sfkgtbor

Happy about the Sonnet cache read price cut.

I'm glad both labs noticed and are trying to improve the models communication styles, they were getting closer and closer to meaningless gibberish.

Very interesting to see another person also landing on using 4 base32 characters for labeling things - it really is enough for a human.

Personally I use it for labeling physical things - mainly boxes. With a corresponding note in my Obsidian vault it really helps with getting content, context, and history about random stuff in my basement.

Python oneliner for generating them I've aliased in my Bash config: python3 -c "import base64; import secrets; print(''.join(secrets.choice(base64._b32alphabet.decode()) for _ in range(4)))"

Multiple people found prompts to make LLM solve the problem, and the $10k has been awarded: https://twitter.com/VictorTaelin/status/1777049193489572064

I really like seeing references to the Culture series when naming things:

https://en.m.wikipedia.org/wiki/The_Player_of_Games

I think the blog post about load shedding by Netflix has some good examples.

"Keeping Netflix Reliable Using Prioritized Load Shedding" https://netflixtechblog.com/keeping-netflix-reliable-using-p...

In terms of safety, vaccines are much better then getting COVID.

One of the big factors of "better protection" of immunity gained after recovering from the virus is survivor bias. In data about mortality after reinfection there are no people who died after getting it for the 1st time. This leaves only ones who had body strong enough to recover.

There's no surprise that later that group fares better.