News for hackers. Stuff that matters.

clzerv

Re:GPT‑6.2 Astra and Sol on Tuesday November 17, 2026 @03:05PM (#50537361)
The Fable5.2 version is the only one that allows user interaction with the model, it's also very clean and minimal. I've been using Sol-5.1-high for a dumb web project and it tends to create really baroque code and UIs. I haven't used Claude in a while and this makes me want to try it again.

It seems like the models have a kind of built in aesthetic sense and I wonder what part of the training influences this. Maybe if more RL time was spent on artistic tasks the models could not only make better visuals but also code.

Currently they all seem to care little about code quality and maintainability (by humans), just functionality, which is fine I guess if you can expect to vibe code forever. But what if there is a critical bug and the models are temporarily unavailable for some reason? It would be hopeless to expect a human to wade through the giant bowl of clanker soup and fix it anytime soon.

While I appreciate the effort and talent to pull this off, I find the results to be awful. I call this kind of overblown sound processing "Bose barf" since pretty much all Bose audio sounds like this, and it just grates on my nerves. Now TVs, Bluetooth speakers, headphones, etc. all seem to perform this kind of aggressive audio processing to overcome the limits of super cheap tiny drivers. It's getting hard to find devices which have a more neutral sound profile. All this audio processing makes music sound like it's coming out of a garden hose duct taped to a 1950's grade school hall speaker. I can't understand the appeal.