Monday, July 5, 2010

Of hypotheses and tools, models and frameworks

My thanks to everyone who has commented on the Confluence and Better confluence diagrams posts. One particular comment on the Confluence post (by Paula Thornton, twelveth in the list, starting with "Seriously") got me so motivated to write (thanks Paula!) that I ended up writing something too long for a comment; so I moved it to this post (possibly one of my shortest ever).

Paula asked about the "known" and "simple" labels on the Cynefin framework and what the framework had to say about situations where things that were "believed" to be known were in fact "apparitions" of truths, which people may believe in something in the face of evidence to the contrary, and indeed people may be paid to defend the appearance of truth. She also questioned the "validity" of the label "simple"  and "challenged" me to "find such a reality and offer it as evidence."

To start, the confluence model can't address the terms "known" and "simple" because it doesn't use those terms. Those belong to Cynefin only. The terms I prefer are Manuel de Landa's: hierarchy (ordered structure) and meshwork (unordered structure). So in general I leave it to those in charge of Cynefin to explain what they mean by "known" and "simple." However, I found Paula's points to be so interesting that I wanted to address them anyway.

What-Is and What-Appears

The term "model" implies that something "out there" is being represented or imitated "in here" in order to better understand it. In an ontological model, one that represents What-Is, anything "known" must be absolutely known, not just "'believed' to be known." With this sort of model, one must, as Paula suggests, "find such a reality and offer it as evidence."

But of the four models mentioned in the Confluence post, only Strum and Latour's social-link model is ontological in nature. This is because it is a scientific hypothesis, and hypotheses are all about What-Is. The medicine wheel, the confluence model and Cynefin are not hypotheses. They are tools. The purpose of a tool is not to explain but to augment understanding. For that reason these models mix ontology and epistemology, What-Is with What-Appears. What-Appears always comes with a colon attached: To whom? Where? In what context? After what? With whom? And so on.

With an ontological, What-Is hypothesis it would be ridiculous to compare what is known and what is believed to be known. If people differ in their beliefs about what is known, one is right and one is wrong, and offering evidence is the only way to decide which is right. But with a tool whose purpose is augmenting understanding about complex topics we can and should and must mix What-Is with What-Appears: not because it's how things are, but because it helps us achieve a goal. Hammers don't explain nails; they pound them in.

To illustrate this distinction, say I am an architect and I am building you a house. Let's say that in order to get a better idea of what sort of house you want, I provide you and your spouse and children with a set of blocks and ask each of you to build your dream house. We then discuss together how we can reconcile your ideals with the realities of the site and your budget. After this process is complete I build a blueprint of the house that will eventually be built. The blocks we use, and the toy houses you build, are tools we use to augment our common understanding of what you want in a house and how we can build it together. These blocks are like the Cynefin model, the medicine wheel and the confluence model. They combine What-Is (the realities of architecture and construction) with What-Appears (your ideas, fantasies and priorities, and possibly your differences of opinion on these). The blueprint I create, in contrast, is a plan, a hypothesis if you will, about what we can and will build. It is about What-Is alone. If a blueprint had lines on it that said things like "Tommy wants a big rocket ship here but Mom would like a breakfast nook" it would be difficult to build a house based on it. But these elements are critical to the sensemaking process.

This is why it is a legitimate and recommended, one might even say required, use of any sensemaking model to explore how things look from many perspectives; considering many aspects of a situation; from different ideologies; in different social classes; with different motivations; and from the perspectives of what is known and what is believed to be known, at the same time. This is not a weakness of such models; it is their strength. It is exactly what they are for. 

Native Americans talk about "walking the circle" of the medicine wheel, or considering many perspectives on the same issue. Indeed most sources I've seen say that the medicine wheel cannot be used without walking it. (I see this as similar to my statement that bounded Cynefin doesn't exist in the abstract but only in sufficiently detailed context.) Walking the wheel shares attributes with the sensemaking activities we use when we consider stories and use any of these frameworks (or others) as mapping devices.

So when Paula asks the question about what "is 'known' in a business and how is it 'known' and who qualifies its validity," the answer given by these models is not "this is known and this is not, and this qualifies its validity." The answer given is "yes, yes, let's explore that."

The confluence I am most excited about between these models is not a confluence in ontology. Agreeing on What-Is about everything is impossible and of limited utility, in my view. What excites me is finding new augmentations to our augmentations. These provide the greatest improvement to our collective ability to make sense of our world and improve it. It is a tragedy beyond words that such powerful tools as the medicine wheel came so close to being eradicated, and we all owe a debt of gratitude to those who have kept such ancient tools alive, and I hope we can correct our errors in time.

On terms

I have often wondered about whether those ISJ reviewers were right to say we should avoid the use of the term "model." It certainly does conjure up images of blueprints and other noun-ish representations of What-Is. And I've seen many interpretations of the Cynefin framework as a representation rather than a tool over the years, and they always make me unhappy; because when any of these tools are used simply as representations their power to augment understanding folds up into a tiny seed of untapped potential. Wikipedia says conceptual models "are used to help us know and understand the subject matter they represent." So that should work. But "model" on has no matching description; it seems limited to models as representations. So "framework" still seems the best compromise as a communicative device. What do you think? Is there a better way to connote the intent of such tools than either model or framework?


Unknown said...

Great post Cynthia. I have mentioned to you before that one of my favorite aspect of the paper you wrote with Dave was the notion that the framework does not exist in the abstract. For myself, I do prefer the term framework, but I often refer to it as an "emergent framework." Yes, this usually perplexes people more, but it then offers me a way to talk conceptually about the framework while maintaining that "creating" one requires some type of interactive "sense making" process(es) from which the framework can emerge.

In my opinion this is the most difficult aspect of explaining the framework. In your previous post you placed items on the framework to provide some examples, while helpful they can also be confusing because it makes it appear as though you are placing them definitively in a specific spot.

I see the entire framework as negotiated space. Even when you look at the "simple" or "pure hierarchy" area, what is known to one may not be known to another. Each of us have our own framework, and any group framework is a negotiation of meaning for that specific context. This is exciting because it offers (depending on how sense making is set up) the opportunity for many voices (from all levels of power) to get incorporated into a general understanding. And, if the process is done ongoing that understanding can change and grow with the community context. Those are a few of the a-ha's your recent posts have inspired for me. Thanks so much for you incredible willingness to share your thoughts.

Cynthia Kurtz said...

Stephen, thanks for this useful point! Yes, yes, yes, the beaver-dam-etc positions on the confluence diagram in the previous post should not be seen as definitive. I should have been more clear about that and thanks for pointing it out!

In fact it was such an important point that I posted a new blog post about it. (Hope you didn't mind that I put some of your comment in it - it was that good :)

Thanks again!


Paula Thornton said...

The biggest problem with the model is in the fundamental premises of complexity and simplicity. They are not different dimensions, but instead, one in the same: