Hao's Preprint Reading List August 1 2025 CDK9 Erythroid enucleation Evolutionarily conserved atherosclerotic lesional mesenchymal cells Fibroblast-Oligodendrocyte-demyelination Long-distance mechanical signals of Caveolae Sulfatase 🫀development VEGFR2 Aerocyte docs.google.com/document/d/e...

Show Post

"i hate curl | sh, it's so insecure" listen rust installed in 30 seconds but getting a newer than 2022 clang installed on a debian machine has already taken like half an hour of my life

Show Post

Does it matter as long as they’ve arrived at the correct answer regardless of how? It demonstrates their problem solving skills which are highly valuable in employees.

Show Post

if a candidate answers all your questions by looking it up on an LLM, but you can’t figure it out just from talking to them do you: reject: they’re not independently capable insta-hire: they have perfect AI fluency

Show Post

this is starting to be a consensus. same report across lots of people basing their opinions on anonymous models on LM Arena 1. OpenAI open weights is very good 2. GPT-5 is extremely good

Show Post