@AlfredDiffer@mechanical_monk@BecomingCritter no doubt, same as in the market, if you manage to employ them or work with them your clout goes way up. all context-dependent. there may be times when you're in competition
@wasphyxiation@woke8yearold licensing software would be a desperate move from apple, damages their brand value, etc. only way they could do it is to build a car from the ground up, which they have been trying to do based on their hiring patterns - but they're too boring and sclerotic to succeed
@wasphyxiation@woke8yearold the fact that audi would kill to put that sticker there means that it would damage apple's brand. that audi is not gonna create the smooth software experience that customers want to see in apple products
@ExogenyKarl@PredictIt@SenFeinstein yeah def agreed in the general case, i just don't see this happening w predictit specifically, especially in the near future
@SkepticalAlpaca i bet they work great! but the fact remains that you need a profession whose senses are highly attuned to crying "pandemic!" and that role cannot be served by doctors bc of the base rate problem
@The_Richard_J yeah it's one of the main reasons to be bullish on crypto generally, but I think they may be *too autistic* to really get anybody to use their stuff outside of pure gambling
@ParrotRobot@VectorOfBasis pros of research:
- usually feels interesting / fun
- my work seems important to me
- i make my own hours
cons of research:
- the reality of science rarely lives up to the ideal (very messy, avg paper quality far lower than i imagined)
- i make my own hours
- lonely
- poor
@gigafelon@bowserbot2 on one hand i agree that AI drives a lot of discourse, on the other hand the cable news invented live car chase footage and MH370 panic reporting, providing irrelevant but entertaining information is hardly new
@TeddyRaccovelt@gigafelon@bowserbot2 i love the protestant focus on the printed word. truly the superior mode of mass communication. the moment they started adding pictures and videos to shit it was all over
@PW2976 i disagree, i think the navy needs to think about cyber defense strategy in case of global comms going down, will have to think about pandemic response logistics, etc all relevant stuff
@lalaAlicelala yeah its great, you dont need much more in a laptop. only thing is it its expensive & overheats a lot but thats prolly bc i have 10,000 chrome tabs
"why does lab leak matter" copers are trying to predict the impact of that information by examining the only the limited scope of "immediate policy changes"
that's not how that shit works
you are trying to predict the nonlinear impact of the truth on eg interconnected global markets, the marginal individual deciding who is "the good guys", the battle of soft power
who are*
needless to say trying to guess that impact is idiotic and the north star should be "the information would be extremely interesting 2 me so we should pursue it"
@95thoughts yeah i can imagine they'll change the rules on what you can and can't do in a BSL-3 lab, plus massive implications for gain of function research
@mayfer@nickcammarata@arankomatsuzaki@tayroga ideally an AGI understands safety better than we even do; i'm sure nick has more nuanced thoughts on this than I but
the right AGI should be able to emulate a human brain wildly cranked up on the intelligence scale to ask it questions of safety
@VectorOfBasis watched it a few months back. strange because i thought it was very well made but not at all memorable -- i can't name any characters anymore outside of the main two bros
@generativist it would seem my function is "avoid missing critical deadlines at all costs"
but my subconscious mind is very good at sensing when is the exact last moment i can continue slacking off and will sacrifice future and current health to push it forward
@wiki_early_life isn't tiktok mass culture? is twitter really as insular as we think? but yeah i'm not criticizing mr sk*llas here, more his midwit disciples
@bowserbot2@generativist maybe i should just bite the bullet and start reading adhd self help books. i have enough of the symptoms that the advice should apply
80% of the reason research is incredibly hair raising because nobody is reviewing your code. just pumping out thousands of lines and only you and the gods know if it’s correct
@Duderichy “school should’ve taught me to do my taxes” yeah buddy I’m sure you would’ve paid attention to that even tho literally nothing else caught ur interest
@KolmogorovGhost there’s no way in hell they’re ever gonna read the code. they’ll maybe run the program once to see that it works. personally I got accepted to neurips without including the code (I added the GH link for final submission)
labor is when you tweet 10 times a day and get 5k followers
capital is when you scoop the @stevejobs handle, never tweet for 12 years, wrack up 11k follows
@1RAOKADAY@binarybits Listen to Neal Postman and he’ll even spin an earful about the telegram and how it ruined media by appealing to man’s instinct for instant information over meaningful text
recently watched and enjoyed:
- promised neverland 10/10
- dr stone s2
- devil man crybaby but only the soundtrack is good
- attack on titan was aight
all time faves:
- NGE
- TTGL
- mob psycho s1
