@tszzl — page 43/103

@SkepticalAlpaca @halvorz yeah I’m jk virology is fuggin epic

@SkepticalAlpaca @halvorz any interesting scientific discipline has a small chance of destroying humanity

the subtext is that people with scores so high that they hit the ceiling effect on standardized tests are a literal commodity https://x.com/wwwojtekk/status/1411534579899551744

@chaosprime 🥲

like lambs to the cosmic slaughter!

@bowserbot2 i guess the dispassionate way they talk about the candidate suggests they probably get quite a few that are similarly impressive

@ryan_landay “Ceiling effect” not “ceiling” as in the predictive power after a certain score may be nil https://en.wikipedia.org/wiki/Ceiling_effect_(statistics)

when someone is going through your thread and liking tweets you can infer their verbal skill based on the speed of traversal :^)

just one more thing for you to worry about

@ipsumkyle meaning there’s not that many or?

@halvorz fug

@DistractedAnna lmao

@VivaLaPanda_ owned

@AOCummies based

@halvorz as far as I can tell everybody who posted on Saturday had perfect verbal reasoning which means the test is sus

@halvorz understanding my verbal vomit is the real challenge

@halvorz yea unfortunately vocab is a pretty g loaded indicator, which sucks for foreigners

@SkepticalAlpaca @95thoughts @halvorz @petergodofsky gdi

@halvorz anecdotally I found the verbal pretty easy and the math somewhat difficult to finish (took it December ‘20) didn’t practice at all so maybe verbal skills are stickier

@LudditeHacker Nice

@foxstoryDA @halvorz yea agreed kinda bullshit to make you wast time on a test that doesn’t matter

@__femb0t @halvorz @sgodofsk

@95thoughts @halvorz @SkepticalAlpaca @petergodofsky is your boss Greg egan

@deknightberight 🤣

@__femb0t @halvorz @sgodofsk it’s genuinely difficult to understand what you people are saying half the time; requires massive inferential jumps and often quite good theory of mind

@__femb0t @halvorz @sgodofsk obviously not as good as reading a damn book but better than a spoken convo

@Treelochana my happiness is way greater on the green day

@RemnantPodFan @nominalthoughts keelhaul the lot!

@95thoughts @halvorz @SkepticalAlpaca @petergodofsky axiomatic

@95thoughts @halvorz @SkepticalAlpaca @petergodofsky specifically I was thinking of this one https://en.wikipedia.org/wiki/The_Moral_Virologist

@orthonormalist @ouranometrian congrats to both of you!!

@gbrl_dick https://zerohplovecraft.wordpress.com/2018/05/11/the-gig-economy-2/

@Duderichy down

@alth0u lmaooo

@alth0u the dream: man will merge with machine to become as Gods the reality: man will merge with twitch terms of service

@EggrollShogun wow anon … your wpm is so high

the twitch terms of service are actually more powerful than many real legal codes

every revision and update creates and destroys entire worlds

@alth0u like when a new law is passed and there’s a lag before the courts work out the corner cases

RIP to the hot tub meta. fun while it lasted

@alth0u Federal Reserve forward guidance but for egirls

@IvyAstrix incredible isn’t it

regulation shapes entrepreneurship https://x.com/IvyAstrix/status/1412313156269928452

@darpyu hot damn!

@aniiyengar @micsolana

@Virtual1nstinct https://x.com/tszzl/status/1396281641698537477

@sameQCU true

inshallah we’ve hit longevity escape velocity https://x.com/Y2K_mindset/status/1412131747475509250

@bowserbot2 @thetinduck something domain specific or? nips is the gold standard

@powerbottomdad1 that’s nothing compared too $50/month in passive income

@HargroveAlex @powerbottomdad1 ur NGMI Bro get on the grindset

seems straightforwardly meaningful to confirm your relationship before family community & country, religious or no https://x.com/owenbroadcast/status/1412398105060904960

@weylandswart @powerbottomdad1 imagine being salaried when you could be shrimp farming

@lisatomic5 i think “meaning crisis” is overblown to some degree

@owenbroadcast idk do you sign your home loan in front of everyone you love

@notlinear1 at the risk of my followers calling me a cringe autist, it’s the evolved equilibrium yes. every single successful civilization discovered the ceremony of marriage, it’s clearly effective and Lindy

@arctanno @weylandswart @powerbottomdad1 fucj

when you’re ready to make real money. talk to me about shape farming

we are raising new geometries from hell and getting paid top dollar. sign up for my udemy course

are any of you working at an ML startup right now that is a frontend for a Philippines labeling sweatshop? dm asap I have questions

@dasanil @bowserbot2 Yeah that’s the move

@owenbroadcast @LudditeHacker i mean i firmly reject your dichotomy of mundane vs supraphysical but i don’t think it’s a productive disagreement so w.e.

@LudditeHacker @owenbroadcast no problem at all i do these threads so people will argue and I can observe

@owenbroadcast @LudditeHacker I saw a solar eclipse before; it was entirely natural and not supernatural as I’m sure you (might) agree. still awe inspiring and brought tears to my eyes. can you deny its emotional resonance despite irreligion?

