I'll paste my abridged response to the 'datacenters in space' post:
... It then occurred to me that they (all major AI companies) know all of these facts (uneconomical costs) but still pushing for it so there must be another reason. Then I recalled the offhand statement from the openAI lady about govt backstop for infra, which was strongly opposed by public and AI czar. this might be be a backdoor way of injecting that backstop capital in terms of subsidies now for results in 5 years or so. and needless to say after pilot programs those will fail spectacularly.
this does not makes sense from dollars pov to me, I ran a back of napkin session with claude & gemini on this and the short of it is you need a magical weightless radiator for cooling and even then it wont work because the launch costs need to be sub $100 before this can be feasible. this does not even factors the 5y amortization and LEO orbit drag correction.
It then occurred to me that they (all major AI companies) know all of these facts but still pushing for it so there must be another reason. Then I recalled the offhand statement from the openAI lady about govt backstop for infra, which was strongly opposed by public and AI czar. this might be be a backdoor way of injecting that backstop capital in terms of subsidies now for results in 5 years or so. and needless to say after pilot programs those will fail spectacularly.
As a non gaming engineer, I dont want a union, what I really want is solid (really, harsh) enforcement of basic common sense labor rules for exempt workers like weekly working hour limits, no after hours scheduling or minimum notice/severance period for layoffs & many other abuses. Problem is tech industry does not wants to give an inch and workers don't complain because of higher pay and 'lottery ticket' effect.
I fear the time for fixing this is passing fast. Its because within a decade AI will have enough of labor displacement that labor wont have any negotiating leverage against capital. If this happens with union, so be it.
OK, so if you don't want a union, but you want to achieve goals which are typically fought for by unions, what exactly is the alternative mechanism you propose for achieving these goals?
This sounds to me like "I don't want memory protection, what I really want is for my computer to solidly enforce common sense memory barriers", or "I don't want defense attorneys, what I really want is courts to provide common sense advocacy for defendants". Somebody has to do the work, and you're naming the component of the system which provides the features you want.
The universe maximizes entropy, not common sense. The tech industry doesn't want to give an inch for the same reason a dropped object falls to earth: there is no force suggesting it do anything else.
Did you miss the last sentence or had your mind made up already by then?
the reason I don't want a typical union is because they work out well where there is a clear separation of labor & capital. labor get paid wages & capital makes money off the invested assets. problem with tech is that in a tech ventures labor is a part owner and capital too (engineer equity). introducing a 3rd party (unions & union leadership) in these situations will likely have the worst effect on the capital portion of labor & possibly eroding the incentives altogether. Imagine having a set of union rules to follow when distributing equity to early stage engineers? how do you decide equitable when by definition you are on cutting edge of tech & innovation. & if you think that wont happen you are mistaken, that is where the biggest pie is and unions arent exactly 'shining beacon of honesty and integrity' either. Look all I want is just some basic labor laws are followed and tech employees dont feel the panic when they hear slack-chime on a weekend night.
there are other mechanisms like state level enforcement of labor laws and punitive fines. you know the good old law enforcement. its really not so hard if you actually want to do it.
So you walk in to negotiate with your capitalist overlords with an equivalent number of lawyers that they can field? Oh, they don't pay you enough to do that? Oh, you need to be working while doing this? Oh, you just got replaced because the default T&C's weren't good enough for you?
You are not a temporarily inconvenienced millionaire/billionaire.
> Did you miss the last sentence or had your mind made up already by then?
Is the first sentence of your comment not your thesis? Or am I to read this as you changing your own mind during this comment? I'm confused. You literally started your comment with "I don't want a union".
> they work out well where there is a clear separation of labor & capital
Citation needed. I've studied a bit of labor history, and this is the first I've heard of such a claim. Unions work poorly when they are corrupt -- the same as any democratic system. I'm not aware of other categorical failures of collective bargaining. Please enlighten me.
> Imagine having a set of union rules to follow when distributing equity to early stage engineers? how do you decide equitable when by definition you are on cutting edge of tech & innovation.
I'm not sure what problem you think the union would create here. Equity is simply another form of compensation, and unions have dealt with non-salary compensation for decades (licensing, branding, likeness, etc). It's really not a problem.
If there are union rules that must be followed, then that would be your answer: you follow those rules. Isn't that strictly better than not having any rules, and simply getting whatever the CEOs/VCs give you? Historically, workers do much better in a union than without.
Union rules are created and voted on by its members. If the workers don't want rules regarding equity, they won't create any. If they do, then they would be worse off without a union to help enforce them.
> unions arent exactly 'shining beacon of honesty and integrity' either
They're democratic representation for workers, no more, no less. (Every time there's an election result someone doesn't like, I see "Ha, democracy isn't perfect, either!" comments from other countries. Yeah, zing, you got us.) Unions have proven to be a shining beacon compared to billionaire CEOs, for sure.
