I wonder how well adaptive timbre could work? Could we have timbres that coincide perfectly for fifths, thirds, sevenths(ish) without having to do adaptive tuning? Just, you know, FFTing the input signal, picking out the odd-numbered overtones, raising the third harmonic by 2c, the fifth by 14, the seventh by something like 30c, etc. (With a synthesized signal it would be easier, of course, since no adjustment of input after FFTing it is needed)

Would different tones of a chord need them differently adjusted? Is there a way to uniformly adjust harmonics for all tones so that 12tet basically would have the benefits of JI when it comes to harmonics coinciding? Or is this a pipe dream? Would the timbres sound way too weird?

Could there be a non-equal adjustment that wouldn't be adaptive, such that e.g. Dmaj and Gmaj, despite having the identical 1:2^(4/12):2^(7/12) frequency patterns have different enough out-of-tune harmonics to, despite both sounding good and major still sounding interestingly distinct, a bit like a timbral equivalent to the non-equal temperaments that were in vogue before 12tet?

A reason I don't much like the kind of linguistic prescriptivism based in armchair-linguistic musings of people into a sort of logician/engineering/whatever mindset is basically to do with how language and our minds and our social situations have evolved.

Basically, what happens when you let evolution run on some system that can be designed evolutionarily, it will often take into account features of the underlying systems that the human designer is not aware of.

Secondarily, language as a system is sort of distributed. No one knows every word of their language (with some trivial exceptions, but these are 'degenerate' cases, especially languages on the brink of extinction), no one knows every construction, etc. Constructions spread when people pick up that they are constructions, which means that for a construction to gain ground, there must be something that gives away that it is a construction in the first place, it needs to fill a purpose, it needs to be easy enough to apply by analogy, etc. So, we have a system where there's a lot of evolutionary algorithms going on all over the place, and the best thing is - the protocol by which these innovations spreads is itself affected by these changes. We're having evolutionary algorithms within evolutionary algorithms. This, in turn, in a really distributed system, which also has interfaces with other similar systems (== other languages, by means of bilingual speakers or anyone in a language contact situation!) ...

The number of fields one should understand to grasp why languages are like they are probably exceeds what anyone can learn in a lifetime. Probably, any number of the weirder things in language - quirky case, irregular forms, discongruence (e.g. the Semitic system where numbers over four or somesuch are marked for the opposite gender of the nouns they govern), congruence, differential object marking, aspect marked on the object instead of on the verb (making aspect unmarkable on intransitives and quirky case verbs) are results of optimizations that have made sense for the distributed memetic(ish) algorithm that designs language all the time. The reanalyzis of how to use the non-nominative pronouns in English probably also fit in with this - in lack of other supporting structures, lots of minds have converged on using them in a way that marks some kind of distinction that is more relevant for them to mark than whatever was marked earlier, hence expressions like "my brother and me went to town yesterday".

For a human to optimize this is doomed to failure. He's unlikely to have an idea of the uses of most of the stuff in the language, and will cut out stuff that it is based on, and add in parts that seem logical or easily processable, but haven't gone through centuries of optimization.

I'm also considering calling 'evolution' an intelligent process, since it solves problems.

I guess this kind of subverts the Intelligent Design people's position. At least I hope so.