@owenbroadcast @LudditeHacker is the marriage any different? i can get all sappy about it’s virtues without invoking god. i couldn’t get sappy about the contract that seals a home loan. the difference is obvious even if the people in your doc aren’t eloquent enough to find the words

@lalaAlicelala what do you mean by argument, and semantics. and twitter dot com

@divalbanerjee you are worth 50 of them king🤴

FELLAS: How did you get an L this week? https://t.co/gj5arYkX5x

@retttx lol just use the last 7 days man I’m doing a bit

@normalhandsome 🙏😕

@Historycourses rip and lol I don’t know if I wanna be the guy bringing negativity week over week

@michaelcurzi 😂

@BroductManager paul allen invented microsoft bro

shroomjak is a demon spreading through the noosphere, true. but what they don’t tell you is that there are several morally righteous demons in Hinduism and I’ve just added mr shrigma to the canon

@aniiyengar I’ve finally gone too far

if you understand this tweet you’re fundamentally NGMI

@ITBeHa “not gonna make it”

@anothercohen problem?

@lucaruzzola @retttx hell yeah

@Thinkwert 10,000 mcchickens

@grantadever bro it’s getting impossible to read my notifs so I check only the mentions tab, which that dumfuk @jack in his infinite ignorance has decided does not include quote tweets

@erienneyoung @Thinkwert I will hand them out to the People

@grantadever @gerad_t0d @theashelina Grant. Thanks for letting me crash at your summer home dude. Didn’t know you had two extras

@halvorz @lalaAlicelala wait why

@UneasyCholo buddy it’s just started

“$x net worth is rich” “not in City Y” ok no enough. this discourse is over. you are all rich compared to my boy The Global Poor

@ne0liberal gas this asap

@Lev_Esch @halvorz @lalaAlicelala just searched “gre” in my gmail haha, these things are self documenting now

@postpostpostr only when it’s convenient

@drethelin yeah but guess what the Facebook Dating team is optimizing

@drethelin i mean idk but a good guess is that dating team is optimizing for dating

@SkepticalAlpaca Rip

@lowqualitybot they’re only real when it’s convenient for me, just like god

you’re all rich https://x.com/__hand_banana/status/1412569564588642307

@jeff82874662 it’s all about the malaria nets Jeff

@medved_bjj it’s provincial rich, but who cares

@postpostpostr @jeff82874662 isn’t this just one of my tweets

@darpyu yea neither of my parents had electricity or running water until sometime in their teens

interesting anti-bostrom argument https://maxhodak.com/downloads/simulation.pdf

to be honest im glad to just read one of these that doesnt hinge on "universal simulation is reddit"

@nEquals001 nah

@L0m3z yeah that's a good one

@LudditeHacker things can be both reddit and true

@xyz_nihilo all that hoopla to add one line to the editorial guidelines, which they clearly ignore anyway

@lalaAlicelala @LudditeHacker exactly

@alth0u mfw all of conscious experience is just my body gaslighting me

@alth0u only the p zombies are enlightened

@LudditeHacker https://x.com/tszzl/status/1402415147629481990?s=20

@alth0u Even without your participation, all the warriors gathered here will die. Therefore, arise, Arjuna; conquer your enemies and enjoy the glory of sovereignty. I have already slain all these warriors; you will only be my instrument. B.G. 11:34

@MegaBasedChad democracy is far less important than liberalism. few know this

@MegaBasedChad Singapore >>>>

@BarneyFlames it’s cope all the way down

@PradyuPrasad this tweet inspired by your bio

@PradyuPrasad trapped in innovators dilemma - double down on maintenance or try new things and risk it collapsing

@houellebecq_2 im guy

@alth0u probably not worth doing anything derisked to this degree

@alth0u in my mind once you assemble the architecture of some software project (with high belief that it’ll work) actually doing it becomes pointless

@jeff82874662 this is effective propaganda because it’ll ring true to most countries facing shortages while us vax rates already plateaud from vax hesitance

@ivashko @alth0u true, management risk becomes key. anyway probably good reason to start a company so you can skip several steps on management

@thesravaka and that’s fine

@brawnyman713 @jeff82874662 but realistically nobody expects the Chinese to come through as their savior — they do expect their benign economic-imperial rulers to save them though

by onion futures they meant this

@jeff82874662 because a pangolin fucked a bat 🥰

@tallsnail @Duderichy @NotComradeSnake @anothercohen I have a mf california king. not by choice mind you

@MatiosTV rip!

@generativist evolve intelligent life

@thickwithcolour lol

@gigafelon I wish I was at his level

@eigenrobot yea bearish on the ability of charter schools to create miracle growth

hope they get it down to 5 figure range https://t.co/TiJKh9O6um

@_StevenFan @generativist inshallah we will create and then enslave them to create market value

@fatcat12321 congrats dumbass you just ceded them most of Creation

which automations become feasible under $20/hr low wage worker regime that werent viable at $10/hr https://www.businessinsider.com/labor-shortage-automation-restaurants-hiring-employment-productivity-job-market-outlook-2021-7

I remember at a McDonald’s I went to in India they had a guy who would listen to your order and then manually enter it himself into the self-checkout kiosk lol. Many automations only make sense in high wage labor regimes

@notlinear1 yeah I don’t know what the social effects are but it seems like the usg has figured out a way to give UBIs without really calling it that

@btbytes wouldn’t the elites in India know English? Or why not include a language menu?

@notzoetaylor this feels like it was actually bought and paid for as a pharma psyop

@keeperofsheep1 p much

@AgainstGop yeah, skeptical of boundless optimism abt retraining everyone into python devs

@finnitejest you sure you’re not talking about me?

