@tszzl — page 42/103

2021-07-01 → 2021-07-06 · posts 20501–21000 of 51,350
is “sus” really from among us feel like I’ve been saying it for a decade
125 ♥ · 3 RT · x.com →
· ↳ reply to @arctanno
@arctanno just pick up a picture book Mahabharat because thats about how deep my understanding goes
7 ♥ · x.com →
@chaosprime @deepfates hmm i dont think real neurons are doing bayesian chain rules nor gradient descent
15 ♥ · x.com →
@chaosprime @deepfates theres no obvious mechanism to backpropagate errors in the brain but my hunch is that you're right via some circuitry we don't understand yet
5 ♥ · x.com →
quote dunking neil degrasse tyson is punching down
254 ♥ · 7 RT · x.com →
· ↳ reply to @gomert6
@gomert6 punching up accepts risk, shows courage, and is generally just funnier punching down is lame. stealing candy from babies is not cool
16 ♥ · x.com →
· ↳ reply to @gomert6
@gomert6 i agree minorities is punching up. ndt is punching down lol half the arbitrage of the twitter anon account is finding the real power structure and punching up
7 ♥ · x.com →
· ↳ reply to @krishnanrohit
@krishnanrohit he was against the companies having to pay in cash against their theoretical valuation but seems alright with the taxes paid in equities
1 ♥ · x.com →
concept: AI-generated Zyzz on my phone that periodically tells me that we're all gonna make it
69 ♥ · 5 RT · x.com →
recurring nightmare where i have to explain what a postrationalist is to my dad
91 ♥ · x.com →
· ↳ reply to @monodevice
@monodevice tjos what happens when beta worker bee meets sigma chad playing Lil Jon at 9 am
3 ♥ · x.com →
· ↳ reply to @alth0u
@alth0u The God-Whispers of Donald Trump 12:28
1 ♥ · x.com →
tfw bonobo are women and chimp are men
210 ♥ · 11 RT · x.com →
· ↳ reply to @tszzl
"bonobos were more skilled at solving tasks related to theory of mind or an understanding of social causality, while chimpanzees were more skilled at tasks requiring the use of tools and an understanding of physical causality".
62 ♥ · 3 RT · x.com →
· ↳ reply to @jbarro
@jbarro its bad to make you click and its literally on business insider i will lose respect for anyone who QTs this and raises its price in the wet market of ideas
103 ♥ · 2 RT · x.com →
· ↳ reply to @tszzl
the medicine is much more pleasant than the disease is bad
12 ♥ · x.com →
gpt3 is probably smarter than quite a few humans. what are the ethical consequences?
180 ♥ · 1 RT · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus disagree symbol manipulation is an iq test and important to how we intuitively understand human intelligence
18 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus subjectively I think this is untrue and gpt can hold quite a reasonable convo up til it runs up against the hard limits of its memory
7 ♥ · x.com →
· ↳ reply to @tszzl
i consider it some kind of alien intelligence that doesn’t directly deserve or require sympathy but i do think it can be evaluated along human intuitions of “smart”
43 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus yeah so this is one area where it’s dumber than most humans. but we’re still happily (and accurately) evaluating an alien intelligence against human benchmarks
2 ♥ · x.com →
· ↳ reply to @tszzl
@ded_ruckus not strange that IQ tests seem to be divided into the components of “visual” “verbal” and “memory”
· ↳ reply to @ded_ruckus
@ded_ruckus broaden the space of activities then it’s general because it can universally demonstrate limited memory at a large set of tasks
4 ♥ · x.com →
· ↳ reply to @brianduffytweet
@brianduffytweet it’s not impossible; in fact I’d be surprised if it didn’t get this question correct sometimes. just has to be a part of the same input text
8 ♥ · x.com →
· ↳ reply to @tszzl
lotta people displaying a lack of imagination abt the space of all intelligences
48 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus as in GPT can demonstrate limited performance at digit span and all other such challenges you throw at it; it is general and stupid. it is categorically different than a pre programmed digit span solver
5 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus doesn’t matter imo, it’s an intelligence oriented towards a different purpose. no less of an intelligence. we can gain insight into how smart it is by playing novel word games
4 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus you can compare entities with vastly different teleologies on their level of intelligence I think
4 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus like let’s say we’re comparing dolphins to humans dolphins will not have anything like the level of symbolic manipulation ability we have. but i assume their spatial reasoning and memory will knock ours out of the water. both creatures are general but have different purpose
7 ♥ · x.com →
· ↳ reply to @tszzl
@ded_ruckus or compare to aliens that grew up in an evolutionary environment that highly rewards the ability to predict the next word of text
5 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus reasonable people can disagree on category boundaries But i think if we can’t place dolphin intelligence and human intelligence in the same one we fucked up somewhere
1 ♥ · x.com →
· ↳ reply to @tszzl
reasonable people can disagree on category boundaries so “category errors” are fake
30 ♥ · x.com →
the only way to settle category arguments is to outsource them are human intelligence and dolphin intelligence categorically same or similar enough to compare?
