The concept of ‘self-organising systems’ is in danger of losing its effectiveness and becoming as vague a term as ‘general systems theory’ and ‘autopoesis’, with their abstract talk of generic open and closed systems. For example, Georges and Romme (1995) define a self-organising system as one that is both open and closed; evoking the old debate about what constitutes a closed system. Mingers (1997) definition of autopoesis (self-reproducing) would appear to subsume self-organising systems. However, doing so may hide some of the advantages of using the perspective of self-organising systems for automatic knowledge sharing when wicked systems pose problems that need to be solved. In order to be able to reproduce, a system needs to be organised, and it may be hierarchical. A self-organised system, however, is one that does not need a hierarchy to respond to environmental surprises. It is the assumption of the need for a hierarchy to direct knowledge sharing that is being challenged here. Self-organisation is not being used here in the sense of the development of identity in a hostile environment, (such as the establishment of the early Christian Church, the labour movement or the feminist movement). Rather, this paper is concerned with how a wicked socio-technical system might be designed to share knowledge so as to provide an effective response to environmental surprises when there is no explicit internal hierarchy. Ideas about how these systems might be designed come from analogies with the world of insects. Some swarm, as in ant nests, and some bee nests have no boss, no corporate plan, and no strategic planner, but a higher level organisation has emerged that serves to enable the unsuspecting insects to make a strategic response to unpredictable large scale problems that suddenly impinge upon their world.
There is an extensive literature (e.g. Mingers, 1997) on self-reproducing, self-replicating, and similar systems. This paper will bypass revisiting these and merely synthesise from two other related areas. The first is the empirical scientific biological literature about what insect colonies do to share knowledge to provide an effective strategic response to problems. The second is the small-worlds literature, which has recently moved from the sociometric to the sciences, as more and more biological systems are seen to use the small-worlds structure to share knowledge. These will be discussed in terms of a story from the crisis management literature, which tells what actually happens in response to a rapidly changing, community-based problem, in particular when the strategic response has voided any pretence of controlled top-down knowledge sharing.
There has been a lot written about self-organisation in the biological and related sciences. Much of this literature is presented as a mathematical analysis of patterns that emerge, e.g. waves, sand dunes, tree structures and the markings on animals. Camazine et al. (2001), however, provide an empirically based explanation using insect systems. This paper’s interpretation of what is meant by self-organisation draws heavily on this, and thus draws on analogies from the world of the insect nest. Camazine et al. (2001) observe that some complex actions emerge through simple interactions internal to the system, without intervention by external directing influences. More formally they define self-organisation as:
… a process in which pattern at the global level of a system emerges solely from numerous interactions among the lower level components of the system. Moreover, the rules specifying interactions among the systems components are executed using only local information. [p8]
Camazine et al. (2001) do not accept that the queen insect in an ant’s nest or beehive is somehow ‘giving instructions’ to the millions of insects who have never been near her. The term ‘queen’ is misleading; the term ‘womb’ would be more acceptable from a knowledge sharing perspective. Each individual ant or bee bases its behaviour on its perception of the position and behaviour of its nearest neighbour, rather than on knowledge of the global behaviour of the whole group. Local dynamic knowledge sharing is all that is present, yet the insects are able to make strategic responses to a global threat to the whole nest. A strategic response somehow ‘emerges’ from lower level actions, evidenced by the very existence of a nest that has specialised integrated operations. The individual ants, for example, are not even thinking about this higher order purpose; rather they are only concerned with their own small function in the nest. If this emerging strategy appears different to the actions of lower level activities, then the system may be described as complex. Individual ants forage for food, build the nest, care for the eggs and milk the queen, yet somehow these activities have become coordinated to produce a species that has survived, and very successfully, for millions of years.
Camazine et. al. (2001) summarise the now significant amount of empirical research that has been conducted on insect nests to better understand how a strategic response can emerge. For example, a few ants placed in a Petri dish were found to move sand around in a random fashion, achieving nothing. But when enough ants were added, the probability of the production of a randomly constructed shape that the ants recognised and would respond to, increased. The presence of these particular shapes then acted to suddenly start the ants working in a coherent manner, constructing recognisable elaborate structures. In another example, an ant’s nest was deliberately damaged and metal plates used to divide the damaged area in such a way that knowledge sharing between the two damaged areas was impossible. The strategic response, the reconstruction, matched perfectly. When the dividing plate was removed, the rebuilt sections looked like one singular rebuilding exercise, perfectly orchestrated. In summarising this empirical literature, Camazine et al. (2001) identify a series of conditions necessary to enable the emergence of a knowledge sharing system from the insect activity that results in a coherent strategic response. These include the presence of:
group influence;
stigmergy;
decentralised control, dense heterarchies; and
dynamic knowledge sharing.
Camazine et al. (2001) do not clearly label this attribute of a self-organising group; rather they sum it up as ‘I do what you do’. The idea starts with noticing that members of a group copy or mimic those around them; they are influenced by the actions of others. Children do what their parents do, artisans learn from their masters, business schools teach the ‘echo of lies’ of how management is done, and when at work we learn a corporate culture, we become team players. We learn the preferred way of doing things if we want to ‘get along’. Examples of our compliance to our local group norms include our dress, religion, food and ethics. However, we can from time to time insert some small minor variation based on experiences we have had elsewhere. This is analogous to our genetic make-up; we are only minor variants of our parents, but we are variants. An invention, a new recipe or a clothes fashion change are examples of an individual changing a group’s behaviour, but if we are honest, one person usually makes very little difference to the generic behaviours of a community. This ‘get along, go along’ behaviour seems related to our very strong ‘inclusion’ needs; we need to belong to a group. Horses are trained by threats to exclude them from the herd, which is far more sustaining as a threat than physical pain. Arguably, the worst punishment we inflict on other humans is solitary confinement. The need to belong is seen as an explanation of why herd species and insect colonies are influenced as they evidently are by the behaviour of the whole group; expressed as ‘I do what you do’.
