“Meta spent almost as much as the Manhattan Project on GPUs in today's dollars”
twitter.com
twitter.com
It’s bizarre we put people on the Moon in the ‘60s for an amount of money similar to Apple’s recent stock buybacks.
Please don't forget to close such statements with an /s, lest some notorious software "engineers" on HN might be led to actually believe engineering the most powerful vehicle on earth mostly by hand was a negligible task.
But these days, we have established infrastructure, to manage a lot of the things that the Apollo guys needed to construct from whole cloth.
Also, the way that SpaceX does research is highly effective (and inflammable). Failing fast works, as long as we are set up to learn the lessons, and avoid doing things like sticking people on top of unverified tech.
There was an interview, recently with either Ben Nelson, or his predecessor, where they mentioned that NASA (or its directly-managed contractors) would never have been allowed to blow up a whole bunch of boosters, just to finally arrive at a design that normalizes what used to be considered impossible. On the first crash, he would have been hauled up in front of a Senate investigation committee, and forced to explain why it would never happen again.
It is with some restraint that I note that "democracy" per se does a piss-poor job of preventing this, when Tim Cook's net worth exceeds the yearly GDP of the Seychelles. This is particularly true in a world where capital controls information channels, and therefore public opinion. To really prevent capitalists from capitalizing on their capital to amass ever more capital, you also need a social infrastructure that redistributes wealth away from powerful individuals back to society, so that social issues can be addressed.
Yeah, I sure hope that would be true otherwise Apple has some explaining to do.
I'm not sure if wealth is any better distributed in autocratic countries, think Putin's Russia and its oligarchs.
Democracy is a continuum, any country can evolve into either direction. It's not like this country is democratic and that is not. Countries have more or less of it.
A fairer way to look at it is how many years of the average citizens salary is it.
The manhattan project was $2B in 1944, or 121,000 median-family-years of work using salary figures from the 1940 census.
Apples stock buyback was $110B in 2024, or 122,910 median-family-years of work using figures from the 2022 census.
So, the inflation figures under represent the cost by a factor of 3.5 over those 80 years.
So it isn’t that inflation is underestimated. The average family truly is richer. Far more house. Far more car. Far more and better food.
(Also, IMO, houses today are much higher quality than in the past as soon as you account for safety)
https://www.newser.com/story/225645/average-size-of-us-homes...
Material and building science has advanced a mind boggling tremendous amount - even modern toilets are significantly better than an actual un-upgraded 40’s - if you can find it.
At the same time we can mass manufacture endless cheap plastic shit.
You say it like that's a bad thing.
You only need to worry about perhaps the average of the things you buy.
It is a bit annoying to have to search and find high-quality items, but it's still doable.
Even better is pointing out a perfectly valid concern regarding higher search costs, and then brushing that aside altogether simply because it doesn’t completely prevent search.
1. Environmental damage
2. Higher search costs
3. Higher transaction costs
4. Opportunity cost
The USA in 1940 were a smaller empire than they are now so, I might be wrong, but the Manhattan project money came mostly from inside the country.
This means that it's easier for Apple to find and spend those money than it was for the USA in 1940.
Similarly there where a great deal of households in the 1940’a with two working adults.
https://www2.census.gov/ces/wp/2019/CES-WP-19-19.pdf
I'm guessing the rise of dual income houses is much more visible in the median than are retirees dropping down the labor force participation rate.
Married-couple households however have dropped dramatically over time, 76% of households in 1940, 55% of all U.S. households in 1990, and 46% in 2020. “In 1940, married couples with children represented 43 percent of all households; married couples without children represented 33 percent of households”
I cannot find stats going back, but a majority of current households today are dual income. https://www.bls.gov/opub/mlr/2020/article/comparing-characte...
This effect of the median family being dual income is likely true for now, and false for 1940. It is much stronger if you subset down to "families", but this nuance doesn't likely appear in the "median household income".
In the 1940’s multi generational households with multiple income earners were common. In 1940 only 7.8% of all households had a single person vs 26.7% by 2010. Working adult children would frequently stay at home until marriage and sometimes past that point.
This is one of those things you need actual data based on similar definitions to compare.