@carrr_losss you misunderstand the point if all labor is more expensive, then more things become feasible/necessary to automate

@generativist of course. we can't be abandoning our visual intuitions for the sake of algebraic notation sophistry

@lalaAlicelala moonschool doesnt sound fun but mb can send them for tourism there 😀

@lalaAlicelala yea me too but then you reread the book and its about how much the military ruined his life by sending him to school in orbit

@lalaAlicelala tru

@lalaAlicelala https://x.com/tszzl/status/1365862880927576066?s=20

tfw coworker catches you with leetcode open 😳

@alth0u bespoke: doing this on purpose as a form of career negging

@alth0u yea thats what i meant too

@alth0u low verbal convo rn

at the level of a handful of people organizing and managing a team effectively is mostly an empathy oriented task. at the large scale it becomes mechanistic again and a certain kind of meat computer autist tends to win. many struggling with these disparities of scale

@niccoteine we're all gonna make it brah

@LucioMM1 the best executives seem to manage a spreadsheet in their mind for the technical & financial ROI of whole departments to create & destroy at will. they need to have a great hiring ability, yes, but empathy doesn't seem as important

@LucioMM1 in short the direct reports of C level officers are not as fragile / do not require as much care as normies

@Teleonomic @halvorz what would you consider the quintessential philsci reading list

@R1Jack p much. big company was a big mistake

@halvorz @Teleonomic noted

extremely jealous of hedgehog thinkers

imagine believing in anything strongly

@halvorz foxes are mostly just good at pedantry on twitter

everybody wants to be an "expert generalist", nobody's good enough at doing it

this tweet is about sonic mains vs fox mains

it’s like evolution worked for 1m years to make my cognition good at twitter and not much else https://x.com/generativist/status/1412955809294684160

@lisatomic5 https://x.com/tszzl/status/1398524943663575042

@lisatomic5 and yet there seem to be quite a few absolutely pivotal expert generalists out there rn

@bantablack lmao am I the guy on the right

@PrinceVogel god I wish

@halvorz hmm Am I missing something

@nicholatian it’s not about becoming omniscient more like holding enough disparate info in one head to make novel synthesis and good decision making

@nicholatian rarely possible when a person has a skin deep understanding of 20 subjects, otherwise every high schooler would be changing the world

@MaiqTL yea I’m not saying I’m some sort of twitter genius, I just have a visceral sense that this place matches my cognition. the frenetic pace, the jumping around, the eclectic references

@MaiqTL essentially it’s breadth of knowledge for the sake of making great dinner party convo rather than for some kind of utility

@MaiqTL our minds

@MaiqTL god I hope you’re right

@arctanno I did, it’s pretty fun but generally the recipe is not that coherent

@reconfigurthing https://x.com/tszzl/status/1412634338076741634

@nicholatian @reconfigurthing I don’t see why simulation argument would have to rule out some religion — might just be that Creation was more complex and bigger than once thought. I mean did heliocentrism or the Hubble Deepfield kill god

@nicholatian @reconfigurthing “reality is fundamentally an illusion” this is just concomitant with hinduism

there are only 3 real expert generalists: - Tyler Cowen - Johnny von Neumann - elon

@reconfigurthing @nicholatian And I think while the argument may cross into metaphysics it doesn’t necessarily have to — there would be testable consequences of a universal simulation

@_Juliankaye @geraldbullvevo lol

@MaiqTL interpret that last tweet as a shitpost

@reconfigurthing @nicholatian there may* be

absolutely nothing that makes me dismiss a take as quick as if it mentions “using your right brain” or whatever

@arctanno based

@telmudic damn I read the wrong book

@halvorz @generativist this is how it starts

@Vitrosphere soon we will buy an island compound

@glockwitdanuts it boosted their user numbers by quite a bit and then they got a slap on the wrist fine from the regulators

@Higgs__Field yea i get it i know, im not here to complain

@Higgs__Field its more like, my brain is incredibly bad at focusing on work and even what i have accomplished is via a series of self-medicated copes. whereas it is naturally attuned to the frenetic pace of reading and writing on twitter

@Mihoda @eigenrobot something to be learned here

@eigenrobot i see what you mean yeah fair enough. global poverty is infohazard

@rbranson hmmm you mean spacex will be talent-sucked into spending all their time maintaining starlink?

@GuilleAngeris you gotta admit your hearts gotta jump a bit when you read the phrase "crypto synthetic assets"

@rbranson i guess it's an open question whether even with perfect information there remain topologies that will always be unprofitable for ISPs to serve and very profitable for starlink to serve

@rbranson (remain enough* topologies for the venture to be worth it tbc)

@moritheil @RahmSJ @eigenrobot yeah i simply don't think this is easily avoidable, considering all they do to augment the rice supply with vital nutrients

@lalaAlicelala his IG is so perfect i keep trying to read deeper meanings into his posts

@BarneyFlames yea and its apparently not even the literal poorest parts of the country. so less than -1 SD seems .. fine? obviously still sucks if it turns out to be preventable brain damage from anemia

@nat_sharpe_ gotta be the eyes moreso than anything

@nat_sharpe_ damn we gotta ask helen keller

@rplevy ah, a hedgehog

@judahwantssocks @GKCdaily Lmao

@VectorOfBasis @NibiruDakini ^

@shlevy yea has never once made sense to me

@sonyasupposedly @shlevy you can just modify the concept of existence to include self generation if that floats your boat

