being catalogued…cause and effect can be separated, and by understanding their linkages we can control outcomes. Human systems are complex [they] comprise many interacting agents, an agent being anything that has identity: the system is irreducible. Cause and effect cannot be separated because they are intimately intertwined”.

What ANT (particularly in Latour’s work) refuses to do is to make this distinction: i.e. the distinction between agents (or variables) with and without identity – identity in the sense of self-maintaining and self-reproducing identity - which have also been called ‘variables with attitude’ (Williams 2003), i.e. anything from viruses onwards. It might be possible to write a perfectly adequate ANT account without belabouring this distinction, it is true, but that will unnecessarily conceal the difference between elements within the ‘network’, not to mention within some of the ‘actors’. To complicate the matter further (no pun intended), the properties of complex systems per se are not limited to ‘live’/ human phenomena – the computer simulation, Life, is a celebrated example of what is essentially a complex piece of mathematics/graphics, as is the behaviour at a non-human level, of ants, both are complex systems.

4. Complex Adaptive Systems
We need to systematically explore complex adaptive systems: what are the characteristics of complex adaptive systems, and how should we approach knowledge management, and in fact general management within them? Are they self-organising, and if they are, what are the implications? And most importantly, are self-organising systems inherently desirable?

Complex systems are:
  • Open systems: they interact with their environment
  • There are a large number of elements, which interact freely.
  • The interaction is fairly rich, i.e. any element in the system influences, and is influenced by quite a few other ones.
  • The interactions are at least in part non-linear.
  • Interactions have a short range: “patterns are determined … by local interactions among decentralised components” (Resnick, quoted in Urry 2005:1).
  • There are loops in the interactions: positive and negative.
  • It is difficult to define the borders of a complex system. Therefore the scope is defined by the purpose and influenced by the observer position.
  • They operate far from equilibrium. “A living system is an open system that maintains itself in a state far from equilibrium, and yet is stable” (Capra:37)
  • They have histories. The past influences current behaviour.
  • Each element is ignorant of the behaviour of the system as a whole, it responds to information available locally.
  • Major disproportionalities are possible between cause and effect, outcomes are often unpredictable, and are influenced by self-organising, emergent structures.
  • Multiple pathways: these are “an essential property [or even] the defining characteristic of a [complex] network” (Capra: 41)
(From: Celliers 2005, Urry 2005, Capra 2005, Snowden 2002, Cetina 2005)

What we need to add to this model is the interplay between commodified knowledge (or formalised procedural knowledge) and complexity. It is possible to distinguish between formal knowledge (or commodified procedural information, see above) and strategic knowledge. Strategic knowledge is the fit between procedural knowledge and contextual analysis (Williams 2006 & 2008), or what St Onge (2003) calls the “capacity for effective action”. And it is important to make the link back from
formalised knowledge to ante-formal knowledge, i.e. the knowledge that is circulated face-2-face and, increasingly, in digital media and the networks of social software, which is ‘not-yet’ formalised, but could well be (see Figure 2). Ante-formal knowledge is more than just ‘folk-wisdom’, it’s more a case of the ‘wisdom of crowds’ (Wikipedia is the prime example) and we might add another variant, the ‘wisdom of networks’.


Figure 2: The Knowledge Process Cycle (KPC)

The knowledge process cycle, Figure 2, shows how Formal (or commodified) knowledge relates to Ante-formal knowledge and information (A-F K. & Inf.), Strategic Knowledge, and Communities of Practitioners, and in turn, how these relate to Experience (Exp.) and Data. In terms of both design and management, it is important to note that different activities and resources that are required for the transition, and transcription, from one phase of the KPC to another, namely:

· Language and semiotics: for the transition from Experience to Data
· Cultures, or ‘social-ware:, for the transition from Data to Ante-formal Knowledge and Information
· Commodity Incubators (and meta-semiotics): for the transition from Ante-formal knowledge to Formal Knowledge
· Design and Strategy: for the transition from Formal Knowledge to Strategic Knowledge
· Alliances and Associations: for the transition from Strategic Knowledge to Communities of Practitioners
· Applications: for the transition from CoP back to Experience.

The Knowledge Process Cycle itself is merely the core of the Knowledge Networks that need to be managed within organisations.

