An interchange about optimization

Ooooooh, Bob, I think you’ve oversimplified things just a tad [in “Optimizing the unoptimizable,” Keep the Joint Running, 4/24/2006].

Optimizing “must be about a single variable”? Nah, I don’t think so. In general it’s about a function of several — or even many variables. (Or to put it another way, life is about tradeoffs. You really can’t ever, in the real world, optimize one variable and the hell with all the rest.)

Your own sports-car analogy illustrates the point. “Handle more responsively”? Now *there’s* a fuzzy kind of object function. Getting the car to stick on the road while you fling it around corners is bound to add weight, which compromises its acceleration. Adding brakes also adds weight, which degrades acceleration and cornering ability to boot.

Etc.

Of course I could stay up all night optimizing the clarity and grammatical correctness of this letter. But I think in the interests of making a sensible tradeoff between time to get to bed and maximizing the clarity of what I’m trying to say, I’m ready to call it done.

– Optimizer

Dear Optimist …

Good e-mail optimization!

I think you and I are at cross purposes. I’d say you can only optimize a single variable, but to do so you must juggle (optimize the mix of) a collection of other variables. Mathematically, it’s a matter of solving for the targeted value of an equation’s independent variable.

Not that an organization can be reduced to an equation. We could only hope for that kind of engineering precision in organizational dynamics.

Life is about trade-offs. Optimization is how you go about deciding what the logical trade-offs are. You point out the key challenge – in most cases, the independent variable is squishy: Hard to define; impossible to reduce to a measure.

We should all be happy that this is the case, I think. If it weren’t, there would be one perfect sports car, instead of having the fun of choosing from different opinions about what one should look like and how it should drive.

– Bob