In reply to:

Sushi: how do you measure the eneven performance of a tweeter when given a large burst of power, as opposed to a steady tone? The driver motion is so miniscule that I absolutely think there's room for things we don't yet know how to measure.

A tweeter or woofer doesn't matter. The driver's swing, how small it may be, doesn't matter either -- it is easy to measure them mechanically, either indirectly or directly by today's techniques. If the use-dependent permanent change in the driver indeed exists, it should be measurable mechanically at the driver level -- again, these are simple things including the resonance frequencies, mechanical and electrical Q's, and the mechanical compliance. The difficulty arises when trying to "completely" measure all the potentially audible changes by ACOUSTIC measurements. But, as I said, in the particular context of speaker break-in, the permanent physical change, if it actually exists, should be detectable by MECHANICAL measurements of individual drivers.

Incidentally, I should also point out here that today's "routine" acoustic measurement techniques (such as MLSSA), do not even use steady-state test tones. The results are all time-resolved, and in many cases, spatially-resolved. In fact, my understanding is that today's bottleneck is the methods for interpreting the wealth of time- and spatial-domain data to accurately correlate them with listening tests, rather than the availability of measurement data themselves. The good & ol' days of the steady-state frequency responses and distortion measurements were long gone. I still do not think they can measure and interpret EVERY audible differences. But let us not underestimate the power of today's acoustic measurement techniques.