1514966313-featured.jpeg
Dave Snowden

Scale

RSS Feed

The third aspect of anthro-complexity I want to discuss is the complex and problematic issue of scaling. Now this is something I have written a lot of posts on (one extended 2014 series here), but I am still refining the message and my own understanding. So read the prior posts but I am not going to fully bound by them or consistent with them after three plus years of practice, reading and thinking. The issue is profound in human systems given the nature of our society, our curiosity and out ability to learn. The historical process of innovation has been or partial or incomplete copying and improvisation to allow novel to emerge. We learn from, but are not bound by the past. As practice stabilises it increasingly standardises, acquires ‘generations of knowledge’, professional practice and the like. The apprentice model of learning means that the apprentices are making mistakes, talking between masters and generally developing the body of knowledge it is never static.

But all of that takes time and in the modern organisation or government we don’t have time so there are two temptations that people fall into:

  1. They take one successful pilot, created with time and focus in a specific context, and try and replicate that practice with less effort in multiple contexts.
  2. They aggregate multiple practices that appear to work of themselves and assume that they will combine in the whole without loss of value.

Now this is problematic in that it ignores context and also misunderstands complexity. Brian Arthur in his seminal 2009 book on evolution in technology clearly establishes that innovation happens at the right level of granularity where combination and recombination through novel linkages is easy. Too coarse a level of granularity and the stabilities are too high, novelty is reduced. To fine a level of granularity and any new coherence is impossible. Combination and recombination allows rapid adaption to context and also provides some power to local subjects and practitioners to adapt.

Working in complex solutions, with parallel safe-to-fail experiments, we automatically get the granularity right and as I keep trying to explain to people individual experiments do not succeed or fail, the merge, mutate and change the theatre in which they operate. What we need is diverse access to different ‘grains’ and an understanding of how they can interact, or how to modify them to allow novel interactions.

The second temptation is more problematic but we can design for it. By breaking practices down and formally designed the input/output hooks we can enable a situation in which they will combine and a non-aggregative assembly of capability will emerge. That means we have the benefits of some standardisation and repeatability (at the optimal level of granularity) without the over constriction of aggregative techniques.

Humans have unique capability here, and a lot of it is aesthetic in nature, but that is for a future post. Overall the french word bricolage is important here, from the verb bricoler meaning ’to tinker’. It now refers to the idea of improvisation in human systems and is best in operation with loose knowledge bases with novel insights and applications. Over engineered solutions, despite good intent, make things worse not better. In a sense the human unique of being able to impose order here is taken to excess and produced the opposite of what is intended.

Top