The complete picture looks more like Figure 3, An Architecture for Managing Knowledge Networks in which, for each of the phases of the KPC, the resources and facilities that are required are:

· for Experience: Mobile multi-media.
· for Data & Data-bases: Data-mining and display software
· for Ante-formal Knowledge and Information: Social Software, both ICT and ICS (Interactive and Communicative Spaces: virtual and physical)
· for Formal Information and Knowledge: ‘Blind’ P2P journal reviews, Markets, Law, IPR, Share Value, IC reporting.
· for Strategic Knowledge: P2P networks, GUI’s for design and scenario planning, narrative creation, dissemination and use.
· for Communities of Practitioners: ICT and ICS, collaboration, netware, reflective practice.
These functions and resources need to be managed across the KPC, and across the physical, IT and ICS systems, supported by the Structural Capital of the organisation, and that of its environment, as well as the three overlapping domains, of:

· Relational Equity: CRM, Alliances, Professional Associations, Lobbies.
· Social Equity: Network Presence, Collaboration.
· Human equity: Continuous Professional Development.


Figure 3 An Architecture for Managing Complex Adaptive Networks

4. Complex Adaptive Networks
What then are the implications for the Management of Knowledge Networks, and specifically, for managing Complex Adaptive Networks or Knowledge Ecologies?

4.1 Affordances
However much ICT and digital networking offer new technical wonders, it is important to focus first of all on the outcomes that we want to achieve, and the functions that we require, before moving on to the tools we need to achieve those outcomes, and functions.

One way to approach this is to outline the (very different) affordances that we need for various aspects of the architecture that we require for Managing Knowledge Networks. We can approach this task from a number of perspectives. From the perspective of the phases of the KPC, for instance, the affordances would fall into the following domains:

Formal Knowledge
Affordances for the production, development, dissemination and application of formal knowledge and information: for the production of the procedures of science, finance, bureaucracy, and information systems, and their subsequent application in practice, which require:
· formal methods (the meta-semiotics of science and financial markets), and their related
· business and engineering systems and practice

Ante-formal Knowledge and Information
Affordances for the emergence of ante-formal practices and knowledge, and the capture, dissemination and application of elements of those practices and that knowledge, to be used (and formalised) elsewhere in the knowledge process cycle [This provides the substance of social life] require:
· Interaction
· Participation
· collaboration
· play, and fun
· integrated digital media and networks.

Strategic Knowledge and CoP
Affordances for the production, development, and dissemination of strategic knowledge, alliances, and communities of practitioners [This provides for the governance of business and social life] require:
· Stories and narratives
· Maps, Animations
· Meta-information

5. Managing Complex Adaptive Networks
At this point the premise that has been tacitly suggested in this paper needs to be made quite explicit, viz:

Information, knowledge, communication, and change at local and global level continue to increase in volume, reach, speed and complexity. Some of these changes, particularly the development of global digital media, have crossed a unique event horizon, in which complex adaptive networks can, if judiciously managed, provide a welcome response to these issues, and provide a host of new opportunities, via the self-organising and self-sustaining affordances of complex adaptive networks.

5.1 Characteristics/ Functions of CANs
In addition to the properties of Complex Adaptive Systems that we have discussed so far, it is worth noting the following features, particularly as they apply to Complex Adaptive Networks (CANs):

· Retrospective coherence rather than predictability
· Emergent properties, frozen accidents, event Horizons.
· Self-organising and self-sustaining
· Basic, widely accessible infrastructure and a few simple rules

CANs specifically require:

· Widespread and low-cost interaction
· Efficient and cheap communication
· Transparency, and immediacy
· Good affordances for people to create, revise, and disseminate a full range of texts and media artefacts: the emphasis on ‘writing’ rather than ‘reading’ (producing rather than consuming) texts.

The Internet and Social Software provide this.

5.2 Management Issues

The characteristics of CANs, particularly immediate, high-access/ low-cost, transparent communication, and the primacy of local interactions, often shielded from broader interactions and controls, mean that collective responsibility, rather than control and hierarchy, or dogma (religious or secular), within a shared ethics is the most desirable, and the most efficient, mode of governance.