It's utterly ridiculous how much of the productivity increases of the last decades ended up not distributed to the workers either in terms of wage/salary hikes or reduced working hours. Something like billionaires rarely existed for a long time, usually it was the royal family and that's it - and now, alone in the last 20 years the amount of billionaires is 6x higher than at the start of the millennium. All of that wealth has been looted from the lower classes and in quite a few cases the taxpayers.
[1] https://de.statista.com/statistik/daten/studie/220002/umfrag...
Most people don't actually seem to want to have mist most of their waking hours free to do whatever they want.
Personally I work part time on software projects and have a small farm/homestead. I usually don't know what day of the week it is and have the freedom to take each morning yo decide how I want to spend my day. I wish everyone could experience that freedom, even if they find they don't like it.
This gives the price of the Manhattan Project as $131 billion in 2024 dollars.
What? We are living in a golden age of unprecedented peace and prosperity. By and large, people never had it so good.
The world at large has other ideas, especially continental Asia and the Middle East where people seemingly can't stop brutally killing each other.
PR China has also seen enormous advances in prosperity, and hasn't been in a major war in 70 years, either. In India obesity is a bigger problem than starvation. And approximately everyone can afford a smartphone, too.
South East Asia is doing fairly well, too, compared to historical averages. Even Sub Saharan Africa is finally starting to catch up.
Yes, there are still wars, alas. But by and large people are richer and living more peaceful lives.
We could look at statistics, if you wish. (I was looking for a graph of casualties of war over time for my previous comment, but couldn't find one that was recent enough, sorry.)
The founding Prime Minister of my own adopted home of Singapore wrote a book called 'From Third World to First'. And he's right.
During the Balkan wars about 5% of Europe's population lived in a country that was at war.
Yet that time period was relatively peaceful compared to the first half of the 20th century where Europe went through two total conflicts.
Yet for me the war in Yugoslavia was of personal significance and I felt it and its longer term effects closely.
But it wouldn't be fair for me to just dismiss that the second half of the 20th century was a time of peace and prosperity in Europe, because a part of it was burning in flames and committing atrocities to who were their brothers 5 minutes before.
The precariousness of peace is what makes it so valuable.
Yes. There are conflicts all around the world. Wars. Oppression. Injustice. Racism.
I don't think that recognizing that we're living unprecedentedly peaceful and just times is going to make those conflicts and injustices worse.
Quite the contrary. I feel that never recognizing the successes and progress that humanity has made in many of those fronts is what will eventually doom us to regress.
Like if there's a global war I bet people are going to want canned food much more than a gold chain. So should we use the price of chicken soup in 1944 vs 2023?
---
But yeah I also agree with eru; being "close to yet another global war" is a way better place to be than "actually in a global war". Considering pretty much every century has them.
Just to nerd out a bit: You can dissolve gold in some particularly strong acids, but it doesn't really diminish the value of the gold. The effort for getting it back into elemental form is small relative to the price it fetches.
Compare this to eg aluminium, or to make an even starker contrast: wood.
Historically, gold has been a decent enough base money. But people soon make up notes to stand in for that base money, and those notes (and fiduciary coins!) win in the market. (To explain, fiduciary coins are those that like notes promise to pay the bearer. By themselves, they are not made of valuable material.)
If you don't move your base money around much, and just let it sit in vaults, then the chemical qualities we talked about don't matter all that much.
See eg https://www.cato.org/policy-analysis/rise-fall-gold-standard... for some background.
Cu and Ag are far more prone to oxidation.
Its thanks to the relativistic effects that gold is so much more oxidation resistant. But the other adjacent metals are too hard and or too rare. Eg. Pt Ir Pd Rh etc
That coincidence of lustre (which is both pleasing but also a rough way for us to ascertain purity/quality before we had better tools), malleability, oxidation resistance and ease of purification (gold has a relatively low melting temp) all made it a fantastic candidate for storing wealth.
As someone who studied the periodic table I find it very fun to then appreciate how the properties of elements played their role in civilisation
Meanwhile things like diamonds are hard / impossible to subdivide and standardise so things like diamond coins or tokens are impractical
Nobody wants to see the US dollar collapse overnight. Neither the US nor China (which holds trillions of US bonds). But everyone wants to move from that standard including the US (which, being the reserve currency, has suffered a horrible NIIP since the end of the gold standard[1]).
