Thursday, September 9, 2010

Deeper confluence: Explanation in the uncanny world

I'd like to introduce yet another new framework for sensemaking in this post, one I've been pondering for a while. This morning I woke up and said to myself, "It's time." So here we go.

Note: The framework described in this post has been superceded by an even better, shinier framework described in the next post. So don't read this! It's drivel! (Unless you came here looking for an explanation of the later drivel....)

Surface and depth

Let me start with a bit of background for those who haven't been following the story so far. I've been developing and using what I call the confluence framework (described here and here) for about ten years, earlier in collaboration with Dave Snowden (and his Cynefin framework) and more recently on my own. Just so you don't have to go look it up, here is a picture of the confluence framework.


A while ago, through a beneficial accident of multiple perceptions, I came to consider ways to think usefully about the ways in which organization and self-organization (hierarchy and meshwork) intermingle and interact. That led to this post and what I've started to call the "mixing" framework, which (again so you don't have to go find it) looks like this:


After that useful accident, I was juxtaposing these two frameworks (confluence and mixing), and I thought: Why stop there? Why not come up with a series of frameworks that together cover the confluence space more fully?

So I thought about what sorts of other frameworks might be useful to sensemaking across the space, and I came up with this:



What that diagram shows (badly because I can't figure out how to get PowerPoint to do what I want) is that each of four "deeper confluence" frameworks is most useful in one area of the surface-level confluence framework, and its appropriateness declines as you move away from that area.
  1. The mixing framework (orange) is most useful where hierarchy and meshwork mix the most strongly, in the upper right-hand corner.
  2. The explanation framework (red) is most useful in the lower left where connections are few and/or weak.
  3. The complex framework (green) gets more useful as you move upwards.
  4. The complicated framework (blue) gets more useful as you move to the right.
The way I see these frameworks being used in sensemaking is this. People might start using the simplest "surface" framework with regard to the situations or issues they care about. Then, after considering things at that level, they might choose one or more of the deeper frameworks and consider other aspects of the issues that are most pertinent to specific areas within the larger framework. Different issues and contexts and goals and groups will gain benefit from different constellations of deeper frameworks, so that the whole package of five frameworks can adapt to the needs of the moment. Also notice that in the middle area of the space, where it is most difficult to guess what is going on (and interpretations are likely to vary the most), all four deeper frameworks can provide help to sensemaking.

Explaining explanation

So, having the mixing framework already, my attention has turned in recent weeks to the lower left corner of the confluence space. This is the area where the connections between components are few, weak or both, and where prediction is impossible. I call this space by various names: the uncanny space, the space of sparse or loose connections, the space of isolation, the space of incoherent and undirected patterns, the void. I also like Tom Graves' term that this space is one of inherent uniqueness. In general the important thing about this area is that patterns do form here, but it is hard to say why, and we have little or no control or predictive power over them.

I've watched people address this uncanny space in quite a few workshops where we derived the Cynefin framework (in its confluence-like pre-boundary form). I've noticed two things people do with this portion of the space. Sometimes they avoid the area and place nothing there, essentially creating a limited framework. Other times they use it as a dustbin and put the most unpleasant or uncomfortable things there, apparently in the belief that things placed there need not be considered further because they are impenetrable. I've long felt that people need more sensemaking help in the sparse area.

As I started thinking about patterns in uncanny space, I kept recalling that old television series "The Unexplained." Paranormal activities, auras, crystals, divination, ghosts, life after life, angels, aliens, NLP, reiki, acupuncture, alternative medicine, and all sorts of things some people believe in and others dismiss, disdain or make fun of fit here (even a list of what fits here will vary depending on your beliefs -- excuse me if I offend anybody on either side!).

On thinking about it, I realized that the reason people call things in this area "unexplained" is not because nobody has explained them. It's because everybody explains them, but everybody explains them differently. My explanation is your un-explanation, because you reject it; and vice versa. So I began thinking about variation in explanation as important to this area.

As a sort of method, I looked up several phenomena and pondered the explanations of a variety of people (myself included) about them. I asked myself: What do people say when they are talking about these things? How do they explain these things? What are the elements that can be found in all such explanations? And how can a framework help people capture similaries and differences in those explanations? How can I help people explain explanation?

This is what I have so far. There are three questions I see people often asking and answering about patterns in the uncanny, sparse area of the confluence framework.
  1. Is this pattern true? Is it real? Is it valid?
  2. Who is behind this pattern? Who is doing this? 
  3. Is this pattern tractable? Is it predictable? Can we control it?
I noticed soon after deriving this list of questions that it matches Harrison White's three evaluations of social identity interaction, which came from observational work done by Bales on how people evaluate (or explain) the actions of others in social situations. The first question here is one of purity for selection; the second of power for mobilization; and the third of quality for commitment. (For more on White's identity interaction categories and their relation to sensemaking frameworks see this paper.)