@sonyasupposedly @shlevy like if you look at many of the older polytheistic mythologies they often present the gods as players on an eternal stage and nobody knows how it got there — hinduism is like that

@sonyasupposedly @shlevy yea I understand the value add and many times I wish I had it; but emotional value add alone doesn’t imply truth for me

@alth0u based

@chrisgardenuk @sonyasupposedly @shlevy “Do as I say, not as I do”

The utility created by using debug strings like “poopoopeepee” is incalculable

@sonyasupposedly @shlevy in truth ive never heard a satisfying answer to “why is there something instead of nothing” so i don’t begrudge anyone theirs

@shlevy perhaps but this kind of verbal trickery is hardly satisfying — even a child can wonder why does all this shit exist

@shlevy I’ve heard the anthropic arguments and the “why is nothing privileged over something” argument and of course the god argument, but none of it satisfies the itch of the underlying curiosity

@PatrickFIanagan @tehsvenn lmao

@arctanno @shlevy true and I figure we can study for long enough that we figure out the how

@sonyasupposedly @shlevy Yea sorry to be more specific, the Brahman is supposed to be eternal, but brahman is taken to be an impersonal force (nobody prays to the brahman itself) so I’m not sure it’s the same thing

@sonyasupposedly @shlevy Similar in greek cosmogony https://en.wikipedia.org/wiki/Chaos_(cosmogony)

@hamandcheese of course

@Plinz using a weak definition of “materialism”

@Plinz my weak definition being “not religion” whereas the strong definition being “nothing more exists than matter and it’s governing laws”

love it when there are massive reorgs and then the operational metrics for your team don’t even change. the half year reorg is like the fake software project you create and destroy to get a promotion except at the director level

@BoldPoorYorick @MaiqTL if there is utility I haven’t seen it yet

@alth0u what do you mean by upward management

@Aiden_CDN @lalaAlicelala lmao

both ngmi https://x.com/briannekimmel/status/1413219695285522433

@finnitejest ooh whats that like

@niccoteine SF + bay area + seattle

@finnitejest sorry finn i don't have a good knowledge of the Outpost Colonies

as a habitual contrarian, extremely torn about simulation theory bc many a house party in sf is full of ppl who take it somewhat too seriously but then my whole twitter feed reflexively just calls it cringe scientism

@bronzebarbarian thats literally what i do

@bronzebarbarian i tell everyone how stupid it is irl and then sing its praises online

@LucreSnooker yes

@pee_zombie sorry low verbal iq rn : you talking about sf? or twitter? or what

@andrej_haulis the thing is it is absolutely not a religious belief structure and to call it that would be to degrade religion lol it's a theory of cosmogony that doesn't actually offer purpose/validation/salvation/any of those things like "hey your grandpa lived and died in a fake universe"

@andrej_haulis strawman: its not wish fulfillment enough steelman: its not beautiful enough

i didnt feel any earthquake therefore its a psyop

@Alephwyr lol

@andrej_haulis hmm but its not clear that the creator would be almighty at all -- after all theres no point in running a simulation if you're omniscient and already know what's gonna happen

@BecomingCritter does it make any real difference if the Higgs boson exist

@_djpn @bronzebarbarian too late for that, we need a new yolo stonk

@_djpn @bronzebarbarian love you too bro 😊

@BaruchKogan nice

@0knaomi @kevin_bowen @elonmusk von Neumann probe type deal

deleted lindy

@BecomingCritter yeah I’m also uninterested if it’s pure metaphysics / nontestable but it feels like it might be testable

@LudditeHacker @vbronjames lmao

@pee_zombie still don’t get the inner demons thing but maybe I will at any rate the weird thing is that there are no aliens

@BecomingCritter well potentially we could use this info to hack/ mess with our own sim or try to manipulate the sim runners or something. it’s hard to know the utility of knowledge before you have it. what you personally do if you knew the corona was a lab leak?

@p_millerd it’s not good lol

@aniiyengar lolol

@moultano yeah. I mean fuck rms but yeah

@thetinduck google last name

@LudditeHacker https://x.com/tszzl/status/1413267345523449856

@LudditeHacker disagree tbh — is all Big Bang cosmology religion then?

@andrej_haulis @vbronjames lmao yea similar

@LudditeHacker @vbronjames oh wait I remember this — lotta non-R1A1A copers VC @razibkhan

@LudditeHacker the difference is that I don’t think this is a metaphysical thought experiment. I have a strong hunch it would have physical consequences

@disconcision holy based

@LudditeHacker yeah I’m not thinking of a Cartesian demon — I’m thinking about a regular physical universe inside a computer. what special properties would it have?

@lalaAlicelala compared credit cards recently and felt this

@0x543211 bro it doesn’t matter tho bostrom didn’t try and answer every of life’s questions incl ultimate origins

@zcknln @pee_zombie lmao at any rate. twitter is far more hyperreal and bizarre than sf will ever be

@0x543211 i mean idk none of this is satisfying. the idea that the earth is not the center of the universe is unsatisfying

@LzSeishi @L0m3z seems like a stretch, why infinite?

