Dave Snowden

Babies should not be thrown out with bathwater

RSS Feed

201111172306.jpgFor some years now I been distinguishing systems thinking from complexity thinking in a variety of ways. Technically I suppose I mean systems dynamics, but the two are largely conflated these days. I should also make it clear that, unlike Stacy in his early books, I am not hostile to systems thinking, but I do think its basic assumptions make its various tools and approaches suitable for complicated but not complex thinking.

For me the main difference is that most systems thinking approaches, in particular those which are shall we say “popular” focus on defining an ideal future state, then seek to close the gap. You can see this in Porter’s approach to strategy, BPR, Sick Stigma and also Learning Organisation. In the latter case the ideal future state may be common values etc so its a little more fluffy than the more engineering approaches but the principle is the same. They also tend to focus firmly in changing individual behaviour; think about the amount of books on competences, leadership behaviour etc. etc.

Complexity thinking on the other hand, recognising that complex adaptive systems are not causal (in the common meaning of the term) but are dispositional in nature should focus on describing the present, then acting in that situated now to test and enable the evolutionary potential of the system in multiple parallel safe to fail experiments. Complex approaches should also focus on building collective capability through ritualised forms such as crews and network creation with techniques such as social network stimulation.

So in summary systems thinking is about closing the gap to an ideal future state using and focusing on individuals while complexity thinking is about understanding the present, and evolving through collective action to a future state which could not be fully anticipated but which is sustainable and resilient. Now I know its a lot more complex than that, and also that some of the greats before popularisation are probably turning in their graves, or their emeritus chairs when they see what has happened. However systems thinking is pervasive, and its linked with models and ideas such as memes and Dawkin’s attitudes in general, i.e. we have an emphasis on causality at the cost of evolution.

Now one of the more popular of systems thinkers is John Seddon, who I have heard several times but never met. We came close in Antwerp recently where we were both speaking at the same conference. John seems to be appearing at several Agile/Scrum/Lean events of late and his keynote at redev2010 is available for online viewing and has received some approving tweets. I suggest listening to this before reading on as I will assume some familiarity with it.

Firstly, there are a lot of very good things here that any reasonable person would understand and agree with. Regrettably most managers are locked into a perverted (Lacanian sense of the word) system that does not permit reasonableness or common sense. Ironically most of those practices have arisen from the same theoretical base but I will return to the ironies later.

The main thrust of Seddon’s criticism is that managing to reduce costs in effect increases them. Creating standard times for jobs and targets for completion just results in multiple job closures and start ups and multiple inefficiencies in the name of efficiency. All of this is well said, true and I commend for listening. I wouldn’t go as far as to rave about it and call it “magic – absolutely brilliant” but its good stuff. The attack on specialization and back office practices, the clear statements that services are not the same as manufacturing and that its all about flow are more than welcome.

The other important point Seddon makes is that managers have to discover the stupidity of their own processes for themselves. He does this by getting them to study the situation. Its not about a consultant coming and telling them, its about a consultant enabling them to discover things for themselves. Again I an in full agreement here, but I’d like to extend it a bit. Over twenty years ago I was pushing managers into field ethnographic work, getting them to take apprentice roles in the field to see how things were done. I think this is better than managers acting as internal consultants. We’ve also discovered that mapping information flows between decisions based on narrative audits (I really need to blog about this) then contrasting that naturalistic this is how it is result with idealised process maps can be very powerful. It forces contrasts into the system that prevent easy complacencies dominating management discourse.

You can probably start to see where I am going with this. I think Seddon is outstanding in diagnosing what is wrong, but I think he is weaker in determining what should be done in a way that can scale and is sustainable. More accurately, the interventions appear (from how he presents them so I could be wrong) over dependent on his presence. There is also a great irony when he complains that people called Ohno’s work at Toyota “Lean” when Ohno said that you should not name thing. Seddon says that this is an American marketing trick; his whole presentation starts off with advocacy for the Vanguard Method, which I think is a name!

The point where I think he gets is badly wrong is in his denigration of the transforming potential of tools (which include technology) in human systems. I also think there are significant issues on scale. The example he gives, which he uses a lot is a workflow system for repairs to housing stock. Its a good case and I’ve enjoyed it over several listenings in different contexts. Seddon firmly states that IT should not be involved until a manual system has been put in place which works and that they should be firmly discouraged from coming up with improvements based on technology capabilities.

Now this is understandable in a low cost system such as the one he describes, and one which is a very basic workflow. Many modern systems could not be reduced to T-cards even in design. This also ignores a key concept in complexity thinking namely co-evolution. In modern systems design we need to enable rapid co-evolution between user needs (often unarticulated) and technology capabilities is safe-to-fail interventions. I will blog about that later this week by the way. Seddon presents an either-or when we need to think more in terms of both-and which is a common problem by the way.

As I wrote this and listened again to the keynote I realised something. Seddon is in fact anti-systems thinking through practice. He has a lot of experience of what is wrong, and an ability to articulate that well. He has solid experience of what can be done differently and the charisma to carry it through in practice. However the theoretical underpinning are weak, or are not articulated which means its very difficult for the practice to effectively scale or adapt to changing circumstances. Co-evolution is, for me, key to understanding a new dynamic to the nature of IT system design and implementation and along with the three basic heuristics of complexity management (finely grained objects, distributed cognition and disintermediation) can provide a more sustainable way forward.

So my main concern about the Seddon keynote is that he is throwing out several valuable babies with the bathwater. More on alternatives tomorrow.

Top