Increasing Transparency
Global digital media provide the basic infrastructure for CANs. In the late 1980’s one of the dividends of transparency, per se, (communications as well as satellite based surveillance) was the ability of the two sides in the Cold War to effectively and immediately see, and monitor, what the other party was doing. There were many factors that led to the end of the Cold War, but the arms limitation treaties, and the arms reduction that followed would not have been possible without global transparency. Lower resolution spin-offs from these developments include Google Maps and Google Earth, as well as GPS-based motoring route guides.

Reducing Transparency
On the other hand, the more recent and more elaborate Internet and Social Software facilities offer the affordances of comprehensive Complex Adaptive Networks. As Cetina (2005) has show in her work, this has enabled the development of micro-global networks for global finance, as well as more recently for global terrorism.

The Chinese model of liberalising the economy without liberalising politics is attempting to separate out the use of CANs in finance from their use in politics – internally and internationally. This seems to be difficult but not impossible.

The West, led by the USA, has re-instated many of the restrictions on CANs in global finance and communication, and civil liberties, following 9-11. This reversed not only the degree of global transparency, but also reversed the substantial shift away from controls based in individual nation-states – it really did reconfigure global governance, and was a good example of de-volution.

Jeffrey Garten (2002) wrote that the historically unique way in which late 20th Century global markets cut themselves off from national and local control and interests, and operated in a hyper-discourse of global capital and ‘free’ trade, in which money and trade seemed to flow almost transparently across the world was cut back: ‘downsized’, in its own parlance.

He pointed out that the regulators are back, national borders are once again in fashion, and the “‘primacy of geopolitics over economics … is once again a fact of life” (ibid). One of the key elements of this global hyper-discourse is precisely its transparency – the extent to which transaction costs (particularly in capital markets) have been reduced to close to zero: zero cost and zero time, with no regard to distance or place, all of which has been possible because of the exponential growth and efficiency in global ICT networks.

It is generally true that new technologies always tend to increase existing discrepancies in power. International capitalism triumphed after the end of the Cold War, and it quickly took advantage of ICT to consolidate and expand its hegemony. However, these global technologies and markets were in fact fully fledged CANs, which can be exploited by anyone, and so other trans-national organisations and networks like Al-Qaeda smartly took up the opportunity too – the more so because Al-Qaeda has deliberately structured itself as a virtual organisation, highly aware of the distortions that fully fledged CAS’s and CAN’s offer (see Cetina 2005).

After Enron, the dot-coms crash, and 9-11 concerted efforts were made to curb the negative aspects of these CANs, or to put it anther way, to limit the abuse of virtual communities of practice. The excesses of fundamentalist across the spectrum, from the religious fundamentalism of Bin-Laden to the market fundamentalism of international global capitalism (or what Robin Cook called “feral capitalism”) exploit the affordances of communication, local interaction and transparency to such an extent that these affordances unfortunately have to be reigned in. Some measure of de-volution is necessary to correct the overshoot in global CAN’s.

Threshold Conditions
The theory of complex adaptive systems, or networks, is closely aligned with evolutionary, and ecological theory, which requires a degree of stability and need satisfaction to function. If these basic conditions are not met, evolution goes by the board, and important evolutionary mechanisms like neoteny (delayed reproductive maturity, seen prominently in animals like higher primates) are put on hold.

The development economist Rihani points out that CASs at the level of national development require very basic needs to be met, even prior to meeting the threshold conditions for economic stability. He argues, with evidence from many countries, that:

“progress is not a function of cost… The danger in isolating wealth as the primary impediment is the temptation to conclude that little can be done until the economy has been uplifted substantially. That, fortunately, is not the case” (2002: 211 – emphasis added).

On the previous page he compares the case of India with that of China, and says:

“China decided to seek local solutions, while India depended on imported ideas. A near obsession with basic needs, indigenous remedies, and communities that function as supportive units has enabled that country [China] to cross the capability divide [emphasis added]. Having moved up the human development ladder, China is now ready to tackle economic development”.

And he goes on to emphasise that “the capability for most people in a nation to interact effectively is determined to a large extent by their state of nutrition, health and knowledge. Without that capability, they will be powerless to take an active part in the natural evolution of that nation as a Complex Adaptive System”.

The debate on the ‘digital divide’ could usefully focus on Rihani’s basics, before focusing too much on technology.