Being influenced by the behaviours of others, especially those immediately around us, is central to self-organisation. An insect seems to be born perceiving that the world will be intimately integrated with what the insects colony around are doing. An ant will merely do what the ants immediately around her do, using whatever genetically received devices she has at her disposal. More empirical evidence of this from insect research includes the behaviour of fire flies. When swarmed, fire flies, with their flashing tails, will all end up synchronising their flashes. The fire flies will alter their flash time and speed under the influence of the group. Infectious yawns, synchronised reproductive cycles, synchronous breathing and ‘mobbing’ are all examples of human group behaviour that influence individual behaviour. Wilson (1983), giving the example of a librarian thinking about the demand for books, uses the term ‘cognitive authority’ to identify who of those around us we choose to mimic. In an insect colony it is assumed the individual insects can only choose to mimic, to listen to, those immediately around them. Modern people, who have access to the media, books and different corporate cultures, and are able to travel to numerous different communities, have a much wider choice of cognitive influences to mimic.
In this social setting, knowledge sharing between those in immediate contact is expected to have already largely occurred. When a crisis occurs, more than one insect knows the same things.
Camazine et al. (2001) are very cautious about the idea that colonies of insects carry in their heads a detailed recipe or fully laid out blueprint of what, for example, a nest should look like; a detailed vision of what the finished construction should do and be. This is justified with the empirical evidence for how nests respond to different physical situations. The insects build allowing for the physical conditions encountered, so every nest is slightly different. Yet overall common design features are observable. This is not attributed to the insects’ knowledge sharing, but to their merely responding with a set of alternatives.
Stigmergy is a term attributed by Camazine et al. (2001) to Grasse. It refers to the mechanism whereby a swarm insect (such as an ant or bee) is stimulated to work constructively towards a common purpose by the presence of work in progress. The half completed work of other similar insects is recognised as an ‘event’ that induces automatic responses from those that see and recognise it. For example, an ant may see a pair of pillars and respond by building an arch between them, without having communicated directly with the earlier builders. This is an indirect form of knowledge sharing; an event is driving asynchronous knowledge sharing. The human equivalent may be the response of rescuers when a building is seen to collapse or a child is seen to be treated badly. In place of stigmergy, Michener uses the expression, ‘indirect social interactions’. In systems management this may be called asynchronous knowledge, or sharing through design. For insects, it may be the result of the quantity of pheromones on the half finished building works, or it may be the physical shape that acts as the asynchronous stimulus.
It is possible to appreciate the importance of asynchronous knowledge sharing to the running of a complex system (one that has emergent properties), such as an ant nest, by drawing on the analogy of a modern corporation. The existence of multinational corporations has been attributed to asynchronous knowledge sharing of faxes, (e)mail and web pages. The size of organisation that can be controlled by means of oral knowledge sharing, is restricted. In large organisations, while oral synchronous knowledge sharing remains very important, time zones, legal records and very detailed specification require ‘written’, asynchronous knowledge sharing. Nations that have developed joint synchronous and asynchronous knowledge sharing have been dominant in economic and scientific terms. At a more modest level (in insects) Camazine et al. (2001) are suggesting a more subtle form of asynchronous social interaction, and thus motivation in the presence of half completed tunnels, pheromone paths and other work in progress.
Given the centrality of purposeful activity to systems thinking and design (Ulrich, 2002; Checkland, 2000), it seems necessary to mention that purposeful activity is presented by Checkland and Ulrich as an emergent property from the large, self-conscious human brain. This is thought to enable us to appreciate the purpose (drivers) of our actions and stand outside ourselves. But should insects be thought to be engaged in purposeful activity, rather than operating like parts of an alarm clock? Are insects living out genetic drivers to bring up young and continue the gene pool? Surely any human self-organising system would need to anticipate that the participants would be able to ask themselves why they should act. Moreover, in human systems, language could be used to provide a driver to act. Therefore, in human self-organising systems it may be necessary to emphasise why people should act if it is not otherwise obvious to them that they should. However, Camazine et al. (2001) understate the influence of purposeful activity, even those of genetic survival, for insects compared to group interaction influences of humans. They place much more emphasis on the insects’ response than to their driving forces for achievement. The genetic drivers of gene survival are not emphasised, possibly because most of the insects never see or come in contact with the young.
The decentralised control attribute identified by Camazine et al. (2001) is defined as a particular ‘architecture of information flow’. Each insect responds to other insects immediately around it to learn what is to be done, rather than from messages from well-informed individuals (leaders) in the upper echelons of a control hierarchy. The organisation chart is one of small clusters of interacting insects responding to one stimulus, such as a half built arch in one part of the nest, or a food retrieval clique at another location in the nest. There is no tree of hierarchical knowledge flowing up and down; rather the structure is more a series of independent clusters of workers who, ninety percent of the time, only communicate directly with the other members of their cluster (described as cliques, or small-worlds). Only when they are unable to solve a problem with local knowledge sharing will they venture out to ask another cluster. The dense heterarchies attribute reinforces the image of a series of separate yet connected small clusters, each focusing on different but loosely interconnected tasks. Heterarchies are inter-independent groups; they are neither hierarchical and nor totally independent clusters. This raises concerns about how a strategic response from these roughly independent responding clusters is possible. The small-world literature may help here.