Taking these three questions and making them into dimensions, I end up with:
  1. the dimension of value (purity)
  2. the dimension of agency (power) 
  3. the dimension of mystery (quality, though this is actually the reverse of quality, because I want to talk about the presence of unpredictability)
Trying the dimensions out in the case of auras, some people believe there is a mysterious force operating in the universe that only some gifted people can perceive (or that everyone can perceive with adequate training). Thus their explanation has strong elements of mystery and value, but little to say about agency. Other people believe there is no mystery involved in auras, only the agency of some smart people making a fast buck by duping some credulous people desparate for something to believe in. These are strong elements of agency and value, but little to say about mystery.

It's difficult to support sensemaking in three dimensions, so I needed to choose two to graph and one to superimpose. Here is the graph, with mystery and agency only.


I show mystery as a cloud, agency as an arrow (of intent) and the lack of either as a simple circle (for want of a better visual representation of absence).  Laying on the value dimension using color, we get these states:


High mystery, no agency. Here the situation is unpredictable for one of two reasons. It could be that science has tried to penetrate the mystery but has failed because of poor information or poor models. This is how we now explain the "dead ends" of science such as phrenology and the Ptolemaic solar system model (though it's not how they explained it then). Another explanation might be that the mystery goes beyond science and can never be predicted. Something like what happens after death might fit here (in some worldviews). The unpredictability of chaos theory might go here (in some worldviews). The colors on the graph show evaluations of value: positive, neutral or negative. I've put the "beyond science" label as neutral, but that's only my view.

High mystery, high agency. Here we find mystery and agency mixed together. I call this area the area of mastery, because mastery of a skill or subject always contains some element of mystery. This is where people say "there is an art to it" or "you have to be born to do this" or "some people have a knack for this" and so on. Many aspects of scientific practice (not theory) fit into this area. True mastery is where the mystery and agency come together to create useful results. False mastery is where the connection misfires and the result is not helpful but harmful. Hitler would be a good example of false mastery: he was by all accounts a master of all sorts of things, but with the worst possible outcomes.

High agency, no mystery. This is the area where positive explanations will highlight true skill and ability and where people say "there is a science to it" and "anyone can do this if you follow these steps." This is the area of recipes and instructions. Negative explanations in this area will reference agency with negative purposes and/or outcomes: lies and the lying liars who tell them. By the way, by using the term "agency" I mean to refer to any agency, meaning the actions of any intelligent being, not just human beings. Some highly religious people (well, at least in some religions) would place many or most of their explanations in areas where the agency of a higher power is the dominant explanatory element. Again since my goal is to build a comparative tool, not a way to get at "truth", the point is explain explanations, not create them.

No mystery, no agency. Here positive explanations will highlight science as it should be, or science when its work is done. Causes have been revealed, experiments have been replicated, and predictable results can be relied upon. Here also are the negative explanations where people make mistakes, not because of deceptive agency or intervening mystery, but simply because they are limited and forget things.

Now remember, as with the confluence framework, these are not descriptions of broad bounded regions; they are caricatures of extreme conditions. Most real explanations will fit somewhere between these extremes.

Comparing explanations

Now let's try this out with some stories (which are essentially bundles of explanation). What I've done is to place four linked explanations here, from different perspectives.


I start with Caisson disease on the Brooklyn Bridge, a pattern I find endlessly interesting. (To put it very briefly, workers on the Brooklyn Bridge got the bends, and they didn't know what it was or how to stop it from happening. More about it here -- scroll down to where it says "Caisson disease.") From the point of view of the Brooklyn Bridge engineers in 1860, the situation was so high in mystery that it went beyond (the current state of) science. From my perspective, what happened in the caissons in 1860 was also in the area of mystery, but for another reason: the science was wrong (thus the box is red). From the perspective of anyone getting the bends today, the situation has been removed from mystery and rests in the firm bed of well-known science.

Now look at pattern number two, a traffic jam (say one that arose from rubber-necking around a broken-down car well off the road). From the perspective of an irate driver in the traffic jam, the situation is negative (red) and most likely involves a high degree of agency (some idiot screwed up somewhere). From the perspective of a traffic engineer, the same traffic jam involves a little agency (the original designers could have improved traffic flow through here) and a little mystery (hard to tell how irate drivers will react to these things).

Pattern three is about auras. A true believer in auras who believes they can see and use them would place the situation high in mystery and high in agency, and mark it as positive (green). They might see themselves as masters of their gift. I myself would place it fairly high in agency, but in a negative sense (deception and self-deception combined). I do however allow for a fair amount of mystery, in that the science might be wrong and/or that there may be more we have not discovered about the idea of auras. (Who knows?)