@alth0u software margins are better than ML margins thats all

@alth0u https://a16z.com/2020/02/16/the-new-business-of-ai-and-how-its-different-from-traditional-software/

@MiloJKing yep

@alth0u it doesnt even have to be that the original chatbot business was a grift -- just a classic example of innovator's dilemma'ing yourself. make a high margin product on the pathway to making a much more interesting low margin product

@alth0u but i also can't rule out bubble / the real AI application will never appear / etc

@alth0u discord?

@alth0u lmao wow

@alth0u the moment Starlink goes public mars is doomed honestly. they will just get innovator's dilemma'd to fuck

@niccoteine ty plemya

@egodrifter https://x.com/tszzl/status/1412634338076741634?s=20

@JustAFactsMan @LucreSnooker lol elaborate

@IvyAstrix meme stocks hits so so different

this and all similar images should send chills down your spine. if you see this in a powerpoint you are about to get robbed blind

@eigenrobot @WidowsSon10 i believe in you

@BarneyFlames @alth0u lmao

@alth0u somehow make a zero sum economy out of this and you have an interesting concept

@alth0u reminiscent of AlphaStar League

@BarneyFlames @alth0u realistically, i don't see why you would need multiple D's or in this setting. just train one big one that takes extra inputs to disambiguate the "mode". bet you could do about as well, just have to make the model larger

@alth0u yea there was an interesting paper that nips last year that generalized alphastar league as an example of a zero sum loss game, i'll see if i can find it

@alth0u @BarneyFlames notably i see that they only compared vs pairwise trained CycleGANs and such. 1) less cynical possibility is that training a single discriminator for all datasets is unstable / doesn't work that well 2) cynical is that 1G:1D just would not be 'novel' enough to get published

@guavault the sigma grindset..

@guavault passive income

@guavault ok the real one is cultural/memetic evolution

@chaosprime we’re just not measuring goodly enough yet :)

@monodevice just you onio. it was localized in Somalia

worst part of any project is searching for the tools to do the job and continue googling in the hopes that something even better and more specialized exists

@monodevice dang you got me there

@ded_ruckus just do OMAD on accident, ez pz

@kyzcralotSxaM one time i started thinkin abt operating systems and it fucked me up programs running other programs broh

@kyzcralotSxaM but yeah the screen tends to make me dizzy

@Noahpinion ,get in the pod, eat the bugs

@NathanielLugh @MatiosTV i can tell you something unsatisfactory like "yea its called samsara" but i think no in that the telos suggested by sim theory is entertainment / science experiment

@NathanielLugh @MatiosTV however the suggested mechanism of action would change through the ages: clockwork, steampunk, industrial machinery, electronics, computers, quantum computers, etc

@AFPoster @Noahpinion that is bugs bro

@kyzcralotSxaM great fuckin book

@samson_hu slack is bad format for Q/A

@pee_zombie @deepfates @BenRatkey EquityZen also works but you gotta fudge some numbers

@samson_hu my big co is pretty open about technical QA

@alth0u gotta somehow asterisk that I got it for free

@finnitejest the values of the producer are somewhat different than the values of the consumer

you say you’re a “postrationalist” but you don’t ape tens of thousands into random stonks based on good vibes only

@finnitejest you don’t think that the American puritans looked down on profligate spending, gambling, drinking, etc?

@lisatomic5 @alth0u 🧐 lol interesting concept

@eigenrobot yeah this is solid

emh / I can buy only s&p is high modernism!!

@AdamNeumannsCoS tomatoes

@RichDecibels guess not

@garybasin least of your worries if you have an intelligence that powerful

>tfw you deliver miracles but several fiscal quarters late https://x.com/elonmusk/status/1413253018720612360

@telmudic ya

@rasalghulvc https://x.com/tszzl/status/1413397293785436160

@andrej_haulis got lost in the quadruple negative

@telmudic lol based?

@shaispiegel93 bro insights dont always have to be novel, sometimes you just forget and remember things

everyone log off. that's an order

@seconds_0 https://x.com/tszzl/status/1407839441528377345?s=20

a huge W for sim2real transfer. i remain extremely bullish on sim2real based AI robotics. if anyone knows of any startups working on this outside of pure research lmk https://x.com/AIatMeta/status/1413468328937639937

sidenote: if you believe sim2real you have to become a tesla bear of sorts. the data moat isn't tenable if you can generate a bunch of productive, but lower quality in-simulo data

@dataghees sim2real** is training in simulo and testing in real world , yeah

@dataghees https://ashish-kmr.github.io/rma-legged-robots/

@dataghees no, they mostly count on real-world data. waymo is heavy into sims

@Willyintheworld yeah, although it seems like from first principles simulating really wacky stuff that won't ever actually happen irl is viable with sims. like what happens if there's an elephant on the road? we can find out by simulating!

