The problem of modelism

Herman Kahn's thoughts on a common pitfall in strategy and planning

In 1957, Cold War strategist Herman Kahn wrote a paper with a co-author about the "ten common pitfalls" in military planning and strategy. 66 years later, it still holds up, particularly the lead-off sin he cited, "modelism," or "the use and abuse of models."

Some caveats are in order. This was a draft paper and as far as I can tell, was never formally published anywhere. Second, some of the examples and language are badly out of date (though the ideas that underpin the ideas are still fresh and useful). Third, Kahn leans towards the quant-side of strategy, which is to say those that prefer to make data-driven decisions (think late-50s “Moneyball”). But when you step back he’s concerned with improving strategy in “an uncertain world.” Aren’t we all?

So let’s talk “modelism.”

I’ve seen modelism at work in many different places. First and foremost is the academic tendency to mis-apply a particular tool or theory to explain some phenomenon in the real world. I remember hearing a Singaporean diplomat give a talk that not-so-subtly attributed the past few decades of peace in the Asia-Pacific to the peoples of the region being simply better at dialogue than the rest of the world—to which a geographer might point out that the vast blue distances between these maritime/island states and the lack of large-scale naval capabilities has something to do with it.

At the broadest level, we want the right tool for the right job—the right model/theory to fit the right problem/challenge.

And Kahn's fairly expansive in his explanation of this issue. "Instead of designing the analysis so that it really can answer some important policy questions (which may get one into some mathematically untidy questions) many analysts prefer to study only the interesting (to them) portions of the whole problem. They often end up by studying an irrelevant or over idealized question. Or what is sometimes almost as bad, the question that is being studied is relevant but not complete."

Kahn focuses on the problem-design phase—put another way, how we approach the challenge at hand, from the very beginning. He digs even deeper in an early footnote:

"We are indebted to Albert Wohlstetter for pointing out to us the extreme importance of emphasizing design over analysis. While the point may seem obvious, it is surprising what a difference it makes in one's approach to problems. For example, if one is studying the bombing of civilians, when it may be crucial to find out where the civilians are likely to be when the bomb goes off. If, however, one is designing good shelter programs, one can merely assume that the civilians are in the shelter. It is part of the design problem to figure out good ways to get them there. This last problem is not only simpler than the first one, but also a more fruitful one to work on. (It should be clear that well designed system is almost always easy to analyze. If its performance were not clearly satisfactory it could scarcely be a good system. Therefore, if one concentrates on design and is successful, the analysis often takes care of itself.)”

The ins and outs of civilian bombing research is far from my expertise. But the takeaway here that in order to defeat modelism, we ought to put far more effort into the front-end thinking.

I hate to rely on mis-quotes, but in this case two seem to apply. There’s a saying, falsely attributed to Einstein, that given an hour, he “would spend 55 minutes defining the problem and then 5 minutes solving it.” There’s another great one, similarly falsely-attributed, to Lincoln: “Give me six hours to chop down a tree and I will spend the first four sharpening the axe.” These are both fantastic, even if unattached to such famous individuals.

Think early to defeat modelism, and steer clear of a preferred or self-serving theory. Find the right axe for the right tree.