Pattern four is my very useful pattern about synesthesia, which is something many people think is made up and silly, but I conveniently have and so can use it as a counterpoint to the auras. I place synesthesia fairly high in mystery, even though I've read the neurological explanations, because I find it pretty amazing. Unlike the aura believer, I don't place it high in agency, because it's not so much something I do or use as something that happens to me. Now the skeptic about synesthesia might believe that I and others are making this up to gain notoriety, or perhaps we are deceiving ourselves in thinking we have "superpowers" and so they would place their (negative) explanation far into agency and with little mystery involved.

Here's one more pattern, though I didn't draw it on the diagram. When I was thinking about this I kept going back to the old Scooby Doo television show I loved as a child. In every episode of Scooby Doo, the group encountered a scary mystery that seemed to have no explanation. Through a combination of painstaking scientific inquiry and silly bumbling, the group always found out that the mystery could be explained either by a simple scientific explanation (the moaning cave is just the wind blowing) or by the agency of a person (usually it was somebody who wanted to scare people away from their nefarious-purposes lair). The first of these moved the mystery down into the comfortable lower left of my explanations framework (no mystery no agency), and the second into the open-and-shut-case area of the lower right (no mystery high agency). I always waited for the Scooby Doo episode in which they concluded that they had found a bona fide mystery, but it never happened. Pity.

Moving forward, whatever that means

I'm excited about the possible utility of such a surface-and-depth system for sensemaking support, and have been talking with various people about building and testing and refining it in theory and practice. In particular I'd like to thank Tom Graves, Carol Mase and my husband (yes he has a name, it's Paul Fernhout) for discussing these frameworks with me and offering helpful suggestions.

Now let me say something about frameworks in general. I know that lots of other people have developed and are developing other sensemaking frameworks, and sometimes I think these things I am writing about may just be irritating to people who are tired of navigating all the various theories and models out there. Why bring another framework into an overpopulated world? I used to think that way about writing: if there are so many great novels in the world, why do people keep writing them? They should all stop so we can catch up!

But then I realized that some of the novels that have most spoken to me have not spoken to other people in the same way. For me Dostoyevsky without The Idiot would be much less of a touchstone; but for others it is The Brothers Karamazov that stands out, for their own reasons. Should somebody have stopped Dostoyevsky from writing  The Brothers Karamazov because I didn't get that much out of it? Or The Idiot because others didn't? I've come to understand that for every good book there is an audience who needs it. In the same way, for every good sensemaking framework there is an audience who needs it. It has occurred to me that it might be just as arrogant to think nobody can gain benefit from your creations as it is to think everyone can.

So I press on, not knowing what will come next, and not really much minding if the end result is just me talking to myself. (There's value in that too.) My end goal is to help people resolve conflicts and solve problems together, in whatever ways I can, so I'm eager to continue working in this area. I'm also eager to work with others who are interested in the same things. As I've said elsewhere, I'm not much interested in models or frameworks if they can't provide benefit to real people dealing with real problems, so I'm the most keen on testing and improving these ideas in practice. If you are interested in discussing these ideas or improving them in practice, please do let me know.

4 comments:

  1. [apologies - too long - will have to split this into two parts]

    Great work as usual, Cynthia - thanks.

    A variety of comments that I hope may be useful:

    On 'mystery': perhaps take a look at how Roger Martin uses the term, in contrast with 'heuristic' and 'algorithm', in his book 'Design Thinking'. In classic Cynefin terms, you could say that he places 'mystery' in the border-region between Chaotic and Complex, 'heuristic' between Complex and Complicated, and 'algorithm' between Complicated and Simple (i.e. rule-bound). Might well be useful for cross-comparison here.

    On 'agency': another way to put 'degree of agency' in your frame might be inanimate (i.e. 'natural forces') versus animate (result of conscious choices, human or otherwise, as you say). We could note, for example, that 'Caisson disease' was _not_ attributed to malign spirits etc - in other words even at the earliest stage it was assumed to be the result of some unknown yet natural cause. Hence in terms of your model, a natural-scientist would tend to look for explanations over on the left; a social-scientist (or, for that matter, a religious person) would tend to look for explanations over on the right.

    One difference between our approaches, perhaps, is that (as you say above) your aim is to _explain_ explanations, whereas I'm more looking for ways to _use_ explanations - the Scientist versus the Technologist, if you like. In a technology perspective, explanations and other beliefs are tools that enable different choice: to quote psychologist Stan Gooch, "things not only have to be seen to be believed, but also have to be believed to be seen". So whilst to some people the focus in assessing a model is whether or not it is 'true', I'm more concerned about whether it's _useful_, or effective, for any given purpose - which, recursively, means that I also need some kind of frame in which to identify the current purpose!

