It did actually answer my question though 😅

Haha it’s certainly HPC the “hard” way….

We’ve gotta do that ourselves!

Hence why the big platform firms push these things so strongly… just like PyTorch/Pyro/JAX+Tensorflow jockeying, these tools don’t *remove* friction per se, they displace it in the stack. I love llms for learners, but I’m not sold that they end this “front” of the “language war…”

I wonder whether this will make the *library side* of these debates even more winner-take-all? Lower barriers for users make it more important for language library developers to “keep parity” with one another, and even possibly slightly ahead. It also intensifies the value of network/ecosystem!

For users, this is probably true! Who can be territorial when the most basic barriers to usage (like installation and familiarity) get ground down by Copilot-type things…but!

if you're curious: doi.org/10.1177/0309...

I never looked into it, but it’s a solid idea—Ron used to have this “classics that weren’t” project w/ books/papers he thought were primed to be impactful but, for whatever reason, never were. That catmog and the associated paper about London housing is *very* ahead of its time.

Some, like #27, are startlingly contemporary! Also, @fcorowe.bsky.social is editing a revival w/ the Digital CATMOGs at ERSA’s REGION journal! AFAIK, Serge’s got the first! doi.org/10.18335/reg...