think of the Mixture of (locally hosted) Experts we'll have as well. right now the trend is near-instaneous single-player 0-shot, what happens when we revert to async, git-style, & multiplayer-edited output for common ideas?
whoops, set the historical cost too early.
watching Oppenheimer just inspires me to think of all the loci in Farcaster where individual projects are working out the unknown potential of social media. there's probably going to be a Manhattan Project where all the angles converge.
sanity check: it occurred to me that insurance is an unorthodox debt, the more insured everything is, the cheaper the premium for noncatastrophic outcomes. feels like prediction markets really should be more of a secondary monetary policy that redeems against distressed vs catastrophic outcomes.
the weird flip I only noticed on rewatches was Cooper's insistence on taking his daughter to the game, then initially refusing to take her to the coordinates. opposite of the don't leave tesseract scene, wanting to initially meet her again on the station, only to leave her behind. attachment/detachment is a huge theme.
100%, I think medical tech should solve medical problems, mostly w/ early detection & correction, but anyone living >110 years is going to have a lot of challenges that end being a ship of theseus solution.
either 2nd Renaissance or Beyond. Renaissance's style was a meme for a bit in pop culture.
not in polite company.
is it a nerdsnipe? it's probably a great idea for research vs a great idea for a market.
I think I saw a pretty interesting conversation on mastadon: https://mastodon.social/@colacharm/101180459575107194
https://warpcast.com/nor/0x5ad69a suppose planned bots are the double-edged sword that makes farcaster-market fit. how would registering/calling those get enshrined into the network? e.g. a parse/order autonomous agent limited to a dedicated permissionless channel.
energy breakthrough + water purification + cross-continental network of pumped storage would probably eradicate famine, arid environments, and climate control for starters.
not a small improvement, an excellent improvement. search bar and contained "manage following" are going to definitely alleviate a permissionless channel overload in the future.
it should have far better transcription (who's training a reward model for Whisper 2?)
no, he tends to be organized and less figurative than this was (I read the whole thing).
but the issue is straightforward to address in a system that is frugal and public. if the trust assumption of reporting misalignment is 1/(# of psychopaths in the world + 1) & in the future we'll err on sharing & editing prompts before agentizing those w/ the weakest possible models that do the job, then that's good.
I would say it's partly informed by the research coming out. technique has to work its way through scarce inputs (e.g. API cost in FrugalGPT) + the theory that symbolic regression/weaker models (LATM paper) is more ideal after using the most powerful intelligence exclusively for discovery. alignment is an issue,...
admittedly, this conjecture should be debated and explored to a much greater degree, I'm not sold with any of the AI alignment discourse tbqh.
will give this a listen, but intuitively I do have agree that humans being outcompeted in a microcosm does not generalize to a world that run by AI singularly designed to outcompete humans in a qualifiable way.
downscaling/multiplying the Parker Solar Probe (https://en.wikipedia.org/wiki/Parker_Solar_Probe) & launching as many of them periodically, along with an Earth-Venus cycler (https://ui.adsabs.harvard.edu/abs/1969JSpRo...6..366H/abstract) (bonus points for doing it with NEOs) just to claim the first Dyson Swarm.
conversely, the first decentralized social network that can dogfood up to Github's functionality is going to be pretty impactful, imho. what are your thoughts about Radicle?
🌶️- nobody has the audacity to turn gambling & metagambling into a full-fledged monetary policy. numer.ai might be the most adjacent so far. but w/ e.g. polymarket, you have to bring USDC. there's no free guesses w/ "good guess" emission. there's no profile page w/ epochal PNL betting. no bond mechanism either.
as long as crypto adapts by adding those rails as an option. in terms of centralization & censorship, it definitely has tradeoffs as a panacea. I've heard that it's been due relative to the rails other cpuntries have.