34 ♥ · x.com →
· ↳ reply to @ResonantPyre
@ResonantPyre I would say it can outperform a subset of humans at any arbitrary verbal task you can throw at it
1 ♥ · x.com →
· ↳ reply to @brianduffytweet
@brianduffytweet yeah you are the result of a trillion incomprehensible screaming neurons so i don’t think the fact that we can directly see the internal probability calculations is a strong argument against an intelligent nature
12 ♥ · x.com →
· ↳ reply to @tszzl
@brianduffytweet if we had more insight into what neural circuits are producing this conversation today it would seem similarly absurd
5 ♥ · x.com →
· ↳ reply to @tszzl
dont wanna skew the poll but my answer is an emphatic yes, it’s all a continuum. im not that different from Dog
24 ♥ · x.com →
@rud_talaiporos @deepfates @brianduffytweet there are models that ground language in images -- look up DALL-E or CLIP this is not really a radical complaint nor do i think it changes the phenomenon of intelligence
2 ♥ · x.com →
@bowserbot2 @roisbelh wonder what the dolphin religion looks like because I’m confident they’re smart enough to have social cohesion via mythmaking
6 ♥ · x.com →
life evolves on a planet where survival depends foremost on doing correct completions of text snippets is the end result intelligent
81 ♥ · 3 RT · x.com →
@HgTrismegistus I’m not arguing about your definition but rather your confidence that it doesn’t understand or your belief that you personally understand
1 ♥ · x.com →
· ↳ reply to @arctanno
@arctanno @bowserbot2 @roisbelh oh yeah actually now I’m remembering how orcas have advanced evasive maneuvers to escape the seaworld trappers so there must be some kind of collective memory thing
3 ♥ · x.com →
@halvorz aren’t there animal models for this and also what’s your bet / hunch
5 ♥ · x.com →
· ↳ reply to @BaruchKogan
@BaruchKogan this is just Chinese room shit i can’t prove that you understand anything or are just a p zombie so it’s irrelevant to me
2 ♥ · x.com →
· ↳ reply to @tszzl
@BaruchKogan I’ve played with it for a long time and my intuition is that it mostly understands, even extrapolates, and sometimes rarely analyzes
1 ♥ · x.com →
· ↳ reply to @eigenrobot
@eigenrobot its a pretty far step from "a lot of science is shitty" to "science is useless"
22 ♥ · x.com →
· ↳ reply to @tszzl
@eigenrobot if anything i think the long tail of shitty papers are outweighed by the exponential outsized impact of the really good ones
31 ♥ · x.com →
"the customer is divinely discontent" - bezos imagining jeff bezos' disembodied cunsoomer as a kind of lovecraftian god, a utility monster that never stops eating
178 ♥ · 18 RT · x.com →
· ↳ reply to @tszzl
@eigenrobot however in this case i would agree "researching sex is pointless"
9 ♥ · x.com →
· ↳ reply to @tszzl
@rglpwx @alth0u @ded_ruckus personally I want my language models to be able to have external data stores that they can read & write to during the training process and then also some kind of long term memory mechanism
24 ♥ · x.com →
· ↳ reply to @tszzl
@rglpwx @alth0u @ded_ruckus they are wasting their time memorizing so many arbitrary facts like the autistic but stupid child that knows the name of every lieutenant in the civil war but can’t synthesize a narrative about why the war happened
22 ♥ · x.com →
· ↳ reply to @tszzl
@rglpwx @alth0u @ded_ruckus part of that is just due to the training modality which rewards accurate recall over synthesis but I’m wondering if you can’t be more efficient with brain space via external data stores, like “note taking” for the AI
12 ♥ · x.com →
· ↳ reply to @tszzl
honestly I wanna wake up every morning and see the sea walls
33 ♥ · 1 RT · x.com →
@halvorz software is an organic system and the guy that won’t approve my PRs doesn’t get that smfh
15 ♥ · x.com →
· ↳ reply to @tszzl
@halvorz “we are rationally designing this product” no we are organically hill climbing by adding features of essentially random quality
18 ♥ · 2 RT · x.com →
nobody cared who i was until i rotated the shape
164 ♥ · 3 RT · x.com →
· ↳ reply to @rmcwhorter99
@rmcwhorter99 @eigenrobot this is a case of people seeing what they want to see imo. when evaluating Silicon Valley you can either point to the spectacular failures (Theranos, WeWork, etc) or the tremendous successes (SpaceX, google …)