@Willyintheworld the latter is likely easier than the former. we can tune the friction constants willy nilly with a sophisticated friction simulator. whereas creating new art and such for the deer is difficult

@Willyintheworld lmao tru. tesla brain will be so used to navigating hellscapes in simulo that it's positively happy to drive on US-80

@sirsfurther sirsfurther is made up

@Nexuist not really, you don't need it to simulate shakespeares brain generating Hamlet, you need it to simulate his sheeplike leminglike crowd behavior driving thru the streets interacting with fast moving objects

@Nexuist consider that ai is superhuman at starcraft

@Nexuist @karpathy yeah this stuff is rough but they're edge cases in perception rather than control

@NLRG_ reinforcement learning -- algorithms that learn via trial and error

@rkarmani the author mentions the robot they purchased somewhere in his thread

@garybasin yep

the three body problem guy joined as a “scifi director” at an AI startup that sells uyghur recognition programs to the Chinese government lmao

@Nazaaakat @peachblvd men can only rotate shapes in situations where it offers no practical value

@jachaseyoung @_O_Null_ true but supporting the internment is probably the normie chinese opinion, would be strange if the author didn't toe the party line there. more interesting that he is actively involved

@djmicrobeads having a hype guy is not a bad idea, like Catherine having a resident philosopher

stop comparing him with orson scott card nerds saying some mean things about LGBT is not comparable to abetting internment camps lol

@HgTrismegistus yeah but doesnt mean i cant empathize

@BarneyFlames true. would be eternally loyal at any company that keeps greg egan or ted chiang on retainer

@notlinear1 agreed i don't feel strongly about this but notably blue checks who believe the normie narrative QTing and comparing to OSC

@notlinear1 fog of war, hard to see through

@HgTrismegistus @oppositenames this is fair but i don't believe in the ideology of "only x,y,z slice of the population can comment on the struggles of x,y,z". i wouldn't try to speak over a gay person on this matter, but neither would i hold my tongue on something obvious just because i'm not gay

@alth0u only outcasts would go it's lindy

@MegaBasedChad its god tier for reading kindle and such

@0xrohan @boobahub ty

@0xTomoyo @0xrohan ty

FOOOOOOOM https://www.wired.com/story/fit-billions-transistors-chip-let-ai-do/

disclaimer: it doesnt actually mean much that there are research teams at google and nvidia (each of which employs hundreds of researchers) experimenting with AI chip design. but ive seem some awesome sim2real papers in the area so i'm simping

@ollyrobot its the rationalists' term for a recursively self improving AI explosion https://www.lesswrong.com/tag/intelligence-explosion

@halvorz lmao yeah, take this one "seriously but not literally". im just cackling at the yudkowsky-hanson infighting going strong

@LudditeHacker is robin hanson "in silicon valley"?

@eigenrobot most scientific and technological progress is made this way the long tail

@eigenrobot dare i say "political progress" may be different tho if we can permit some whig history altho i keep seeing things about how young the american revolutionaries were

@ollyrobot want to rt this so bad but am coward

@postpostpostr @eigenrobot i dunno feel free to prove me wrong via an exhaustive accounting of all famous scientific papers and theirs authors ages, seems liek 2 much work

@kristineCreates basically reinforcement learning AIs training for hundreds of realtime years in simulation dojos and then being tested in the real world since its so hard to gather real world data

alright convince me not to become a Carlos vtuber https://x.com/codemiko/status/1413971287563767810

@Jelqmaster lmao fr. inshallah everyone can become an egirl

@GuilleAngeris true lol. better heuristics replace worse heuristics. anyway do you have a read on how impactful this is? google published some eye popping results but it was for a toy environment

@GuilleAngeris https://ai.googleblog.com/2020/04/chip-design-with-deep-reinforcement.html

@eigenrobot @LudditeHacker yeh letting early 20s idealists run things seems dumb & not really the steelman here

@eigenrobot @LudditeHacker the steelman is more like in areas where new ideas are more important than conservation should take young ppl seriously (science, tech, business, etc). the bright side of all those areas is that nobody can fuck around and become an accidental Maoist in charge of shit that matters

@eigenrobot @LudditeHacker you have to earn your keep by eg providing better products and services or writing papers that people want to read or things like that letting thousands of high variance flowers bloom works when the superstructure is bound by some kind of darwinian fitness

@eigenrobot @LudditeHacker in short thesis would be: young ppl prolly have worse ideas on average, but higher variance in that the outliers will b quite important

@GuilleAngeris why? RL would be the right approach due to the discrete action space and can't get rewards til the end

@GuilleAngeris lol big if true. will read more later but thats a higher level of grifting than i would've expected if they're trying to RL optimize convex opt problems

@no_lying_online @ViciousRicardan wrong

@GuilleAngeris isn't Go trained via RL and some tree search guidance?

@halvorz nice https://x.com/ESYudkowsky/status/1414676892674711554?s=20

@GuilleAngeris @DogIsABot seems hard to believe that nvidia and intel, some of the largest companies in the world, aren't doing all they can to optimize routing & placement with modern optimization methods. wish google could compare with, say, the state of the art internal tool at nvidia

@halvorz and to be fair, what we know of transformer models suggests that indefinite scaling is real https://arxiv.org/abs/2001.08361 can't easily translate those claims to bio brains tho obv nor more holistic intelligence criteria than "test set accuracy"

@isaac_lewis correct on all counts, but its still neat how far modern AI optimization has come. to the level where it's infecting the chips that run it

@candrianillborn @GuilleAngeris hahaha

@ded_ruckus yes agreed, but if simple layout changes work you allow incremental improvement without retooling the fabs. eg if you can somehow gain 5% power efficiencies here and there just by moving around nanowires w.o needing totally new fab techs that's a nice and cheap win

@ded_ruckus layout optimization would never get you a step size improvement, gotta look to the fundamental physics and engineering of fabs for that. even still, chip companies keep moore's law going via micro optimizations at every layer of design

