the software gas law is showing itself very clearly on nostr “The resource usage of any piece of software will expand until there are no more resources to use”

gm 🍍

we *could* have better REQs

the pain will continue until people realize how to properly choose models and don’t cargo cult <latest model> from Anthropic/Z.ai/OpenAI/etc View quoted note →

gm mmmmh 🍍

LLMs are decent if what you are trying to do is in the training distribution. They suck otherwise. 80% of stuff on Nostr here is. (Not sure if that is good for innovation) and even then you can see the tons of rough edges. For the rest, 20%, like scalable infrastructure, good UX, good relays. AI cannot fix that, only competent developers can. And Nostr is having a hard time retaining competent developers I think. With the amount of hype-driven development, “specs” changing constantly, and low quality code, only few competent developers want to develop here.

apparently MARA is now an AI inference company instead of Bitcoin mining probably a net positive

If you have more than 2 LLMs working on the *same task* in parallel then you’re doing it wrong

why is OpenAI resetting Codex ratelimits every 5 minutes?!

gm 🍍

apparently pi supports subagents calling subagents what the fuck

gm 🍍