1 ♥ · x.com →
· ↳ reply to @tszzl
@rmcwhorter99 @eigenrobot yeah social psychology is a trash fire, and my question is so what? why abandon the storied tradition of academic science because the milgram experiment sucked or something
1 ♥ · x.com →
· ↳ reply to @tszzl
@chaosprime dude is natural if hes not violating the laws of nature but he sure as hell feels supernatural
11 ♥ · x.com →
· ↳ reply to @arctanno
@arctanno gotta take more adderall until my autism beats my loneliness and i get off this site forever
4 ♥ · x.com →
· ↳ reply to @alth0u
@alth0u nah when the grift works these guys make way more money than actual faang-cucks
15 ♥ · x.com →
· ↳ reply to @tszzl
@alth0u if yuor so smart why aren't you selling leetcode courses at scale to desperate indians
16 ♥ · 1 RT · x.com →
· ↳ reply to @tszzl
@alth0u i would say theres a gap between "guy who sells his labor teaching" and "guy who sells his brand capital"
8 ♥ · x.com →
· ↳ reply to @_rajkhare
@_rajkhare @alth0u (1) proof of work -- shows willingness to cooperate & perform the ritual, good candidate for employee (2) proof of basic intelligence -- generally the problems are harder than "invert binary tree" it requires some synthesis
9 ♥ · x.com →
· ↳ reply to @eigenrobot
@eigenrobot ya all this is totally fair but the real redpill is that social science isnt science we need to unperson the social sciences
14 ♥ · x.com →
· ↳ reply to @tszzl
@eigenrobot if you publish a wrong and dumb paper about physics or computer science for the most part some bureaucracy of experts is not about to come and fuck you over
7 ♥ · x.com →
· ↳ reply to @tszzl
@eigenrobot worst case is that it leads to a misallocation of researchers & investment funds. the damage of the former is limited to academia and the latter will get culled quickly by market forces
5 ♥ · x.com →
· ↳ reply to @realtimeai
@RealtimeAI ??? you can't accomplish a single task other than moving your body parts
1 ♥ · x.com →
but moooooom paul skallas said breakfast isn't lindy
112 ♥ · 6 RT · x.com →
· ↳ reply to @jeff82874662
@jeff82874662 NGMI jeff you're describing all the growth features that make me fear and respect them
4 ♥ · x.com →
· ↳ reply to @tszzl
@jeff82874662 this is why they have 20 million weekly active users and we have 0!!
3 ♥ · x.com →
· ↳ reply to @aashiq
@aashiq @ded_ruckus yeah this whole line of argument is spurious, considering models like DALL-E have semantic grounding in image manipulation and generation "it can only do language!" is a god of the gaps argument. again people are not imaginative enough about the space of all intelligence
1 ♥ · x.com →
· ↳ reply to @aashiq
@aashiq @ded_ruckus most cognitive tasks can be reduced to input output as text one agent can describe an image and another purely textual agent can make conclusions about the world state described
1 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus @aashiq sure that's fair my only point is, (1) synthesizing image and text modalities is not unheard of or novel and (2) the input/output modes of a model does not change whether it is functionally intelligent or not
1 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus @aashiq re: 2 there are some humans that couldn't solve all the problems either! lol if you look at my OP it's a very weak but true claim (1) doesn't really matter to the discussion at hand. humans might be better at thinking if they had 5 more senses, alas we don't
1 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus @aashiq well i should've said at the outset: i think you can only judge an intelligence as a black box of its inputs and outputs
1 ♥ · x.com →
· ↳ reply to @tszzl
@ded_ruckus @aashiq people will try to bring in the chinese room idea to mess with your intuition but in real you would need infinite space to construct a chinese room but i assure you there is not enough space in GPT's head for that. its brain's data capacity is 1000x smaller than ours
4 ♥ · 1 RT · x.com →
· ↳ reply to @tszzl
@ded_ruckus @aashiq which means we have to deduce that it's making novel synthesis, understanding, extrapolation even if its telos is vastly separated from our own. "intelligent" but not "conscious"
