November 18, 2023

The International Staff Band of the Salvation Army in 1941. (I wonder about that guy's trombone with the extra handle...not sure if it's a Sacbut or just a trombone with a handle) I think about professionalism in The Salvation Army brass bands. In most all music there's a spectrum from raw to polished. In the 1800s Salvation Army bands were known for their ad hoc, street-preaching, draw-a-crowd-with-whatever-horns-and-players you got. Obviously by 1941 there were some different aspirations at least at higher levels. I see the same thing in Honk!-style bands of course. Some bands are deep dives into esoteric time signatures and can build complex chords and harmonies. My bands tend to be a bit more built for robustness. Similar splits happen in a lot of arts, like, typography say. I was thinking about this thing on AI generated fonts:
LLMs work differently, of course, but since the local conditions of a typeface cannot be anticipated, and these deltas cannot be reliably inferred after the fact, the size and quality of the corpus is irrelevant to this effort. A typeface is less like writing an essay, and more like inventing a new language. The things that make it successful just can't be quantitatively measured.

That said... I think it absolutely certain that an unsupervised AI will create what certain segments of the lay public consider to be a good typeface, and if this doesn't happen in the next six months, I'll be very surprised. This is because non-specialists aren't burdened by the need to look at type critically, in terms of its mechanical requirements, its utility, its cultural relevance, its originality, or its style.
I'm really torn. On the one hand I know how the ultimate stupid guy thing to do is think that the smart guys don't exist. On the other hand, I am kind of suspicious about gourmet connoisseurship - the kind that seems to be operating in a different space than mere amateurs can detect (though there's always the claim that everyone subconsciously recognizes the truly great.)

These ideas get into some really deep issues. I may have lost my erstwhile debate partner Peterman with his frustration in my refusal to accept his vision of expertise and special knowledge.

Like one of the most consistent themes in epistemology for me is the capacity for self-deception - "Never believe a thing simply because you want it to be true" to quote Diax's Rake. And if you're on the outside looking in at audiophiles, say, paying thousands for special cables and whatever, it's really hard to know. (Though I have noticed that the headphones in the store always sound better - but much of that is because you're actually LISTENING for the nuance.)

But also I think the value of something is an emergent property of how it interacts with its audience. So maybe experts understand the subject so well they predict the emergence. And sometimes there are some things of interest but mostly to experts.

I don't know if I'm a real gourmet about anything in my life.
Maybe you need a frog?
AHAHA fuly fledged David-Attenborough-Narrator-Bots!!!!