1: https://en.wikipedia.org/wiki/Net_international_investment_p...
As to the claim that "fiat currencies are fleeting," it is a fact every country of 200+ has one, so it's safer to say every non-fiat currency has failed.
There's also significant empirical evidence such economies are on the whole less volatile than older ones limited by how fast one can dig material out of a hole. See a list of severity and frequency of US recessions, or study the Great Moderation, or simply look at the volatility of gold prices, to see how unstable non-fiat pricing is.
Pretty much all the important things in life have gone up more than stated inflation.
I don't care that I can get a cheap very environmentally unsustainable cardigan for less than ever before. Likewise for a watery nutrition-less tomato.
Meanwhile education, housing and healthcare are the least affordable they've ever been.
If you want to understand cost of living, just look at your personal bookkeeping. It will paint the most accurate picture. There isn't much to be gained in reporting cost of living in the news – unless you think aliens on some far away planet will take interest a thousand years from now when the signal arrives at their location.
Housing might be somewhat more valuable than it used to be, but the extreme valuation swings we've seen in the last 20 years were due to monetary policy and macroeconomic conditions, and not actual demand. Or I should say, demand for housing is always proportional to population. It's not like when people can't afford a house, that demand disappears. It just can't be acted on.
It is not. A CPI measures the change in cost of a fixed basket of goods. A COLI measures a change in cost of a fixed level of "well-being".
> It's not like when people can't afford a house, that demand disappears.
It is like that. That's literally what defines demand. Perhaps you have confused "demand" with "desire"? They do share the same first couple of letters.
Like I swear to god I see variants of this thread almost every other day where someone is like 'i think that this economic thing is broken' and someone else comes along and is like 'no,no, you're just holding it wrong -- The thing you want to use to do this is...'.
And despite witnessing this kind of encounter regularly I'm no more informed about economics. I've taken two first year econ courses and I still don't feel informed about it.
Why is it that all too often economics discussions get further and further away from some objective measurement of reality and turn into semantics arguments over what obscurely named things like M1 and M2 or CPI actually even mean to me anyway?
Is there like an authoritative source that defines all of these things and has a recent dataset of verified data to work work?
Like what's the lib-economy or python package for economic stuff?
The basics aren't hard. But also not particularly useful outside of the work of economists. That is no doubt the problem you keep seeing: 'i think that this economic thing is broken [because I cannot find any applicability in my daily life]'.
Economics can become hard as it is ultimately a study of human (possibly animal) behaviour, and these creatures seem erratic to our limited understanding.
Read any modern and popular university textbook, and good data can be found at FRED, BLS, Census, CBO, despite the usual conspiracy claims.
> Why is it that all too often economics discussions get further and further away from some objective measurement of reality
Like physics, or biology, or math, economics is not getting further from the object of study - terminology needs defined precisely to prevent people from making wrong conclusions from misunderstanding or bad reasoning. Learning not only terminology, but the problems the terminology was created to avoid, takes time and study. But due to the complexity in each field this is needed. One cannot get proper conclusions using sloppy hand waving or touchy-feely ideas with little solid foundation.
collecting data about atoms from spectrometers just doesn't have the same subjectivity and political interests as various economic data and definitions do.
As an outsider there is so much orthodoxy to economics it feels more like the church of pythagoras and we look to burn the Hippasus of our times. In physics MOND seems more entertained than some of the people who dare to question economics...
I find nothing in this page is about political definitions. Politicians like to conflate and mix things up since laymen have all sorts of ideas about economics, mostly because they understand so little of it.
What economic definitions are political?
BTW, I've done grad work in math, physics, and computer science, have degrees in all three, PhD math, and have taught graduate level mathematical economics. I went to grad school to do superstring theory, ended up doing algebraic geometry, but have still done all three professionally since (with publications in each). So I do have some decent understanding of hard science and econ. I find econ to be vastly closer to hard science than the pop beliefs and the ones above.
There's a reason phd econ programs recruit math majors, not econ majors. It takes some high level math aptitude and tooling these days.