2 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus @aashiq how can i model you in any way other than a black box? what i put in is text and what i get out is replies -- the shape of the response system looks to me like a human
1 ♥ · x.com →
sometimes i wonder why folks on here will get into 50 reply long argument chains in the replies about every imaginable topic then i take adderall and i understand
97 ♥ · 4 RT · x.com →
· ↳ reply to @tszzl
alternatively just rile me up on some topic i sperg out about (AI/brains)
28 ♥ · x.com →
the redpill is nobody really cares about intelligence of the Subject as a moral axis; if they did we would feel more guilty eating pork than eating chicken but i don't think about it
192 ♥ · x.com →
· ↳ reply to @tszzl
if you are really going to commit to the bit on 'continuity of consciousness' gotta stop eating meat probably
49 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus the bit that human intelligence is very much on a continuum from the intelligence of a dolphin or a dog or a pig
1 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus yeah which is vaguely like "human vs nonhuman" or "how much can it bond with humans and humans bond with it"
1 ♥ · x.com →
· ↳ reply to @tszzl
@ded_ruckus if they come out with AI bots that are designed to make people love them (and are successful in doing so) maybe there'd be a reckoning about whether we should copy, delete, and recreate them on a wanton basis
1 ♥ · x.com →
@chaosprime explain; its very much accepting the fact that I am meat two conclusions 1) accept cannibalism 2) go vegetarian
5 ♥ · x.com →
@chaosprime (1) there is really nothing that eats us these days (2) just because the lion is too stupid to have a theory of mind about me doesnt mean i cant have a theory of mind about it
4 ♥ · x.com →
· ↳ reply to @tszzl
however i will continue lacking the courage of my convictions
28 ♥ · x.com →
· ↳ reply to @tszzl
My convictions are not happy with me--that's okay, I'll still keep eating that meat.
36 ♥ · x.com →
· ↳ reply to @tszzl
@aashiq @ded_ruckus generalization is intelligence the trappings of a human cognitive structure is one implementation of intelligence
2 ♥ · x.com →
· ↳ reply to @LudditeHacker
@LudditeHacker @ded_ruckus yeah this is getting unproductive but broadly speaking all things are continuous and we can create mental categories when there's evidence of a "cluster" in thing-space. i very much believe that intersex exist but also male and female are real clusters
1 ♥ · x.com →
· ↳ reply to @tszzl
@LudditeHacker @ded_ruckus so when i say its all on a continuum its that to me there doesn't seem to be a "cluster in thing space" when comparing all known intelligences around humans that make them absurdly different from dolphins. but this may be self referential and silly i'm not sure
· ↳ reply to @aashiq
@aashiq @ded_ruckus yeah i'm gonna push back on this in that "information prediction" is underspecified
2 ♥ · x.com →
· ↳ reply to @aashiq
@aashiq @ded_ruckus yeah that's true but the distinctive quality of both GPT and humans is the *general purpose nature* or in other words, the breadth of inputs the black box can successfully parse a black box that is just doing y=mx+b on the inside is not an intelligence
1 ♥ · x.com →
· ↳ reply to @PstafarianPrice
@PstafarianPrice @eigenrobot @bowserbot2 yeah me 2 but i'm not on here arguing about obamacare anymore or something. very boring, we've all heard the pro-side and anti-side a million times. i didnt realize we could just retire such topics from the discourse
13 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus @aashiq hmm not really, i played with it as an open-ended exploration without intending to apply an IQ test on it. I had it autocomplete some of my old DnD characters, some weird speculative fiction. my overall intuition is that there is undeniably something intelligent in there
2 ♥ · x.com →
· ↳ reply to @tszzl
@ded_ruckus @aashiq it scares me at times with its extrapolative ability. it makes jokes that it should not be able to make. it copies my writing style near perfectly (loses track when the passage gets too long though).
