[Jtsummers]

I keep going back to this quote from The Psychology of Computer Programming by Gerald Weinberg:

> Fisher's Fundamental Theorem states—in terms appropriate to the present context—that the better adapted a system is to a particular environment, the less adaptable it is to new environments.

It's a systems thing. You build a system that is resilient to changes in the environment, it is less efficient if the environment doesn't change versus a system that is optimized for that specific environment. But the counterpoint is that the optimized system can't handle changes in the environment.

[andrewflnr]

Nice. I've always thought of it as shaving away error margins, but I think your quote is much more general and easy to understand.

[Jtsummers]

Well, shaving away the error margins is one way you gain efficiency. But in doing so the assumption is made that the environment is static (or sufficiently static relative to that margin). How does the fastest ground vehicle hit that mark? By being the lightest, most streamlined it can be with the most powerful motor it can contain, running on one of the flattest longest spaces it can find. Put it on a NASCAR track and the operator would be dead.

An important note is that assumptions are often not made explicitly. People don't sit down and say, "I think this vehicle will never be used on an oval race track, therefore 'steering' will be minimal, and the throttle will be full-open or full-closed". They just know they're making the fastest ground vehicle, and so those decisions are consequences of that specific objective. They happen to have an implied assumption that the vehicle will never need to operate at lower speeds or make turns.

This leads to an interesting tangent, System-Theoretic Process Analysis (STPA) by Nancy Leveson has an objective of improving system safety. But one of the ways it does this is to raise those assumptions to the level of conscious awareness by creating an explicit system model. It's not the only process that does this, but it's making headway in the systems safety community these days.