@alth0u @ded_ruckus i tried this for a compilers project in college and the results sucked ass but i did it the wrong way probably need a language model to really get things going actively researched tho : https://research.fb.com/wp-content/uploads/2020/09/Machine-Learning-in-Compilers-Past-Present-and-Future.pdf

@alth0u @ded_ruckus i can also say that inside {BigTechCo} theres a whole group actively looking into rl 4 compiler opt

@jekbradbury @alth0u @ded_ruckus ML moves fast these days althou says things and they just happen

@monodevice so uncivilized onio

@jekbradbury @alth0u @ded_ruckus compilers already have heuristic cost models right? so presumably you could just drag and drop replace and leave it at that but doubtful

@jack_forlines @GuilleAngeris yeah; RL seems great for combinatorial problems

@SigmaNick dont do anything fancy, just go OMAD (one meal a day). easiest diet in the world

@MahmetTokarev i would be very surprised em is scifi technology -- we are not anywhere close to brain scanning i can clearly see the onramp from here to general intelligence

@monodevice tru

@ded_ruckus @alth0u ai that turns the universe to grey goo in an effort to get unengineer the GIL

@SigmaNick 1000 calorie deficit a day is p manageable for like a month

@ollyrobot yeah 100%

@ollyrobot the esoteric reading is of a country that made a faustian bargain with the technocapital gods

i ghostwrote this book

@argosopentech @robinhanson i mean, he is plain wrong, its hard to see the underbelly of jobs being automated

@jane_gatsby lmao

@dyson_spear you got me there

v2

@ConejoCapital lmao based

@perceptions420 finish the essay broh. low effort memes are a bad standin for clear logic

@djcentos6 they shouldve given that mf a computer in jail (and pray he doesnt smash it to pieces)

@alth0u @beaconsai @Linktree_ commoditize your complement (modern social media) :^)

@alth0u @beaconsai @Linktree_ linktree will rule the world

a piece of software has truly Made It once ppl start building accelerator chips for it

@bryancsk TPU

@TurboRational slack is built on electron eg javascript v8 engine & chromium and i'm sure many modern SoCs like M1 are benchmarked on browser performance and such

@TurboRational this can't be an accident https://www.macrumors.com/2020/11/20/chrome-m1-macs-80-percent-faster/

@allaboutchemist @bryancsk losing in some narrow sense, many still happily using TF

@GuilleAngeris hmmmmm https://x.com/tszzl/status/1415061488683782144?s=20

@GuilleAngeris they seem to at least have implicitly optimized for the number of chrome tabs you can open lol

@Teleonomic idk doesnt this suggest we can cut like 35 gigatons of co2 based on just renewables transition?

@GuilleAngeris yep, plus add in the video accelerators for streaming netflix

@Teleonomic sure steel requires carbon inputs, but again not a huge chunk of the pie surely we'll always neutralize some amount via CCS

@Teleonomic yeah i understand -- carbon inputs for heating blast furnaces and such but iirc this is like <5% of total carbon emissions

@ollyrobot could be wrong but i thought they lowered the albedo of all starlinks after the complaints came in?

@ollyrobot yeah erasing them is impossible but making them dark enough such that they don't overexpose astrophotography

@ollyrobot and i think they're invisible to the naked eye even before they lowered the albedo

@ollyrobot you can see them right after they launch but they gradually become darker as they gain altitude and adjust to the right orentation https://x.com/SeanMombo/status/1335722942701907968?s=20

@ollyrobot i c, i think it was re: this visorsat thing i was reading https://micky.com.au/spacex-makes-starlink-satellites-invisible-to-the-naked-eye/

@halvorz @edmundrollett so its not the symptoms themselves (coughing, sneezing, etc) that make ppl more infectious?

@Willyintheworld @timhwang lmao

@timhwang kardashian scale

@__femb0t @GuilleAngeris no one should speak about web development again. js talk should be forbidden in polite society

@__femb0t @GuilleAngeris "yeah i use the CLOSURE TRANSPILER to reduce my PACKAGE SIZES" - statements dreamed up by the utterly deranged

@natal_female like a kernel function?

@ollyrobot so my read is that the current starlinks at their terminal orbit are at like +5.5 mag brightness where +6 is roughly invisible to the naked eye in a dark field so yeah you will see this things in the darkness but they will be dim

@ollyrobot and if/when they add these DarkSat things (which amount to basically black hubcaps) it will go past +6 to be invisible in dark sites and the operational time of a starlink is like 2-3 years so if they launch the new program all bad sats should be outta the sky asap

@ollyrobot but of course its valid to wonder if spacex will actually follow through on using the darker sats if it adds cost and eats into their margins. my read is probably yes so that at the least they dont get into trouble with any governments but could be wrong. can see the concern

@ollyrobot ok so it looks like the VisorSat program was launched as of last year& successful >The result: the visors may have darkened the satellites to 31% of their former brightness, according to a new study. >Starlink may now be largely invisible to the naked eye https://www.businessinsider.com/spacex-starlink-visors-satellites-mostly-invisible-to-naked-eye-2021-1