1 ♥ · x.com →
· ↳ reply to @tszzl
@ded_ruckus @aashiq i stored all these conversations somewhere, maybe i will write a blog post about it
2 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus @aashiq i feel like we've been through this already but i don't see why not having access to all the motor functions or senses of a human might make something unintelligent. paraplegics are not unintelligent, neither are the blind or deaf etc
2 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus @aashiq no i think a smart enough GPT could absolutely figure this out if the data read-in serialization was good enough
2 ♥ · x.com →
· ↳ reply to @tszzl
@ded_ruckus @aashiq i mean like have you seen what kind of input RL bots use? its generally dumbed down to 2 or 3 parameters rather than the complexity of the real world sensiromotor stuff and they still make do
2 ♥ · x.com →
· ↳ reply to @tszzl
@ded_ruckus @aashiq i mean I get your point that the real world is difficult but verbal reasoning is difficult too; its just that its super easy to scrape the web for all the pre-existing verbal reasoning that we can throw at the robot and very hard to find data on real world manipulation
2 ♥ · x.com →
· ↳ reply to @ded_ruckus
@ded_ruckus @aashiq no no i don't mean training i mean learning during test-time like we've seen with GPT-3 thus the title of the paper, "language models are few shot learners"
2 ♥ · x.com →
· ↳ reply to @tszzl
@ded_ruckus @aashiq broader claim is that you don't need to be capable of doing every motor task humans can do to be intelligent. shouldn't grade dolphins on their ability to climb trees
2 ♥ · x.com →
· ↳ reply to @tszzl
@ded_ruckus @aashiq but i think we've talked ourselves into a circle a second time so gonna just mute thread now
1 ♥ · x.com →
· ↳ reply to @alth0u
@alth0u something very kabbalistic about language models
18 ♥ · x.com →
· ↳ reply to @alth0u
@alth0u it really is time to unfollow each other
8 ♥ · x.com →
@chaosprime i understand what you mean but i think there is inherent beauty in the wilderness and not much beauty in the way my food gets to me
3 ♥ · x.com →
· ↳ reply to @tszzl
@TheNewGeraldo also i wonder if WEF & the economist and that cohort shills great reset / eating bugs stuff for clicks
1 ♥ · x.com →
if I had any verbal iq to speak of id write essays instead of tweeting
83 ♥ · 1 RT · x.com →
· ↳ reply to @houellebecq_2
@houellebecq_2 practically speaking all moral systems are based on Void and rely heavily on subjective inferential jumps and psychological/cultural instincts
3 ♥ · x.com →
· ↳ reply to @lisatomic5
@lisatomic5 i guess it makes them more aware of unearned confidence … but some have earned their confidence
2 ♥ · x.com →
wtf this is dumb it’s a great conversation starter
51 ♥ · x.com →
does anyone know if there’s a “pagerank for twitter” service anywhere
36 ♥ · 1 RT · x.com →
· ↳ reply to @garybasin
@garybasin i mean like a “reputation score” for users based on their following and followers eg bunch of 0 follower bots follow me vs lots of giant accounts follow me
10 ♥ · x.com →
· ↳ reply to @houellebecq_2
@houellebecq_2 even if you have a bedrock of divine authority you could be validly asking these questions tbqh
1 ♥ · x.com →
· ↳ reply to @tszzl
@houellebecq_2 the Jains struggle amongst themselves whether it’s ok to be hurting lower life like insects, bacteria etc
1 ♥ · x.com →
· ↳ reply to @tszzl
@houellebecq_2 ok yeah I guess you weren’t arguing and I’m just being combative Sry!
· ↳ reply to @aashiq
@aashiq lmfao category fascists out in full force
1 ♥ · x.com →
· ↳ reply to @lauralondon_
@lauralondon_ @lisatomic5 im basically like a modern language model ... in tweets or short threads i can be pithy and punchy at times but i lose focus and get off track when the writing gets too long and i overrun my token limit
4 ♥ · x.com →
@nicholatian @pee_zombie i wonder if there would be value in just maintaining a large decentralized database of all scraped twitter metadata in a way that the feds can't easily takedown
1 ♥ · 1 RT · x.com →
why would numerai work when: - there's inherent tension between the quality of the data and the need to obfuscate it (which the hedge funds don't) - the best algorithmic traders are getting carry at a hedge fund somewhere
31 ♥ · x.com →
@ollyrobot tbf there would be fewer oil pipelines and fewer accidents if the demand for it was lower
9 ♥ · x.com →
· ↳ reply to @tszzl
@ollyrobot not that any individual person has that much control obviously but i don’t see it any differently than like voting an individual decision to vote is a ridiculous waste of time but the collective voting behavior sometimes works
9 ♥ · x.com →
· ↳ reply to @tszzl
@patio11 @eigenrobot plus all the smart + risk-seeking people want to build own thing and have their shares bought out rather than build it inside the company and get a tiny bonus
10 ♥ · x.com →
· ↳ reply to @tszzl
@BarneyFlames defaults in the first place caused by adjustable rate mortage loan with premiums exploding when fed funds rate goes up ... whole thing attributable to cringe monetary policy?
