I'd want to say that those tiny moments of musicality shouldn't be notated, even if they could be. — J
Agreed. I suppose I shouldn't have put it this way because I was thinking of the musicologist not the musician, someone who is analysing a performance rather than creating one.
I'm of two minds in this talk of "having enough data" I keep using, here in talking about music or above talking about the social sciences.
There's a great forgotten book called The Road to Xanadu by John Livingston Lowes (iirc) in which he traces every image, very nearly every phrase and every word, in two poems of Coleridge (Xanadu and Ancient Mariner) to sources in Coleridge's library. It's not an "explanation" of the poems; I believe the point Lowes made (and I may misremember) was that in a way knowing all this only deepens the mystery of Coleridge's creativity in taking all this material to create these things. It's not like you could train an LLM on Coleridge's library and then say, "Write me an astonishing poem," and out they would pop.
(Coleridge being a particularly ripe case, as Eliot described him, a man visited by the muse for a while, and when she left, he was a haunted man. Coleridge himself didn't understand what had happened.)
So, part of me does want to say that there can never be enough data to explain, much less predict, human action, and certainly not unlikely human action like creativity. The "human sciences" would then be marked either by arrogance or folly, as you like. I could be old school, I'm old enough.
But I'm not convinced. That attitude strikes me unavoidably as a rearguard action, defending human nobility against a godless and disenchanting science, that sort of thing.
Instead, I think it's simply a fact that the data needed, and the theory needed, are evidently beyond us, and so we must make do and aim a bit lower in our expectations, or at least be more circumspect in our claims. When it comes to scientifically informed debates over social policy, for instance, we sometimes know enough to do better, but still less than we think we do and so some caution is advisable.
We know a lot of what was swirling around in Coleridge's head, but not all of it, and we know something about how his brain worked, because it worked like ours, but the specific historical process that took those inputs and yielded those outputs is unrecoverable.
So it is with any musical performance. I'm inclined to say that one of the reasons the musician played this note this way is because of that time she wiped out on her bike when she was 8. That might be a big enough factor to make it into her biography ― if, say, she broke a finger that healed in a way relevant to her playing. It might be a kind of emotional turning point for her, if it nudged her attitude toward risk a certain way. It might be an infinitesimal factor, no more or less relevant than the peanut-butter sandwich she ate that day, but all of which went into making her the person who produced that performance.
We're talking really about what God knows about her. When God hears that performance, does he smile slightly and connect it to that skid on her bike? God has all the data, so how does he understand the world and the people in it?
It looks like that's the standard for science I have been indirectly endorsing, or if not "standard" then "ideal". Which is a little odd, certainly, but maybe that's fine. In practice, science is entirely a matter of making do, and being very clever about what you can learn and how despite not being gods. I guess.
Therefore Bayes.