Hard to vary

Written by  on March 29, 2016

While trying to compose a presentation for the upcoming SpringSim 2016 meeting, I wanted to express Deutsch’s concept of a good explanation being “hard to vary”. Here’s a quote from The Beginning of Infinity:

Fallibilism entails not looking to authorities but instead acknowledging that we may always be mistaken, and trying to correct errors. We do so by seeking good explanations – explanations that are hard to vary in the sense that changing the details would ruin the explanation.

The point I’m trying to make in the presentation is about building simulations that are easy to change. But by “easy to change”, what I really mean is “easy to falsify”. This point isn’t very easy to express to a group of (mostly) engineers, whose expertise and purpose is to design very particular solutions to very particular problems. This sort of stereotype of “the engineer” should be fairly clear to the layperson. Engineers are often quick to drill down into the detailed nitty gritty of any situation. To the synoptician, whose skill is seeing the horizon, the view from 30,000 feet, such a skill is best characterized as “falling down” into an abyss, rather than “drilling down” into detail.

The same can be said of a stereotypical scientist, whose motivation is to discover knowledge regardless of any particular gravity well of detail. E.g. a biologist who discovers she needs to learn, say, fluid dynamics, in order to make progress on the biology, does just that … learns fluid dynamics. Walls, boundaries, disciplines, are all battered down in relentless pursuit of the objective.

One way to think of the difference might be that the engineer accumulates globs of detail and packs them carefully together to build something. By contrast, the scientist encapsulates globs of detail, sloughs off what she can and drags what she must as she wades through the surrounding muck.

The point in my presentation is that cumulative globs of detail are a hindrance to science. And, although it might be counter-intuitive, this gels nicely with Deutsch’s “hard to vary” concept, because what I think he really means is: contains only the critical detail. It’s a kind of duality, the less detail your accurate theory has, the more difficult it is to change/delete any given detail without breaking the theory. That lack of detail makes them both easy and hard to vary. And this is the appropriate paradigm for scientific M&S.

By the way, I’m flirting with learning about the new-ish push toward the Philosophy of Engineering (e.g. here). This guy even seems to claim that Engineering (capital E) is larger, contains, science. At least that’s what I think I heard. But the above paradigm difference seems to argue against that.

There is a middle ground, I suppose, where both Engineering and Science require some form of intervention, activism. The idea is that you cannot understand independent of manipulation. But to me, this would imply that both Engineering and Science are (different) specialized descendants of the more basic condition of consciousness.