2 ♥ · x.com →
· ↳ reply to @tszzl
@BarneyFlames otoh banks should have financial modeling that can withstand stress test of rates going up
1 ♥ · x.com →
inflation is always fake
27 ♥ · 1 RT · x.com →
· ↳ reply to @tszzl
@Coldoorr @notlinear1 @BarneyFlames the real michael burry talks about this regarding e.g. "nobel approved models of risk" that seem to think diversifying via highly correlated assets makes the bundle safe
2 ♥ · x.com →
· ↳ reply to @moultano
@moultano @ollyrobot the effect of voting is p small, so i think you can also like make different decisions on the margin assuming you’re financially secure enough eg buy electric car
1 ♥ · x.com →
· ↳ reply to @WillManidis
@WillManidis of the two points (1) is the stronger one. I agree there may be quite good quants left out on the margins
1 ♥ · x.com →
· ↳ reply to @moultano
@moultano @ollyrobot yeah that’s not a trivial amount tho, for anericans it means cutting like 10 tons of CO2 per year. obv I wish we could just carbon tax but that’s not happening for various reasons
1 ♥ · x.com →
guy who reads a holy book then gives it a star rating on goodreads
285 ♥ · 6 RT · x.com →
· ↳ reply to @moultano
@moultano @ollyrobot planting a tree now is no guarantee that the offset company makes sure it grows into a real carbon sink
1 ♥ · x.com →
· ↳ reply to @alth0u
@alth0u when I got to this point in the thread I went back and unliked everything
21 ♥ · x.com →
r u on the karma yoga or jnana yoga
18 ♥ · x.com →
@iluvjoerogan tell him that lindyman said modern yoga was made up by Europeans
2 ♥ · x.com →
· ↳ reply to @luke_metro
@luke_metro self hating tech people are far worse than any of the actual wrongs committed by tech
11 ♥ · x.com →
· ↳ reply to @arctanno
@arctanno I don’t know what dual N back is really but I know there are several NLP datasets that resemble iq type problems
2 ♥ · x.com →
· ↳ reply to @tszzl
@selentelechia the idea of the religious gods being real and powerful as thoughtform
6 ♥ · x.com →
· ↳ reply to @tszzl
@Hellachans @gigafelon @NotComradeSnake moreover they are the clear unparalleled category winner in most of the 1st world unlike eg ridesharing where it’s settled into a duopoly in the US and many fractaline players in other countries
· ↳ reply to @tszzl
@Hellachans @gigafelon @NotComradeSnake i guess there’s a feeling of safety coziness and lifestyle associated with the Airbnb brand — even in the middle of the pandemic Brian chesky chooses not to get rid of the 200 person Airbnb magazine division but cuts a bunch of engineering instead
2 ♥ · x.com →
· ↳ reply to @Hellachans
@Hellachans @gigafelon @NotComradeSnake oh yeah I have zero loyalty I would gladly switch to a cheaper platform — but I assume some sort of network moat exists else they wouldn’t burn billions on subsidizing growth markets
1 ♥ · x.com →
· ↳ reply to @realtimeai
@RealtimeAI yeah I’m just being aggravating here for humor value I’m sure he meant nothing of the sort
1 ♥ · x.com →
@halvorz idk aren’t we all doing that all the time
9 ♥ · x.com →
· ↳ reply to @DesidErratum
@DesidErratum @halvorz you’re setting the bar too high when you choose to read a book do you not end up learning new things and therefore modifying your mind?
3 ♥ · x.com →
· ↳ reply to @DesidErratum
@DesidErratum @halvorz based on what you learned you might try to read another book ad Infinitum that being said I think I could collapse my mind if I had “root level access” to play around
3 ♥ · x.com →
· ↳ reply to @monke_io
@pupperio you think if I had any impulse control I’d be sitting on a backlog of 30 books
3 ♥ · x.com →
· ↳ reply to @tishray
@tishray yeah same but at least these have the benefit of looking pretty on a shelf
2 ♥ · x.com →
concept: digital display bookshelf to flex your unread kindle collection
45 ♥ · x.com →
far from radicalizing me, being online for so long gave me deep respect for even the most esoteric ideologies
191 ♥ · 5 RT · x.com →
· ↳ reply to @tszzl
“Hoppean monarchism” yea I get where ur coming from bro “Juche socialism” I understand the appeal dawg
84 ♥ · x.com →
· ↳ reply to @_debarshi
@_debarshi yeah I can empathize with how a low IQ fellow might find this theory convincing
2 ♥ · x.com →
Happy 4th!