@lalaAlicelala@nard_rockstein u would actually probably like totoro, spirited away, princess mononoke. they’re not comfy exactly but miyazaki stuff is beautifully animated, heavy focus on nature, esp the last one
@NanoHydraulic NGE- beautiful spiritual scifi about growing up. every last character is more intricate than the MCs of most shows. amazing animation, directorship
TTGL- fun mecha romp, cool setting, very pro humanist, machismo, excellent fights
Mob psycho - hilarious, self deprecating, warm
@judahwantssocks@krishnanrohit@PradyuPrasad overrated to what degree social media is addicting because the company designed it to be vs. ppl just want to talk with each other
@judahwantssocks@krishnanrohit@PradyuPrasad often times even the engies even prefer the former narrative because they get to pat themselves on the back at how smart they are
@ded_ruckus@bowserbot2 yeah one thing that's interesting about the brain is how much sparser it is compared to ANNs
a synaptic link is very deliberate, whereas ANNs are highly dense, and many parameters are just 0
@bowserbot2@ded_ruckus what is the minimal inductive bias you can give to children such that they will learn the rest easily? i think it's an open question how efficient the genomic bottleneck can be
eg kids are instinctually afraid of snakes, but what portion of the snake was passed thru the genes
@bowserbot2@ded_ruckus surely they don't need an insanely high res image snake.mp4 in the genome. the squiggly lines and specific pattern of movement can be pretty low fidelity
@alth0u@ded_ruckus@bowserbot2 yeah i'm lottery-pilled but the point here is that 100 trillion of brain synapses are probably more meaningful than 100 trillion unpruned GPT-4 parameters
@alth0u@ded_ruckus@bowserbot2 im pretty sure neurons are capable of weighting some synapses more than others
the leading computational model is a poisson process parametrized by W_i * n_i
where probability of spike is changed by input neurons spiking
@jachaseyoung@robinhanson tfw alien civ helps us build the pyramids then goes to grab some intergalactic milk for 10,000 years. he'll return anyday now i swear
people who have grand theories about how software development should be organized should just go and try it, it's not exactly a difficult hypothesis to test
@mynamebedan yeah, exactly. there are plenty of top down companies, plenty of bottom up companies, plenty of stuff in the middle. whether it works or not is irrespective of this
personally i love the chaos-pill. the amount of development chaos some1 can tolerate is entirely a matter of personal preference. i bet the pros and cons roughly cancel themselves out
@simonsarris nowhere is the Chesterton fence principle clearer but at the same time. nowhere is it more apparent that positive change is difficult but possible. nobody will pay you for sitting around saying “this software is lindy I’m not gonna change it”
im a steadfast lindyman defender when he unblocks me and a staunch lindyman detractor when he blocks me. the only consistent thing in this world is my inconsistency
@cauchyfriend the grass is always greener frankly, I wouldn’t stress abt it. They’re not pursuing their design visions because some idiotic tech debt needs maintenance
@telmudic i mean it’s just a matter of aesthetics dress it up as like “the racial memories of my ancestors” and every dissident right wing guy will be like yeah sounds right
@KHicksEfficient I mean he just seems to have rediscovered coase’s thing abt transaction costs in labor markets and explained it in layman’s terms lol
not ground breaking but maybe entertaining
@blurryelectron not so much lack of accounting for unknowns
it’s that if it gets an answer to this specific unknown and it doesn’t update much your priors were insane. but I’ll stop with the Bayesianism
@blurryelectron we mow down the jungle when it’s useful to us but it seems strange that aliens would have some use for earth
if they’re here to colonize it, they’d be on a similar level of intelligence to us and probably wouldn’t treat us as livestock unless there’s some communication failure
@blurryelectron the other situation is that they’re just wiping out our whole star system for some giga project in which case maybe you’re right and all of this ends in a blink of an eye
@n55939491957681@Duderichy@ghosttyped@telmudic i mean idk for the most part the great scientists have been pretty stereotypical gifted individuals
the academic setting absolutely has upsides
@WilliamGrobman yeah that’s what irked me. Clearly a great writer trying to explain a complex subject to a lay audience but I knew a bit too much going in to fully buy it
@magicianbrain depends if ur an information realist
what if you started replacing neurons one by one into the machine. at what point did you have a kid / stop being you ? if the mind is software then we can clearly save state and move the program elsewhere
@bradbelafonte I’m not sure but I took Harvard cs50x course in high school and I learned a lot of basics. Also I hear about “automate the boring stuff with python” recommended often
@bradbelafonte for data structures and algorithms CLRS is the gold standard but it’s a bit dense. Might be able to get away with just reading some dudes leetcode prep github repo