Nicholas Nolan

[Nicholas Nolan @NicholasNolan64]

@TheStalwart @dioscuri Simulation hypothesis. If running high-fidelity simulated worlds (such that the inhabitants are sentient) becomes common post-singularity then even if only a fraction of those worlds are simulacra of the historical Earth they outnumber the real Earth and we're likely in one.

[dax @thdxr]

we really had it figured out with art deco

[Nicholas Nolan @NicholasNolan64]

@JeremyNguyenPhD @phokarlsson Was it this one?
https://t.co/t1quGLyhyN

[Nicholas Nolan @NicholasNolan64]

@KelseyTuoc One of these things is not like the others: administrators of food stamp and affirmative action programs may be indifferent to applicants lying, but owner occupancy fraud is a principle-agent problem and if lots of people are doing it then lots of people are in the wrong.

[Nicholas Nolan @NicholasNolan64]

@KelseyTuoc RF engineering relies on being able to accurately predict how waves will behave. With five digits of pi, we wouldn't be able to reliably produce antennae, resonators, filters, and so on. Simulations would be catastrophically wrong; stuff would be within tolerance only by chance.

[Nicholas Nolan @NicholasNolan64]

@ciphergoth Org members assigned specific permissions: admin, whitelist, blacklist, contextualize certain actions. The real solution is fixing the government's system; it should not be physically possible to add randos to the natsec group chat.

[Nicholas Nolan @NicholasNolan64]

@AmandaAskell Rewinding the clock 70 years is not a feasible plan but that doesn't mean the alternatives are workable.

[Nicholas Nolan @NicholasNolan64]

@AmandaAskell Artificial womb research faces catastrophic financial and regulatory barriers to reaching the prototype stage, we already have extended fertility spans via egg freezing, AI-assisted childcare sounds like marginal productivity gain for daycares, UBI is expensive, etc.

[Nicholas Nolan @NicholasNolan64]

@carlo_l_fritz @alz_zyd_ @waltuuuhr Seems like it's from https://t.co/1YcSQH0XsZ

[Nicholas Nolan @NicholasNolan64]

@ESYudkowsky @erikbryn @cosmosPixelado @gcolbourn @MatthewW_eacc Crystalline styles, a bivalve digestive organ.

[Nicholas Nolan @NicholasNolan64]

@historyinmemes Alcohol, although I can see the argument for opium.

[sydney @demiurgently]

- brain desires new good
- ask brain if this is desire for positional good or functional good
- brain doesn't understand, pull out illustrated diagram explaining relative income effect & hedonic treadmills
- brain laughs "its a real desire ma'am"
- acquires good
- its positional

[Grant♟️ @granawkins]

STANFORD TORUS MEGATHREAD
You’ve seen the pictures, but you probably don’t know the full story.

It's the result of a 10-week engineering study at Stanford University in 1975, funded by NASA and led by Girard O’Neill - the most thorough study on free-space settlement to date.
1/

[Nicholas Nolan @NicholasNolan64]

@parafactual I am not positive that slow takeoff is likelier than fast, but I do think that any task which cannot be evaluated on the order of milliseconds is unlikely to be RL'd into superhuman territory.

[Nicholas Nolan @NicholasNolan64]

@parafactual Current AIs that have achieved something like fast takeoff did so using reinforcement learning (e.g. AlphaZero became superhuman at chess in nine hours by playing 44 mm games). Doesn't seem to generalize well to tasks which are not purely informational, c.f. self-driving cars.

[Nicholas Nolan @NicholasNolan64]

@micsolana International acquisition of community banks in the United States. Biotech startup explosion. NFTs.

[QC @QiaochuYuan]

it's a little-known fact that god actually didn't die until the widespread use of pseudorandom number generators by modern computers, each use a small act of divination, exhausted god's ability to respond to all divination. god was killed mostly by microsoft in the 90s

[Nicholas Nolan @NicholasNolan64]

@alyssamvance Availability heuristic?

[Nicholas Nolan @NicholasNolan64]

Xenosemanticity has good mouthfeel but is sadly less applicable to us. Even if mind-readers were real you wouldn't be able to defeat them by kludging disparate concepts together to make verboten thoughts – when the rest of your thoughts are plaintext, cyphertext stands out.

[Nicholas Nolan @NicholasNolan64]

Interesting term from LW post on interpretability. Autostigmergy: encoding information into the environment so it doesn't need to be in your own memory. Humans do that all the time, albeit to augment our memory and not defeat mind-readers.
https://t.co/4ArygVdsNS