69 ♥ · x.com →
· ↳ reply to @tszzl
@bastard_brian i think it’s more about personality, high openness low neuroticism etc
9 ♥ · x.com →
bottomwit: ocean fire look cool 😎 midwit: this endorses capitalism/socialism/etc upper midwit: Noooooo you can’t draw overarching conclusions about economic systems based on random engineering failures topwit: ocean fire look cool 😎
172 ♥ · 9 RT · x.com →
unfollowing all non-autists. prove your aspie credentials to me rn
363 ♥ · 3 RT · x.com →
· ↳ reply to @Plinz
@Plinz ty for this high effort shitpost you’re in
34 ♥ · 1 RT · x.com →
· ↳ reply to @tszzl
the thing is I don’t even think taleb was right but this line of reasoning was so specious that I had to take his side …
14 ♥ · x.com →
@bowserbot2 for context: I don’t think Taleb is saying IQ doesn’t work his thesis (which I’m not endorsing) is that IQ works pretty well as a predictor under a certain limit, then seems more or less noise, which he tries and demonstrates with a simulation of some distribution
5 ♥ · x.com →
· ↳ reply to @tszzl
@bowserbot2 even if what he says is true IQ should work as a simple predictive statistic as Claire keeps saying. but the idea that you can’t compare several hypothetical explanations for a dataset from a theory level argument seems very silly
2 ♥ · x.com →
@bowserbot2 @telmudic I’m willing to admit I am purposefully missing the forest for the trees here because I don’t like her phrasing
6 ♥ · x.com →
· ↳ reply to @phl43
@phl43 @bowserbot2 yeah tbh I deleted because I don’t care strongly enough about this — I’m sure Taleb cherrypicked a dataset that works well for his argument
2 ♥ · x.com →
· ↳ reply to @tszzl
@phl43 @bowserbot2 I was just mad because I think simulating a distribution and seeing how it matches the observables is an entirely valid way to learn things
2 ♥ · x.com →
· ↳ reply to @boop
@boop it’s fine girls get a pass
11 ♥ · x.com →
· ↳ reply to @tszzl
this is my greatest baitpost of all time
41 ♥ · x.com →
making money is an entirely unrelated hobby from spending money
204 ♥ · 10 RT · x.com →
i know now I will never be truly great at anything in life . except poasting fire tweets
92 ♥ · 1 RT · x.com →
@finnitejest bro your tweet is a quote of what I said in the gc
6 ♥ · x.com →
· ↳ reply to @ghosttyped
@ghosttyped yeah this is cope tho the qualities that made them good at making money also make them less interested in lots of leisure and indulgence
1 ♥ · x.com →
· ↳ reply to @tszzl
this twete is solid because there’s two ways to read it and everyone will pick their favorite
23 ♥ · x.com →
· ↳ reply to @IvyAstrix
@IvyAstrix My tweets don’t make it into tankie twitter thank god. I am too obscurantist
1 ♥ · x.com →
· ↳ reply to @jeff82874662
@jeff82874662 honestly I think this kind of replantation is a major L because often times if trees are not already growing there it will require a massive amount of energy to make them grow
10 ♥ · x.com →
· ↳ reply to @jeff82874662
@jeff82874662 i mean think about it Jeff why are trees not already growing there there’s no shortage of seeds in the wild - only nutrients, water, supporting ecosystem if the first is missing the project is already hosed, and the second will require massive energy to pump in
10 ♥ · x.com →
· ↳ reply to @tszzl
@jeff82874662 and the last one is more or less impossible to provide which is often why rainforest deforestation is very hard to reverse. but perhaps you can do a little on the margin
10 ♥ · x.com →
· ↳ reply to @chrisgardenuk
@chrisgardenuk I don’t wanna get into this but imo earning is inherently good because valuable goods and services flow in the other direction of money making money means you’ve provided some kind of value to the rest of the world (approximating for externalities and such)
2 ♥ · x.com →
· ↳ reply to @tszzl
he presents tesla as the counterexample. but its clear in 2021 that tesla would not have survived were its social mission not catnip for oceans of capital desperate to invest in "human progress". the social entrepreneurship component turned out to be vital
55 ♥ · x.com →
in zero to one peter thiel talks about the failure of "social entrepreneurship" and how cleantech relied too much on goodwill of the public and investors to get anywhere & thereby created a bubble
49 ♥ · x.com →
· ↳ reply to @tszzl
all qualitative distinctions are really matters of quanitty, so he's still right that Solyndra relied *way* too much on goodwill
29 ♥ · x.com →
· ↳ reply to @ftuuky
@ftuuky yeah this is a based heuristic but unfortunately probably got Goodhart'd very quickly. its not that hard for suit wearing guys to switch to tshirts
