Dave Snowden

Social atomism, Identity & natural numbers

RSS Feed

Bob Hruzek hinted that I had started but not yet finished my discourse on natural numbers. I had elaborated it to a degree here, but have so far failed to fulfill my promise to look at the impact of social atomism on social systems. In parallel I promised to argue that focusing on identity not the individual as the primary unit of analysis resolves a lot of otherwise intractable problems. It also allows us to scale, within the limits of natural numbers to large populations. Now while both these concepts have significant practical implications for on line communities and society in general I need to start the discussion with some metaphysics …

Navigation: There are three sections here, Social Atomism, Identity and Implications. You may want to skip through to the final section and read that before deciding if you want to wade through what is a long post.

Now readers may not be familiar with the term Social Atomism. Put simply it is the common assumption that the the primary unit of analysis is the individual and that communities (and society) are aggregations of individuals and negotiated interests between those individuals. Related concepts such as social capital, as popularised by Prusak and others in knowledge management, use terms such as the favour bank. The implication here is that we do things for other people in the expectation that they will then reciprocate the favour. Atomism, or individualism is the legitimising framework for liberal democracy and free market economics. I am going to argue that the primary unit of analysis in human system is identity, which should not be confused with the concept of role. As such I am challenging the dominant ontological assumption of the anglo-saxon world so this is no small undertaking and what follows should be considered an overview, developing an idea that will be elaborated in the book.

A quick clarification here. I am using ‘anglo-saxon’ as an abbreviation for ‘North American & England’; I do not want to use the popular west-east contrast as that is not accurate. Most people when they reference western thinking really mean anglo-saxon and are ignoring multiple other european traditions. Eastern thinking is not unified schema either by the way. I am using ontology in its philosophical meaning as the science or study of being, what it means, or what something must possess in order to exist. The more recent corruption of the word by the IT community to mean a form of glorified taxonomy is deprecated.


One of the best books on this subject is Weissman’s A Social Ontology which provides both a theoretical background and a good exposition of the implications. I will quote intact from the introduction:

Individualist – atomist – theories emphasize the self-sufficiency and moral autonomy of persons. They speak of freedoms, rights and exceptions; rarely or never of reciprocities, duties and connections. Individualism dominates our self perception. It encourages us to deny the ligaments and nerves of our social lives. Perceiving every society as an aggregate, it assaults or diminishes the systems – including families, schools, businesses, and states – where personal identity, security and satisfaction are achieved. Every such system is, in atomist eyes, no less an aggregate than the passengers in a bus.

Linked to this we have the assumptions of neo-classical economics and behavioral theories of the firm that see each individual making decisions based on personal self interest. The idea is that society is formed by the the market that arises from the interaction of personal self interest. Various philosophies, such as utilitarianism have arisen to try and account for issues of public good and altruism. The basic general argument is that it is in our interests to do things for other people, on the basis that overall our own self interest will be satisfied. Prof. Bernard Williams excellent, short and eminently readable book Morality effectively demolishes this point of view. At the most extreme we see the denial that the word society has any meaning. This is Margaret Thatcher speaking in 1987 :”And, you know, there is no such thing as society. There are individual men and women, and there are families.”


Now that quote from Thatcher is normally shortened to “there is no such thing as society” which probably serves the purposes of politic critics (and I freely admit that I have always been one) but it is not complete. The phrase and there are families is significant. A family is one obvious type of identity and the best place to start a discussion of the subject. In philosophy the discussion about identity arose early and is associated with the idea of change. The ontological question is What makes this thing the same thing as it was before? and the epistemological question is How do I tell if it is, or is not? One of the best illustrations of this is the paradox of the Ship of Theseus (or Locke’s socks or Grandfather’s old axe, philosophers are great story tellers), given to many a first year Philosophy student. The paradox can be stated thus:

Theseus returns from slaying the Minotaur and the Athenians preserve his ship as a memorial. As a plank rots, they replace it with a new one storing the old planks in an adjacent warehouse. After a period of time all the old planks have been replaced at which point a local entrepreneur reassembles them into a second ship. The question: which is the Ship of Theseus?

I don’t intend to summarise all the arguments around this (and there are many and the debate is not resolved). I use it to illustrate that identity is not a simple issue, even for a ship let alone in human systems. It is not a simple matter of categorisation or aggregation. It is complex, fluid and incorporates issues of change, time and abstraction; it is messy. A few corporate taxonomies would benefit from considering this, but that is an aside.

I am still working on this but I think we can identify five (we may as well keep to natural numbers here) characteristics of an identity. I want to illustrate each in the context of family, but remember that a family is only one form of identity.