> As an outsider there is so much orthodoxy to economics
Then read more. Pretty much everything defensible under the sun is open season to analyze in econ, overlapping lots of fields to address problems. I think no modern academic discipline is very orthodox when you get into it, because everyone is trying to make a name for themselves, and you don't get that without breaking current ideas. Econ is definitley a field where everyone is trying to break anything previous, which is also how it got good (again, despite pop nonsense).
If you measure physics by how well the weather is predicted for a month ahead, you'll claim physics is crap. Similarly, if you follow pop nonsense about "econ" you will miss the power of actual economics, which, like physics, will take you a decade of reading and improving to get to a deep understanding.
I have no doubt that there is a lot of hardcore maths in economics - I remember learning Koopmans Theorem only to discover he moved over to economics.
Just as there is a lot of deep and niche physics which requires a lot of abstract theory (string theory maybe?) we also have benefited from a lot of applied physics - hopefully I don't need to list all the ways mankind has benefited from applied physics.
Applied economics impacts billions of lives, arguably all of our lives. To the casual outsider, what are some modern economic practices that have fallen out from the maths that is driving the needle today? All the more given your point that despite all our advances in maths computing and physics, we still can't predict the weather.
Pretty much every single one of them actually. And that should be no surprise, no academic field is free from its social background, and economics being the heir of what used to be political philosophy, there's really no surprise that it's affected a lot by it.
You want concrete example: “GDP” it used to count only legal economy, now in most OECD countries it also counts stuff like drug dealing and other clandestine activities, but it does not count lots on activities, like community work or even housework. How is the limit drawn? It's entirely arbitrary is actually open to political debate.
The fact that we use GDP as an indicator at all, is also political: we could use Net Domestic Product, which takes depreciation of capital into account (that would be useful when it comes to ageing infrastructures), or even use other tools.
“Unemployement” it is not, in fact, the number of people that aren't working, it's the number of people who are actively looking for a job. And as such, during the 2010-2020 era, the US had low unemployment, despite having less working people than comparable countries, because many people weren't actually included in statistics (“Disguised Unemployment”). Students are considered “inactive” by the way.
In fact, you can take the entire economic lexicon and do the same exercise: all the definition here are arbitrary and the way they have been defined is subject to political debate. And for sociological reason in the field, most of them has been defined from a conservative mindset. Likewise, sociology has been mostly shaped by leftist views, and the definitions are a reflection of that. That doesn't mean economy or sociology are useless, but ignoring this bias is very problematic.
> There's a reason phd econ programs recruit math majors, not econ majors. It takes some high level math aptitude and tooling these days.
The reason why they do so is that Math is not science, so they go well around each other.
Don't get me wrong, I've a Math degree and I love it, but it's not science: assuming your reasoning is logically correct, then it's true, you don't have to confront it to the real world to see if it matches and if it's actually descriptive of real phenomenon. And that's the problem of mathematical economy: people do reasoning in a vacuum, it is “true” is the mathematical sense of things, but its descriptive power is close to zero.
In that regard, my favorite mathematical economics paper that I've read this year is one[1] that mathematically proves that “agents are infinitely-lived” is a poor hypothesis (among others). This isn't about studying the economy at all, it's just proving random theorems and it has no more scientific value than the theorems software developers unknowingly prove every day thanks to Curry–Howard.
Actual economics looks like this[2], but it's actually much more work to get a single paper, and the selective pressure is really going in disfavor of such research, and in favor of cheap intellectual masturbation.
[1] https://sci-hub.se/https://www.sciencedirect.com/science/art...
[2] https://annas-archive.org/scidb/10.1162/003355397555352?scid...
>It is like that. That's literally what defines demand.
It is not. You have confused the essence of demand with the actual capability to pay. We could get nuanced about it and say that people who can't pay are not market participants, but they could be. If I can afford to pay $X for a house and the price is higher than that, my inability to buy does not mean my demand went anywhere.
>Perhaps you have confused "demand" with "desire"? They do share the same first couple of letters.
Perhaps you have confused yourself with the kind of person who can make this kind of bold statement without getting laughed at. Go sniff your own farts some more and leave me alone.
By looking for consumables that people pay for regularly. Coincidentally, the things people pay for regularly are generally the things they need to live. But that doesn't make it a cost of living index. It is literally a consumer price index. That's what CPI stands for.