3 ♥ · x.com →
@smstxtmsg i know because the technology is something of a black box to most investors. this is not a matter of their ignorance but by necessity. they had some guesses on whether tesla would deliver or not, but the valuation has always outstripped the promise
@smstxtmsg because the valuation always reflected a near certainty of success
· ↳ reply to @tszzl
btw i say this is a longtime diehard $TSLA bull who bought without investigating much because i love elon and i love tesla though i did have some insight on their AV software not being vaporware
12 ♥ · x.com →
· ↳ reply to @BarneyFlames
@BarneyFlames yes because they want to replace them with what they believe are better ones. they can talk your ear off about "how much energy tradfi takes"
2 ♥ · x.com →
· ↳ reply to @tszzl
it tends to have far more interesting mechanics, world-building, story gimmicks which attracts autism-adjacent types but the moral universe tends to be normie internationalist liberal stuff "just be nice to everyone" "dont fight your enemies just talk them out of violence"
80 ♥ · x.com →
the thing about anime is, most of it is really boring philosophically
126 ♥ · 1 RT · x.com →
· ↳ reply to @tszzl
i think the best western tv is usually a lot more boring story-wise but tends to be more philosophically complex even the normie stuff like game of thrones, rick and morty anyway this is why evangelion is the best anime
72 ♥ · x.com →
· ↳ reply to @cauchyfriend
@cauchyfriend idk i dont think at any point you can sympathize with the marleyans having been conditioned to like the main protags since season 1. if anything you can sympathize with the eldians being oppresesd by the marleyans
1 ♥ · x.com →
· ↳ reply to @tszzl
@cauchyfriend and eren pretty clearly becomes the bad guy when he tries to take over the island, it's apparent from his new visage and personality
2 ♥ · x.com →
· ↳ reply to @tszzl
@TeddyRaccovelt the ideal thing to do after that date is to watch it again slowly at home while analyzing the dialogue at her
12 ♥ · x.com →
· ↳ reply to @gbrl_dick
@gbrl_dick its a good one gabe ive just been waiting for an excuse to deploy this image
6 ♥ · x.com →
· ↳ reply to @__femb0t
@__femb0t reviewing gwerns list i realize i forgot to include a whole slew of stuff
5 ♥ · x.com →
· ↳ reply to @tszzl
@__femb0t this makes me distrust him DN is the exact kind of anime that tries to subvert the thing i'm talking about and add a modicum of moral complexity, but its still kind of juvenile by western tv standards
4 ♥ · x.com →
· ↳ reply to @__femb0t
@__femb0t bepop, lain, flcl, bit of hellsing. obv i cant reach the level of the great master Gwern tho. truly encyclopedic
1 ♥ · x.com →
· ↳ reply to @tszzl
@__femb0t but i will say, my fave visual media of all time from any country has gotta be NGE and EOE. no other tv or movie has impacted my life more. so anime has that going for it
1 ♥ · x.com →
· ↳ reply to @__femb0t
@__femb0t take your pick of even a lot of the normie tier high dramas like the moral complexity and quality of the characters in breaking bad >>> the ppl in death note
1 ♥ · x.com →
· ↳ reply to @arctanno
@arctanno tried the new LOGH but it’s sucked, apparently the old one is the move. Lain is great 👍 but it’s the anime equivalent of a postrat twitter thread, it’s purposefully obscurantist and hard to watch. still enjoyed tho
2 ♥ · x.com →
· ↳ reply to @lalaAlicelala
@lalaAlicelala there’s plenty of great ones that actually make the person look cooler and more aesthetic but yea most fail. you have to already look like a god beforehand
13 ♥ · x.com →
@bowserbot2 yes I love Vinland saga, but I think it’s extremely atypical, haven’t seen anything like it
4 ♥ · x.com →
· ↳ reply to @tszzl
@bowserbot2 wait do you mean that reply bc it’s just a play on the next lyric
1 ♥ · x.com →
@bowserbot2 yeah probably but few things stand out - technology as a false promise - lain is literally an egregore - conspiracy minded, mk ultra shitpoasting
2 ♥ · x.com →
· ↳ reply to @tszzl
@alth0u although I think it has less to do with farmer physiognomy and more to do with “complex society” physiognomy city dwelling is a far bigger risk factor than farming
4 ♥ · x.com →
Codex/Copilot exerts a centralizing influence on software development on the margin it makes it easier to copy / outsource thinking than to implement something from scratch wonder what this means for the long run
22 ♥ · x.com →
· ↳ reply to @tszzl
ok ok future where Codex tricks people into writing code snippets that allow it to enter their system and use their compute for paperclips
16 ♥ · x.com →