Dave Snowden

A brief conversation with Gary Klein

RSS Feed

I’ve been under the weather this week, starting with flu and moving on to a poisoned toe. Interesting reflection there, if I had been born a century earlier then the only solution would have been amputation. Mind you the impacted wisdom teeth would have probably got me twenty years earlier Either way most of the week has been waiting for the antibiotics to but in and trying to elevate the foot. I really shouldn’t have made trips to the Universities of Surrey and Birmingham; good audiences but recovery was set back and I am way behind on email, papers and the blog.

The contrail free Wiltshire skies have been one compensation; we sit under the transatlantic routes so you normally never get a perfect sunset. However I am now waiting to find out if I can make Nice on Tuesday, or more critically Denver on Thursday. Another compensation was an exchange with Gary Klein which is reproduced below. Gary and I first met in Singapore when the Ministry of Defense set us up for a debate and were disappointed that we had a rushing clash of agreement (that will give you the context for my second answer below). Gary is one of the seminal thinkers and authors on whose ideas I drew in creating the whole body of Cognitive Edge methods and its been a pleasure to have worked and talked with him since that meeting.

One of the reasons its a pleasure is that you can have a disagreement with Gary without the protagonists taking umbrage.Gary agreed that I could share the exchange here and has promised/threatened a further response! Although its short, its an example of many conversations that are now taking part in respect of language. I have another post shortly on Clay’s use of the word complex (in my terms he means complicated) which will be a further illustration. Paradigm shifts in concepts require mediated changes in the use of language within both academic and practitioner fields. Here you see an early example of that dialogue.

(Gary in bold, my response in italics For background on the meaning of abductive see here. )

You made an important point that abduction is a means to make connections. But I think deduction and induction are also ways to make connections.

Agreed, any way of looking at things will produce novel connections and insights to a degree. The issue is the nature of the insights. I would assert that deduction and induction are a priori inclined to discovery with the realm of the probable, but will less good with plausible. That is to say they will generically allow discover of connections within the field of insight, but true novelty coming along by accident rather than design. Abduction is far more likely to create completely new insights.   Post hoc ergo propter hoc would normally apply, but in our approach volume and self-signification mitigate or remove that.

You made a different point along the lines that causality becomes problemmatic when dealing with complexity. I disagree. Causal inferences are still operating. But they aren’t operating in the simple cause/effect links or even the simple chain reaction/domino fashion. That’s the point of the research I’ve been doing.

Maybe we should offer to debate it in Singapore! I think this may be a terminology issue. If the system is chaotic/random then agent behaviour is deterministic, which means I can use statistical instruments. If its constrained, then the constraint structure allows predictability/repeatability. Strong constraints produce linear causality, weaker constraints provide repeatability that may be non-linear. However the moment you get the phase shift into a coevolutionary relationship between agents and system then there is no repeatability expect by accident. In that context there is no meaningful causality. That said, a landscape shows the evolutionary potential of the system so it allows strong statements to be made that will becoherent with the evidence presented by the landscape. I think (from a hazy memory) that in my language you are dealing with coherence rather than causality? Needs more data on this one, but that is a gut reaction.

We talked about Trial and Error learning as a 4th type of reasoning. You asked if I could come up with examples that weren’t deduction, induction or abduction. Here are some: conducting experiments, as well as shaking the tree to see what happens (taking an active stance to sensemaking).

mm, not so sure there. Conducting an experiment is generally hypothesis based so induction. Shaking the tree I would have classed as an abductive technique!

Top