This paper examined the theories of Actor-networks, Foucault’s Discourse, and Complex Adaptive Systems. From a critique of these theories, and a consideration of the Knowledge Process Cycle, an integrated model has been developed to provide an Architecture for Managing Knowledge Networks, or Complex Adaptive Networks (CAN’s).

Key issues in the management of CAN’s have been explored, including issues of Transparency and Threshold conditions.

Clearly CANs represent a major new event horizon for communicating, developing and sharing knowledge. This applies to all of the aspects of the Architecture for Managing Knowledge Networks, and all the phases of the Knowledge Process Cycle. It does not mean that we should step back entirely from traditional management concerns, but it does mean that we need to be mindful of the benefits of CANs for specific aspects of Knowledge Network Architecture, many of which benefit from, or even require, a specifically CAN approach to thrive. This in turn requires a different management mind-set, even in universities, where management models are often still focused on centralised control, and slow, incremental, hierarchical, linear change.

Managing Knowledge Networks is clearly more complex, but if the thesis of this paper is correct, i.e. that CAN’s are a potential solution to quite a few the problems of increasing scale, speed and volume, then we can design CAN’s in which self-organising behaviour can lead to the development of communities of inquiry which no longer require the same level of constant management surveillance and intervention that might have been required in the past. There are already some examples of such developments, in Communities of Practice, (Online) Moderated Peer Learning, and wikis like Wikipedia, all of which are now firmly established. Newer forms, like ‘unconferences’ and online JAMs are next.


Capra, F. (2005) Complexity and Life. Theory, Culture and Society 22(5): 33-44.

Cetina, K.K. (2005) Complex Global Microstructures. Theory, Culture and Society 22(5): 213-234.

Cilliers, P. (2005) Complexity, Deconstruction and Relativism. Theory, Culture and Society 22(5): 255-267.

Deacon, T. W. (2003). The hierarchic logic of emergence: untangling the
interdependence of evolution and self-organization. In B. Weber, & D.
Depew (Eds.), Evolution and learning: The baldwin effect reconsid-
ered. Cambridge, MA: The MIT Press, Chap. 14

Drucker, P. (2001a) The next society Economist Nov 1st 2001, London.

Garten J (2003) A new year; a new agenda.
Economist January 4th, London.

Gibson, J.J. (1979) The ecological approach to visual perception. Boston, M.A., Houghton

Kurtz, C. & Snowden, D. J. (2007) Bramble bushes in a thicket. Narrative and the intangibles of learning networks. Forthcoming in Gibbert, Michel, Durand & Thomas: Strategic Networks: Learning to Compete. Oxford, Blackwell.

Law, J. (1992) Notes on the Theory of Actor-Network. Systems Practice 5(4): 379-393.

Mitra, S. (2003) Minimally invasive education: a progress report on the "hole-in-the-wall" experiments. British Journal of Educational Technology 34 (3), 367–371.
Rihani, S (2002) Complex Systems Theory and Development Practice, Zed Books, London.
Siemens, G. (2006) Knowing

Snowden, D.J. (2002) Complex Acts of Knowing: Paradox and Descriptive Self-Awareness. European Conference on Knowledge Management, Dublin, Ireland.

Saint-Onge, H. & Armstrong C (2004) The Conductive Organization: Building Beyond Sustainability. Elsevier: Butterworth HeinemannSt Onge,

Toffler, A. (1970) Future Shock. Nueva York, Randon House

Urry, J. (2005) The Complexity Turn. Theory, Culture and Society 22(5): 1-14.

Williams, R.T. & Walters, C (2003) Discourse Analysis and Complex Adaptive Systems Theory. 2nd European Conference on Research Methodology for Business and Management, Reading University, UK.

Williams, R.T. (2005) Meta-semiotics and Practical Epistemology. Theory & Psychology, Vol. 15, No. 5, 711-737

Williams, R.T. (2006) Narratives of Knowledge and Intelligence: beyond the tacit and explicit. Journal of Knowledge Management.

Williams, R.T. (2008) The Epistemology of Knowledge and the Knowledge Process Cycle Forthcoming in: Journal of Knowledge Management.

Withagen, R. and Michaels, C.F. (2005) On Ecological Conceptualisations of Perceptual and Action Systems. Theory and Psychology 15(5): 603-620.