    Hence two other themes here: recursion and dynamics. Recursion occurs when we use the frame to explore itself (or, in this case, explain itself). One of the key points in a frame about explanation is repeatability and predictability: the whole point of an explanation is to predict what would happen again under equivalent circumstances. So in scientific terms any unique event is a 'mystery of mysteries', because there's no repetition; in religious terms the same event might be deemed a 'mystery of agency', evidence of the hand of some discarnate entity; and so on. Like many enterprise-architects, I've always found this kind of recursion to be one of the most powerful and practical ways to use the Cynefin frame; it seems it could well be useful here.

    And as with Cynefin, the dynamics would seem to be extremely important here - using the Confluence frame to explain and map the means and/or sequence via which people _change_ their explanations. In your last diagram above, each of your 'colour-change' relations is also a dynamic - not solely 'positions', but how someone might move from one position to another. Which again is often partly recursive, because we can use the structure of the frame to map ('explain') each step in the transition.

    [continues]

    ReplyDelete
  2. [continued]

    One real point of concern, though, is your colour-code, which in essence is a scale from 'grounded belief' ('true') to 'ungrounded belief' ('false'). The way you've framed that does kind of reflect your preference for scientific over technology over anything else, and again doesn't seem to allow for natural uniqueness or personal difference of experience and interpretation - would perhaps be wise to be wary of the tendency to subjective feedback for reinforcement of (literal) prejudice in the way it's currently framed? Perhaps needs more emphasis that it's about subjective _belief_ rather than objective 'fact' - e.g. that green represents a _belief_ that something is true, and red likewise a belief that something is false, rather than that it 'is' or 'is not' true or false in any objective sense.

    I'll use your point about synaesthesia versus auras as an example for all of this. The _experience_ (i.e. personal-belief/explanation) of synaesthesia is well-documented: for example, almost the entire terminology of wine assumes or implies some form of synaesthesia. Credible explanations exist, mainly centred around various views of 'cross-wiring' in the brain and suchlike. The catch is that it is a highly _personal_ experience: hence easy for those who _don't_ experience it to place it under 'false agency' (upper right on your diagram). To move it to 'scientific explanation' (lower-left) implies or requires strong dynamic forces - especially since, given that it is a near-unique _personal_ experience or attribute, there is little to no way for them to experience it themselves.

    Synaesthesia usually implies one sense being used as a _direct proxy_ for another sense: colour for sound, for example, or texture for taste. But so-called 'auras' seem to be somewhat different: in my understanding and (very limited) personal experience of the respective phenomena, they seem more an _indirect proxy_ for merged or interpreted information - not a direct cross-sensing but a perceptual _overlay_, much like current development of so-called 'augmented reality' on mobile systems. Hence, for example, perceived-colour as a proxy for overall health. Something similar occurs in dowsing, for example, where small movements of the hands are leveraged and interpreted as a proxy for other information provided by weighted-sum cross-merging of other information. The catch in both cases is that the source-information may be any combination of objective or subjective - in other words, anywhere on your green/red ('true'/'false') spectrum. Which leaves the entire phenomena wide open for interpretation all the way from false-positive ('newage') to false-negative ('Skeptic'). Probably the only way out of that impasse is personal experience - which again may make little sense to others (i.e. placing it at the 'high-unpredictability' or upper section of your frame).

    [continues]

    ReplyDelete
  3. [continued - part 3 of 3]

    One personal experience of my own may illustrate this. Some years back I had a serious accident which left me in a hospital bed with several broken ribs. As per standard practice, I'd been given a fair dose of morphine to tackle the resultant acute pain. One bizarre side-effect of the drug was that every time someone spoke, I saw the words as if written on the wall, with each person's voice in its own distinct speech-bubble. Each person also had their own distinct colour-pattern, together with kind of extra overlays on each voice that I felt/experience as distinct sensations on my skin. I could also clearly sense the direction and movement of each person in the passageway outside the ward, even though I couldn't see them - I perceived them as a dark shadow _through_ the wall. Until the morphine wore off, some hours later, it was exactly like living in some kind of semi-psychedelic cartoon-strip...

    Under those circumstances, any question of 'real' versus 'imaginary' makes little sense: any boundaries between them becomes blurred (your 'And' versus 'But', in your 'Mixed' version of the frame above), so much so that 'real _and_ imaginary' is the only term that does make sense. It's not repeatable (or at least, I _hope_ it isn't! :-) ), and by definition it's subjective, yet in those specific terms the respective feelings and experiences _are_ themselves 'facts' - whereas any interpretations (i.e. 'explanations') are not.

    So to make the Confluence frame even more useful, it would seem worthwhile to explore how to recurse at least one more level, where we can use 'subjective' versus 'objective' to map and 'explain' the explanations and dynamics of explanations.

    Hope this helps, anyway.

    ReplyDelete
  4. Tom, far from ignoring these comments, I find them so useful that I cannot possibly reply only in the small space of a comment - so they are informing my next post which is coming soon ...

    Cynthia

    ReplyDelete