> Look all I want is just some basic labor laws are followed [...] you know the good old law enforcement. its really not so hard if you actually want to do it.
You're using generic "you" here as a linguistic trick. Workers want labor laws followed, but they're not the same people working in "good old law enforcement". Unions are the mechanism by which workers are effective at getting labor laws followed, and new laws passed.
You still have not proposed any alternative mechanism besides suggesting that people who have no incentive to do a thing should start doing a thing. You claim it's not hard. That may be the single hardest thing to do in this world.
The challenge with having good worker protection laws and high minimum wage is that at some point local labour becomes less competitive with foreign low cost options. Then you'll need tariffs or other import rules to keep your local industries alive.
Because it happend much longer ago. If the US wanted they could leapfrog that step and go straight to federal laws covering the same things. It's just that there is no majority that wants it.
to me bigger problem emerges if you assume all those companies in the 'circular financing infographic' as one conglomerate. then essentially all you are left with is real demand for AI & I just dont really see much of it besides Hype and FOMO. that falls away the moment CFOs get 'permission' to ignore the AI growth story. besides that I only see coding bot and openAI's consumer subscription business. I dont see that becoming $1T business anytime soon. so what gives? I think Burry is right but I am not sure of timing because they just need one funder to extend and pretend for a few more quarters & DJT can do it under guise of datacenter jobs etc.
Circular thing is bad too but from a different angle, Imagine if the whole TPU vs GPU thing erodes Nvidia's moat and its profit margins compress. if that happens how long it can keep feeding the same unproductive 'pets.AI' type startups? one break in the narrative and tragedy of commons strikes. will it happen soon? anybody's guess but given Trump is at helm and there is going to be new Fed chief, I doubt it would be anywhere near soon. Definitely not before mid-terms are locked in.
if you find that fascinating then you'll be blown away by something called 'Wolfarm physics project'. it basically is trying to recreate entire physics using such baseline 'graph update' rules like 'Game of Life'. So far no predictions yet but very interesting.
Wolfram is kind of obsessed with cellular automata, even went and wrote a whole book about them titled "A New Kind of Science". The reception to it was a bit mixed. CA are Turing-complete, so yeah, you can compute anything with them, I'm just not sure that in itself leads to any greater Revealed Truths. Does make for some fun visualizations though.
A new kind of science is one of my favorite books, I read the entirety of the book during a dreadful vacation when I was 19 or 20 on an iPod touch.
It goes much beyond just cellular automata, the thousand pages or so all seem to drive down the same few points:
- "I, Stephen Wolfram, am an unprecedented genius" (not my favorite part of the book)
- Simple rules lead to complexity when iterated upon
- The invention of field of computation is as big and important of an invention as the field of mathematics
The last one is less explicit, but it's what I took away from it. Computation is of course part of mathematics, but it is a kind of "live" mathematics. Executable mathematics.
Super cool book and absolutely worth reading if you're into this kind of thing.
I would give the same review, without seeing any of this as a positive. NKS was bloviating, grandiose, repetitive, and shallow. The fact that Wolfram himself didn’t show that CA were Turing complete when most theoretical computer scientists would say “it’s obvious, and not that interesting” kinda disproves his whole point about him being an under appreciated genius. Shrug.
That CA in general were Turing complete is 'obvious'. What was novel is that Wolfram's employee proved something like Turing completeness for a 1d CA with two states and only three cells total in the neighbourhood.
I say something-like-Turing completeness, because it requires a very specially prepared tape to work that makes it a bit borderline. (But please look it up properly, this is all from memory.)
Having said all that, the result is a nice optimisation / upper bound on how little you need in terms of CA to get Turing completeness, but I agree that philosophically nothing much changes compared to having to use a slightly more complicated CA to get to Turing completeness.
The question really ultimately resolves to whether the universe can be quantized at all levels or whether it is analog. If it is quantized I demand my 5 minutes with god, because I would see that as proof of all of this being a simulation. My lack of belief in such a being makes me hope that it is analog.
Computation does not necessarily need to be quantized and discrete; there are fully continuous models of computation, like ODEs or continuous cellular automata.
That's true, but we already know that a bunch of stuff about the universe is quantized. The question is whether or not that holds true for everything or rather not. And all 'fully continuous models of computation' in the end rely on a representation that is a quantized approximation of an ideal. In other words: any practical implementation of such a model that does not end up being a noise generator or an oscillator and that can be used for reliable computation is - as far as I know - based on some quantized model, and then there are still the cells themselves (arguably quanta) and their location (usually on a grid, but you could use a continuous representation for that as well). Now, 23 or 52 bits (depending on the size of the float representation you use for the 'continuous' values) is a lot, but it is not actually continuous. That's an analog concept and you can't really implement that concept with a fidelity high enough on a digital computer.
