Art leaps back to blogging with the sort of heavy post that makes me wish he’d post more regularly. Go read it.

He wraps up his discussion of perception, assumptions and one’s ability to see with this:

Here’s another thought that I should probably ‘unpack’ more fully in another post. Mental models are reflected in, but less often challenged by analytical tools. Oh sure, insights can most definitely be had with them, but I’m talking about much bigger constructs. Sophisticated computer models, for the most part, tend to only amplify the force and reach of our brains. They do not, as a general rule, make them better able to ‘think different’ (ly) and develop new mental models for seeing the world — for assessing the big risks in truly complex adaptive systems or for finding innovations.

This is a very important point. One can use formal analytic models to challenge one’s self to think differently, but this rarely happens.

The potential, however, is significant. One can use formal models in decision analysis where you reflect a decisionmaker’s assumptions back at him. You instantiate his assumptions and then show him the consequences of them in the simulated environment. Then you compare the results to another simulation with different assumptions. Or you compare the results of the same assumptions in different simulations.

It comes down to how one thinks about using analytic tools. If you think of them as a way to help you consider alternatives and formalize your assumptions, then they can help you to recognize the limitations of your assumptions. At the most fundamental level, this is what mathematics allows one to do - it lets you think rigorously about the essential assumptions for an assertion and allows you to contemplate alternative logical systems based on different assumptions. If, however, one gets married to one’s models and methods and forgets WHY one turned to those models in the first place, then you get the result Art describes. Incidentally, this is why I shifted my career away from a narrow specialization in operations research - I have great affection for the field but I came to the conclusion that due to academic drift it had become an insufficient basis for contributing insight to issues of national security (largely due to the lack of debate I saw over precisely the issue Art raises).

We need thinkers who are utterly conversant with analytic tools but are not defined by their tools. Many organizations make artificial distinctions between those who are quantitative specialists and those who are qualitative specialists. Making this artificial separation makes it awfully difficult to use analytic tools correctly, because the qualitative specialists either ignore them or accept them as black boxes, while the quantitative specialists aren’t spending enough time interrogating their models with the actual dynamics of the world.

I used a Boyd allusion in the title. The potential of analytic models is to formally represent one’s orientation and then compare it to an alternative orientation. If all you do is instantiate your own assumptions, then you haven’t gained much - all you’ve done is automated your orientation (perhaps allowing faster actions to flow from observation) without necessarily improving the alignment of your orientation.