Is chasing a feeling a legitimate research method? I think a lot of people in the coding tools community, myself included, are chasing a feeling we want out of our tools. I want feelings of tangibility.

new paper from our work at Meta! **GPT-style language models memorize 3.6 bits per param** we compute capacity by measuring total bits memorized, using some theory from Shannon (1953) shockingly, the memorization-datasize curves look like this: ___________ / / (🧵)

Slick UI from my colleagues!

Still dismayed that Rima's amazing VideOrigami didn't win best paper. It deserves it. youtube.com/watch?v=U31u_u…

Still looking for professional devs that use AI agents like Codex or Claude Code. $100 Amazon gift card. DM me!

love this parody

very not-vibey

Vibe coding is the future 🤣

A theme of the recent literature on vibe coding is that programming experience is still required...and it doesn't always look or feel like vibing. I think this blog post sums it up well: you basically have to be a senior software engineer. simonwillison.net/2025/Oct/7/vib…

first off no, those are not "routers",

Do professionals vibe code? What is hype and what is real with software development with agents? We observed and surveyed professional developers (unlike most prior works) to find out. Answer: No, pros don't vibe, but they control. Preprint: people.cs.uchicago.edu/~brianhempel/P…

Professional Software Developers Don't Vibe, They Control: AI Agent Use for Coding in 2025 arxiv.org/abs/2512.14012

accidentally called the AI coding agent "he" today instead of "it" need to do penance now

xkcd 303 "Compiling" is now superseded by "Waiting for AI". But the phrase "waiting for AI" is not smooth. What's a better wording? Possibilities: - "My code's generating" - "Vibe waiting" - "AI's slow!" Other ideas?

and so our paper is already out of date 😆