Profile avatar
niplav.site
https://niplav.site/index
1,025 posts 339 followers 218 following
Regular Contributor
Active Commenter

expert networks : apocalypse panics :: mass media : moral panics

People don't argue enough about whether AI can meaningfully help solve the alignment of ASIs They argue too much if the alignment of ASIs is a relevant problem I'd like about 1½ OOMs of debate *more* on the former question and about ⅔ OOMs *less* on the latter Thank you

My favorite (sadly defunct) schizo website ever: web.archive.org/web/*/http:/...

Takeaway from LessOnline: The people who are very active on twitter were on average more interesting to talk to than the non-poasters.

I think parts of the Frame Problem¹ can be used with Trammel-style fixed points in infinite regresses send bisk ¹: en.wikipedia.org/wiki/Frame_p...

"full-throated support for trans folk" can be either be a trivial consequence of advocating for morphological freedom, or a demand for endorsement of a specific gender ontology. Those two are often conflated.

More on how I ~don't have internet at home: niplav.site/notes#On_Hav...

Andrés Gomez Emilsson's EM field theory view predicts that strong [[magnets]] should have "psychedelic" effects?

Word for a good thing done out of ignorance (opposite, in some sense, of "mistake")?

Are the Szegedys an eminent family? Mario: Mathematics Christian: Machine learning Balázs: Mathematics

If Δ is for updating towards an interlocutor's position, then ∇ is for updating on multiple positions at the same time? Deep update/viewquake?

Friendship is owoptimal

cowonstrained owoptimization

Sleep paralysis=co-(lucid dreaming)