@ollyrobot sorry to spam your replies with nerd bullshit im on stimulant fueled bender rn https://arxiv.org/pdf/2101.00374.pdf looks like the biggest problem is that there are outlier observations sometimes when the sat catches the light wrong and becomes pretty bright

so it looks like starlink's VisorSat program worked? >The result: the visors may have darkened the satellites to 31% of their former brightness, according to a new study. >Starlink may now be largely invisible to the naked eye https://www.businessinsider.com/spacex-starlink-visors-satellites-mostly-invisible-to-naked-eye-2021-1

@wowholyfucking yeah i mean forget the astronomers but i can understand complaints if there are 50,000 sats ruining everyone's naked view of the night sky

i wish sf was half as cyberpunk as twitter makes it out to be lmao https://x.com/danielsinger/status/1415074189992595457

@CaesarSalad99 recently went on a night time stroll through the good 'ol 'loin and had a blast

@CaesarSalad99 yeah theyre like catatonic in weird positions. sad but completely harmless

@metorotem @alt1na1 wait i gotta get into bouldering

@ThatsMauvelous @eigenrobot reminds me of something you said

@ThatsMauvelous cautiously taking my rt back bc some of my followers will consider this crass materialism but its tru and i dont think its a bad thing. need more natural MDMA plugs

@gbrl_dick i think its like an early cycle in innovation curves marginally adding a computer + touch screen makes most products worse but is a necessary local minimum before shit becomes cool

@gbrl_dick at first they were making smart fridges that were doing fuck all like twitter for samsung smart fridge or see what items are already inside the fridge among other stupid shit but now they're doing like predictive processing of what produce you're gonna need and auto ordering it

@gbrl_dick lot of the IoT discourse among tech ppl is classic disillusionment stuff, lack of imagination, etc

@gbrl_dick yeah fr many things should have either (1) a frontend that is an app on your phone (2) voice UX rather than creating permanent liabilities to write frontend software for fridge touchscreens that nobody will want to work on after like 2 yrs

@tszzll bro do you not

@parafactual culture war tweeting is easy mode

@parafactual ive already prestiged and now playing Challenge mode

@parafactual was poasting incredible bait like this just 6 months ago. ive grown so much https://x.com/tszzl/status/1343743644638060544?s=20

@parafactual 18 months ago is a total no man's land. who even is this person? https://x.com/tszzl/status/1225904876988833793?s=20

@BroductManager she's for the the streets

@halvorz ya this is the only one that pisses me off usually in these debates the effect sizes are trivial and study results are uncertain it won’t matter either

@GuilleAngeris @__femb0t

@gbrl_dick @AutismU21 thanks for looking out for me buddy. I got hittas down undah

@bekyand8numbers @NLRG_ wtf don’t resurface this cringe @NLRG_

@houellebecq_2 both are wrong 😇

@halvorz way

weary google eng manager: guys please could we ship a single product this year that would be nice google engies: fuck off im documenting https://x.com/mmitchell_ai/status/1414983007350390787

@houellebecq_2 nerve center of the internet

@BaruchKogan @houellebecq_2 the impending doom is a feature

@houellebecq_2 I can spit at the doorstep of twitter everyday

@Hellachans Google glass was the last time a reasonable person could be hopeful about the future.

@lisatomic5 sigma trait

@lisatomic5 autism* trait

@notlinear1 lol

@notlinear1 to be honest the worst kind of SWE is not “product first” or “tech first” but like “process first” gotta write a design doc before they wipe their ass

@halvorz @telmudic wait how tf do you get a PhD and never publish once

@BroductManager @telmudic lmaooo

the world if the Power Structures of Harvard hadn’t Burried Eric’s Work

@bigbadmeatball Hank

@alt1na1 wait I wish my college dorm was like this though

@alt1na1 so much space

@meteor_cultist @alt1na1 bunk beds r not a sigma move

ive actually calculated that this is unfashionable and therefore based and sigma https://x.com/Joshmedia/status/1414381981295685640

@meteor_cultist @alt1na1 yeah the bunk beds create a lot more space but you gotta be comfortable af with your roommate

@LudditeHacker getting laid is just too easy. sigma plays it on hard mode with the climate bed

Pessimists get to sound smart. Optimists get to make money. Me and @ori_living are going the moon and the rest of you are NGMI

@0xGray all part of the ori living experience

@Hellachans @ori_living you’re just not ready for Ori Living

@Hellachans @ori_living you’re simply not ready for the true facts that a $20,000 bed is resource and climate efficient, actually

@nominalthoughts blocked

@eigenrobot @houellebecq_2 the companies are more important than the compute

@WillManidis My Life As A Language Model

@jeff82874662 says the Brooklynite to the Salt Of The Earth Midwesterner

@jeff82874662 when’s the last time you cooked your own meal bugboy

@OswaldHurlem I got back here like 3 months ago

@GuilleAngeris @eigenrobot @houellebecq_2 I dunno I just seem to meet an absurd concentration of interesting people here, especially off of twitter everyone has their own project going in the background that they’re willing to pitch at a moments notice and finally I just like going to work in a high rise

@GuilleAngeris @eigenrobot @houellebecq_2 not even living in a dump rn and my rents $1000/month

@grantadever @jeff82874662 I have seen Jeff’s pod with my own eyes. It’s quite a chadly pod but a pod nonetheless

most suspicious command line tool is jq

@jdoliner if you know you know

@wannabegroncho latter doesn’t work apparently

extreme blackpill https://x.com/chiefofstuffs/status/1415349403385098242

@kitten_beloved 😅