@pathunstrom @xgranade I'm curious about something.

I'm not a developer, but I have a lot of experience writing scripts, and I can 'think like a computer' and understand where things are happening and what's going wrong. I'm the person at work who gets the head-scratcher tickets and has to tactfully walk people back from 'bad ideas/bad design/tech debt'. I'm pretty sure that paired-up with something like Claude Code, I could work out a decent set of PRs to do something like "convert Basilisk II from GTK3 to GTK4, or SDL2 to SDL3".

A lot of people I interact with (not at work, thankfully) are deeply drunk on LLM Kool-Aide. They are weirdly excited about using LLMs and agents for everything they can. They tend to be the kinds of people that take joy in watching their toys 'do stuff' they never could with natural language. They do not have deep technical backgrounds, most have never written a script or taken the time to profile why something is slow or inefficient. I'm guessing that many people like this are just pointing their tools at projects and hoping to score points on GitHub like they would go for 'likes' on Instagram.

Is there a difference to the project owners between us, or are we both slop factories? I ask because I get the impression there are a LOT of the latter, and few of the former, and I'm guessing that one is giving a bad name to the other.