Second (and third) thoughts: Second-order cybernetics and deconstruction

Graham Harris v3By Graham Harris

The old saw says “many a true word is spoken in jest.” and this has been proven by Terry Pratchett who wrote the following in one of his witty Disc World novels:

“First thoughts are everyday thoughts. Everyone has those. Second thoughts are the thoughts you think about the way you think. People who enjoy thinking have those. Third thoughts are thoughts that watch the world and think all by them selves. They’re rare and often troublesome. Listening to them is part of witchcraft.” (A Hat full of sky, 2004, Disc World novel #32, footnote 2, p. 74).

It was Nick Winder from Newcastle University in UK who put me onto this quotation. In a talk to a conference about tipping points in Durham in 2012, he talked about the ways we think about what we think about.

As I have written, ways of thinking about systems have been dominated by first thoughts. It is commonplace to think about systems in terms of closed, final, simulable systems where the feedback loops are inside the system – so inputs give rise to predictable outputs and Newtonian dynamics determine our approach to cause and effect.

Heinz von Foerster – who was the rapporteur for the Macy Conferences after the War – called these trivial machines (closed, final, specified by states and processes) and this approach to the world has dominated the world of modeling and simulation, particularly since the invention of the digital computer. Nick Winder has called this the first tenet of Jonah’s Law: “you can only predict the future of systems you cannot change.”

Second order cybernetics is what Niklas Luhmann (in An Introduction to Systems Theory, 2002) has called a “theory of observing systems,” so it is a form of second order thinking. Second thoughts place a reflexive observer in the loop – so there feedback loops both inside and outside the system, and there are recursive interactions between output and input. Second order cybernetics considers the system and its environment – indeed the two are inextricably linked. These are von Foerster’s non-trivial machines (open, equifinal, complex, recursive; they contain entities that are alive, observing and thinking about thinking).

Essentially what Luhmann did was to deconstruct the trivial systems theory and to base a second order reflexive theory on, “thinking about the way you think about systems.” So second thoughts become the second tenet of Jonah’s Law: “you cannot predict the future if you can change it.” Despite all the fuss about deconstruction – especially in sociology and the humanities – and the snooty attitude of science to it, this is really all deconstruction is; it is just, “thinking critically about the way you think about things.”

In deconstructing the trivial systems theory, Luhmann (2002) defined a system in a different way (there is no environment without an observer), based his theories partly on Humberto Maturana and Francisco Varela’s concept of autopoiesis (a system determines its own operation; states are a function of operations not vice-versa), and brought in the concept of differential form (communication is more than information). In the second order view, information is more than a rationalist statistical concept; it is instead, a “difference that makes a difference,”; (an idea first stated by Gregory Bateson during the same Macy Conferences).

The deconstructed approach to systems theory turns the usual physico-mathematical theory on its head and focuses not on states but on processes and relationships (this is Rosen’s approach from Thinking Systems #3) and places great emphasis on meaning and the role of interaction, observation, communication and recursion. This is not – horror of horrors – lousy relativist thinking, or even witchcraft; it is science, but science of a different kind.

[I will note in parentheses here – to anticipate later blogs – that this is the kind of science that mobile communications infrastructure now enables in the analysis of social media, network structures, Twitter feeds, “Big Data” and the like. In some other spheres of science we have almost none of this kind of data – but that does not mean that eventually we won’t be able to acquire it – but more of that anon.]

Uncertainty is fundamental to any complex system. Uncertainty is not just “noise” to be controlled and averaged out; uncertainty is a sign of process and internal dynamics. In her book Complexity: A Guided Tour (2009), Melanie Mitchell showed how organisms (like ants, for example) sample the statistical properties of their environment by, amongst other things, counting the frequency of communications (meaningful interactions) with other individuals. (To have meaningful features the environment requires a reflexive observer and different observers observe different environments.) Order at the level of the ant colony emerges from these interactions without there being any central processor: there is not, and need not be, any central “meaner”. So system properties arise from the bottom up through probabilistic encounters and communication. Individual agents work together in parallel to create meaning in the context of higher-level constraints. Outcomes can change as events unfold.

Sampling their environment – both by individuals and by the entire colony in second order cybernetics – requires both randomness and probability to explore and exploit the world. Agents must be anticipatory systems (i.e. they must have an internal model of the world they live in – derived from evolution and contingent history – so they may define meaningful interactions; Rosen, 2014).  So these systems are open, equifinal and they show development: they work through communication and meaning, there is no fixed solution and the future is recursive and unpredictable.

This model of interaction, meaning, and differential form can be applied to living systems at all levels: molecular biology, metabolism, the immune system and eco-systems. But a second order cybernetic model of uncertainty, noise, information, meaning, context and design is totally different from a first order (naïve realist) model. It raises a question I shall return to: what then is a system and what model or theory is appropriate?

Third thoughts deconstruct our ontology and epistemology and develop ways of thinking about thinking about thinking about things. Third thoughts are evaluative and consider things like ethics. In isolation from such a context, second thoughts can lead to relativist thinking so that Luhmann, for example, saw no need for ethics; indeed he saw statements about “oughts” as pernicious. Not all agreed with him. Katherine Hayles, who was trained as a chemist, sees the need for second thoughts to be constrained by evaluative thinking – there are facts about the world that are highly relevant at this time. There are global and other constraints on human actions and we are social apes: there are “oughts”.

In his 2004 paper, Towards a theory of knowledge systems for integrative socio-natural science, (Research in Human Ecology, 11: 118), Nick Winder recognized three knowledge communities distinguished by their different approaches to solving system-level problems through policy-relevant research. But first he noted: “the minimum requirement for participation in this work is an acceptance that science is not, as naïve realists believe, a personal quest for objective truth.”

Reductionists use research to work towards a definitive solution. Scientists and engineers tend to fall into this group. Constructionists are theoreticians – usually social and political scientists – for whom research is a way to find a consensus around an agreed problem. They tend to be critical of reductionists and their methods.  Deconstructionists believe that any attempt to define a well-posed problem presents us with severe ethical and intellectual problems. Critical humanists would fall into this group. Winder notes that each of these groups has coherent beliefs that have special strengths. Constructionists formulate new problems, reductionists seek new knowledge and implement it and deconstructionists evaluate.

So again I seek a middle way by employing all levels of thinking to discover facts about the world but also being aware of contexts, communications, meaning, and ethics.  First, second, and third thoughts are all essential components of a successful systems theory. There is no magic in this – wizards are not required – we merely need to extend the ways we think about thinking into new avenues. We must not allow first thoughts to dominate our thinking: we need to develop more sophisticated and higher level thinking skills to tackle many modern problems.

Leave a Reply

Your email address will not be published. Required fields are marked *