Quote:

Success in an air battle for a fighter plane depends on the pilot, the engine, the wings and the guns.

What percentage importance does each have?

Quote:

...technically, the clarinet is not a "lumped system" in control engineering terms. Regression analysis is of the most use when variables, though they may be many, each have a distinct influence and the cross-coupling terms are weak. [The] cholesterol example is a good one.

The analysis may, of course, show up cross-couplings, eg "high cholesterol is not so harmful if your diet is rich in Armenian blue parsnips", but if these exist, then one absolutely cannot have a description that is appropriate to a lumped system (decreasing in importance as you go down etc).

[In the clarinet,] the cross couplings are huge, as simple experiments show: leaks, cracks, resonance fingerings, fork fingerings, sloppy fingerings, bits of lead stuck on the outside(see Benade) etc etc. Thus the sound column does not already exist at the top of the instrument, but is the product of everything. The whole vibrating air column depends on a lot of things and they are closely coupled. I think this is good science, not sophistry. And this is not to deny that _sometimes_ the assertion works, for example a leak on the bottom pad will only affect the lowest notes in each register.