You could do it on an analog computer but then you'd be into the noise very quickly.
In theory you can, but in practice this is super hard to do.
If your underlying system is linear and stable, you can pick any arbitrary precision you are interested in and compute all future behaviour to that precision on a digital computer.
Btw, quantum mechanics is both linear and stable--and even deterministic. Admittedly it's a bit of a mystery how the observed chaotic nature of eg Newtonian billard balls emerges from quantum mechanics.
'Stable' in this case means that small perturbations in the input only lead to small perturbations in the output. You can insert your favourite epsilon-delta formalisation of that concept, if you wish.
To get back to the meat of your comment:
You can simulate such a stable system 'lazily'. Ie you simulate it with any given fixed precision at first, and (only) when someone zooms in to have a closer look at a specific part, you increase the precision of the numbers in your simulation. (Thanks to the finite speed of light, you might even get away with only re-simulating that part of your system with higher fidelity. But I'm not quite sure.)
Remember those fractal explorers like Fractint that used to be all the rage: they were digital at heart---obviously---but you could zoom in arbitrarily as if they had infinite continuous precision.
Sure, but that 'If' isn't true for all but the simplest analog systems. Non-linearities are present in the most unexpected places and just about every system can be made to oscillate.
That's the whole reason digital won out: not because we can't make analog computers but because it is impossible to make analog computers beyond a certain level of complexity if you want deterministic behavior. Of course with LLMs we're throwing all of that gain overboard again but the basic premise still holds: if you don't quantize you drown in an accumulation of noise.
> Sure, but that 'If' isn't true for all but the simplest analog systems.
Quantum mechanics is linear and stable. Quantum mechanics is behind all systems (analog or otherwise), unless they become big enough that gravity becomes important.
> That's the whole reason digital won out: not because we can't make analog computers but because it is impossible to make analog computers beyond a certain level of complexity if you want deterministic behavior.
It's more to do with precision: analog computers have tolerances. It's easier and cheaper to get to high precision with digital computers. Digital computers are also much easier to make programmable. And in the case of analog vs digital electronic computers: digital uses less energy than analog.
For all we know, it could be distinct layers all the way down to infinity. Each time you peel one, something completely different comes up. Never truly knowable. The universe has thrown more than a few hints that our obsession with precision and certainty could be seen cosmically as "silly".
In our current algorithmic-obsessed era, this is reminiscent of procedural generation (but down/up the scale of complexity, not "one man's sky" style of PG).
However, we also have a long track record of seeing the world as nails for our latest hammer. The idea of an algorithm, or even computation in general, could be in reality conceptually closer to "pointy stone tool" than "ultimate substrate".
> For all we know, it could be distinct layers all the way down to infinity. Each time you peel one, something completely different comes up. Never truly knowable. The universe has thrown more than a few hints that our obsession with precision and certainty could be seen cosmically as "silly".
That's a tempting thing to say, but quantum mechanics suggests that we don't have infinite layers at the bottom. Most thermodynamic arguments combined with quantum mechanics. See eg also https://en.wikipedia.org/wiki/Bekenstein_bound about the amount of information that can even in theory be contained in a specific volume of space time.
makes me wonder if its possible to get natural numbers like pi/e using a geometric structure in GoL. it would be interesting to derive them from an emerging order based on fixed set of automata rules. If possible it might lead credence to simulated universe hypothesis.
2) there are algorithms that calculate digits of Pi or e.
so... yes?
but if I just took any old Pi-digits algorithm and encoded it on GoL, its appearance would not be meaningful or "elegant" to our senses. You're probably asking "what does the shortest/most elegant program to calculate Pi in GoL look like, and does it maybe have some unexpected relation to other mathematical terms like, I dunno, Euler's identity or... Mandelbrot set?" And then you would probably need to answer the question "Well, how would you like the digits encoded and represented?".
All of a sudden your question becomes a bit ambiguous. Or did I misunderstand what you meant?
I mean.... I think I feel what you're asking, like... is there some primal version of Pi that can be encoded in GoL initial condition with as few bits as possible but I'm afraid that the answer is something like "well, that depends on what you mean by [...]"
yes you are right the question was ambiguous. WHat I was really looking for is if there is a way to physically derive PI out of some basic cellular automata operations. that does not necessarily have to involve circles etc. much less so about representing digits of PI itself. just that an pattern that evolves into ever more accurate values of pi (or a quantity derived from pi). this definitly must not be encoded or somehow preprogrammed into the GoL's initial condition. In fact I dont even care if the evolution rules of standard GoL are followed. just that SOME set of rules automatically produce PI-derivative.
the reason it would be cool is because then we've taken a most fundamental geometric constant and derived it from purely graph update type of mechanics. if you could do that then likely you can do a whole set of other things like physical laws.
reply