  1. An identity is not the same thing as a role. Within a family I may have many roles such as father, cook, picker up of cat sick, humane disposer of spiders, fault bearer (just keep adding them) which I or others perform at different times. The family has a coherence that is more that the aggregate of its parts or its roles
  2. An identity does not have rigid boundaries, nor is it susceptible of precise definition. When does one’s daughter’s boy friend become a part of the family? A cousin twice removed may be an intimate of one family and an unknown relative in another. It is a coalescence, with coherence but also with fuzzy boundaries.
  3. Identity is not absolute, it can change in context or over time although the point of transition (the establishment of a new identity) may not be clear either at the time or in retrospect. Identities can inherit characteristics from other identities and can also be polymorphic (represent themselves in different ways). In families people die, they divorce (well or badly) they engage in political disagreements (it’s not just people, it’s also abstractions), they may be separated by civil war or disagreements over probate. Families have histories that provide subtle or not so subtle.
  4. Identity in human systems is a strange attractor (to take a key concept from complexity theory). We have historically understood two types of attractor: a point attractor characterised by a pendulum and a limit cycle attractor where activity oscillates between different stable states. Complexity theory has given us a third, strange attractor often characterised by the Lorenz Butterfly. To quote Juarrero “All attractors represent characteristic behaviours or states that tend to draw the system towards themselves, but strange attractors are ‘thick’, allowing individual behaviours to fluctuate so widely that even when captured by the attractor’s basin they appear unique” and “Strange attractors describe ordered global patterns with such a high degree of local fluctuation, that is, that individual trajectories appear random, never quite exactly repeating….” A critical point to realise is that a strange attractor comes into being through evolution not design.
  5. Identity is established by robust resilience and if we understand identity from the perspective of a complex adaptive system then interdependence becomes more important than autonomy. A family tolerates failure, it carries as sense of obligation that is not just about individual choice. A family is never stable, it is resilient because it changes. In contrast something clearly independent is a category, associated with stasis and stagnation. This point (and others) is derived from a paper by Juarrero and some of these ideas are taken up in a recent Kurtz-Snowden paper .

This gives a general, but incomplete idea of identity and I offer it as a work in progress not yet a complete theory. In effect in any system the atomised individual has multiple identities which act as attractors. The pattern of our interaction and connectivity with those identities is in its own turn a strange attractor. The paths are constrained but only loosely, and are never the same, they have subtle differences.

I think this shows why the idea that communities are aggregations of individuals breaks down. The insights from complexity also couple with a greater understanding that our consciousness is a distributed function of our brain, the nervous and hormonal systems and the environment. Rockwell’s book Neither Brain nor Ghost is a good source on this.. In effect atomism, aggregation and categorisation is coming under assault from multiple sources.


I have argued for a view of human systems based on identity as an attractor mechanism, I have further argued that this is a more accurate account than that provided by atomistic accounts. I have also by implication argued that communities are not aggregations of individuals, but a pattern that either is an identity, or forms from the connectivity of identities. With this comes an argument against categories and linear concepts of causality.

Now there is a simple way to explain this, by reference to two types of psychometric test. One type (Myers-Briggs is an example) seeks to classify the individual into a category and having done so to be able to predict aspects of that individual’s behaviour based on the general characteristics of the category. This is an atomistic example. The other type is one that identifies orientations and is team focused (The earlier versions of BELBIN based on seven team characteristics). Here you receive a response which indicates which which characteristics represent your primary and secondary orientations. Your behaviour as an individual does not fit in a box, but it moves in a path determined by the strength of the different characteristics (attractors) in relation to each other, and critically to the other players in the space. The patterns that emerge from this interaction are always different and contextually appropriate – they are called teams.

So that is an example, but also a metaphor for understanding systems in terms of identity rather than atomistic aggregations. Now let us look at few of the implications of this through two examples:

  1. In Marketing we traditionally think in terms of market segmentation, i.e. categories. Now with increasing social mobility and globalisation the basic and simple segmentation no longer works. Companies are moving towards more categories and sub-categories in an attempt to manage increasing uncertainty. Ultimately this leads to one-to-one marketing and commoditisation .If on the other hand we map the identities in the market place we have greater economy of representation. We can attempt to amplify those identities which are positive to our product/service and dampen those which are not. We cannot predict the exact outcome arising from an intervention but we can now adopt a safe-fail experimental approach which enables a co-evolution of supply and demand. Our work here, which is now realised in software is based on taking a similar approach to understand the context and threat of asymmetric threat in anti-terrorism. You can’t understand or predict the activity and motivation of each terrorist. However if you can map the identity structures in play you have cognitive advantage over the enemy.
  2. The HR function in any organisation needs to understand the dynamics or organisational culture, both in a stable environment, but also during change and cathartic events such as a merger or acquisition. Trying to categorise all employees or seek to determine their individual behaviour (on the theory that organisational culture is the sum of its parts) is hugely expensive, not to mentioned flawed in theory and impossible in practice. In contrast if we map the identities in play we can focus our change programmes on amplification and dampening of identities, knowing that by doing so we influence individual behavior as a system. In a merger we can identify similarities and differences in identities between the two organisations. This then allows us to focus on making changes to the environment and influencing the evolution of a new shared culture. We do not try and design it in advance, we create conditions where we know the focal points around which culture will emerge and we seek to influence them.

I could go on to look at applications in strategy, military decision making and elsewhere but I think that is enough for what is already an over long post. However I do want to return to the question of natural numbers. By now my intent should be fairly self evident. If we take those numbers as valid, but apply them to individuals then we see natural limits based on the number if participating individuals. If on the other hand we realise that human behaviour is based around identities then we can manage based on different assumptions of trust based on identity linkages. In effect this is a nodal network, a large number of agents can be in play but they are orientated around self–organising nodes. A system with five nodes or less will behave differently from one with fiveteen and so on. Of course it is easier to amplify or dampen the impact of an existing node than to create one from scratch.

For those who want to know more about how to map identities, then its a new set of methods and supporting software and will be taught on forthcoming accreditation courses. The one thing I do know is that its a lot easier to teach it f2f than it is to write a blog!