niplav @niplav@schelling.pt

Evaluation being easier than generation doesn't mean evaluation is easy

Has anyone tried to ask organisational psychologists about scalable oversight?

s h o r t t o r s o

@cosmiccitizen nice, thanks for the rec!

@scathach
"“Exactly, Houston. Everything is okay. Nothing is wrong. Nothing has ever been wrong, anywhere. The cosmos is like a flawless jewel, each of whose facets is another flawless jewel, and so on to infinity. Except there is no jewel. It’s all light. No, there isn’t even light. From within Time you can’t see any of it, but when you step outside into Eternity it’s all so…full. It’s so beautiful, Houston.”"

"HOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

@flancian @agora done, thanks

@liaizon *spins a last time*

Chaitinudassicon

- How did the polynesians settle all those islands in the Pacific? [[History Questions]]
- This seems just impossible to do, to me. Or *a lot* of people die in attempts

- Since cancers evolve very quickly and diversely, can we perhaps ultimately use *that fact* as a method of detecting cancer, as a sort of OOD-situation for cells in the body?
- https://www.quantamagazine.org/studies-reveal-extreme-diversity-of-cancer-cells-20131113/
- In general cancers need to evolve quicker than the systems they're embedded in
- Applications to [[Demons in the Prior]]?

Gwern must feel so lonely up there

s h o r t t o r s o

I have located your program in the universal distribution, pray I don't locate it any further

I would simply choose a Turing machine in which I am the simplest program

The problem with the Repugnant Conclusion in UDASSA is that you run out of low-complexity creatures to occupy the short programs with

Reading “Anthropics and the Universal Distribution” on public transport and shaking my head so everybody sees I disagree with it