> Many categories are included in CPI, such as housing.
Indeed. The consumable portion of housing (rent, interest payments, etc.) is something most people pay for regularly, so it is ripe for inclusion. On the flip side, ignoring that a house isn't normally considered a consumable, most people will buy approximately one in their lifetime so it wouldn't be a great comparator for seeing how an individual perceives a change in value over time.
> it is intended to measure the price increases that consumers experience.
Yes, exactly, a general increase across a wide variety of goods and services comes as the result of a decline in value of the currency. There is no way to determine currency value in a vacuum, but it can be inferred this way. Which, as it happens, is what CPI gives us. Of course, there are other ways to measure inflation. Someone doing economics work will not use just a single measure of inflation as there are pros and cons to different inference methods, but CPI is the method we have settled on for the "official" rate.
> We could get nuanced about it and say that people who can't pay are not market participants, but they could be.
Demand is defined by desire and willingness. It is impossible for someone who cannot pay to have willingness, but anyone could gain the willingness at some point in the future, sure, and should that happen their demand would return. It's not a state forever set in stone. It's a dynamic state where people can come and go as conditions change. But as far as any single point in time goes...
When property values blow up 50% in a year or two and people pay $20k+ over sticker for cars, it's not "supply chain issues" nor 8% inflation dude. I have looked at old ads for stuff, and the difference is that now the same goods are manufactured for 1/10th the cost overseas and somehow cost way more anyway or at best roughly the same price as before. The BLS has a very complex scheme for understating inflation that involves substitutions, hypothetical costs, and other gimmicks. It wasn't always so fishy but the change was made to make the economy look better. I could argue with you at length about details but I can tell it will not be productive.
>As to the claim that "fiat currencies are fleeting," it is a fact every country of 200+ has one, so it's safer to say every non-fiat currency has failed.
Gold and silver are valuable everywhere and have been since they were first discovered many thousands of years ago. So your claim is utterly false. Every fiat currency that has existed has gone to zero, as well as any that were based on claims of redeemability.
>See a list of severity and frequency of US recessions, or study the Great Moderation, or simply look at the volatility of gold prices, to see how unstable non-fiat pricing is.
Gold prices are only volatile in terms of dollars, due to speculation in the futures market. The banks and governments play games with gold prices to discourage investors from seeking it. There are some cyclical or geopolitical factors in it too. But as JP Morgan said, "Gold is money. Everything else is credit." It can be manipulated but not nearly as much as fiat.
If you believe this is the general reason, then you should invest in all these companies making 80+% profits and get the amazing returns.
Oh yeah, those companies don't exit, maybe because the claimed reason doesn't exist.
> Gold prices are only volatile in terms of dollars, due to speculation in the futures market
That's nonsense. Gold has always been far more volatile than goods, which is precisely the reason the world saw the incontrovertible evidence during the Great Depression, which is why all of the 200+ countries on the planet left it as a standard. If it were some magical better way, then surely at least one of the 200+ countries would do it and win, but it has been clear nearly a century what a bad idea it is.
> Every fiat currency that has existed has gone to zero,
Well, there's 200ish of them that exist right now above zero. And guess what? Every single non-fiat currency has failed, since there are zero left. You can keep claiming otherwise but this is reality.
> Gold and silver are valuable everywhere and have been since they were first discovered many thousands of years ago.
So are cattle and a zillion other old goods. That does not make them a good currency or a basis for an economy.
Tying your economy growth to the rate you can dig up gold is also a reason that those thousands of years saw economic growth at a fraction of a percent. When mankind realized the stupidity of limiting growth in all areas to the growth in digging up gold, and detached the two, economic growth greatly increased. You cannot invest in new production, inventions, and goods, without having capital. Limiting capital, i.e., gold supply, results in lower growth.
I didn't say they were getting great returns. Although their returns are good (and if I could go back in time and invest like 20 years ago I could 100x my money), the fact is that although stuff is cheaper we (individuals and businesses) still can't buy more because money is worth less than it used to be.
>Well, there's 200ish of them that exist right now above zero. And guess what? Every single non-fiat currency has failed, since there are zero left. You can keep claiming otherwise but this is reality.
You can keep denying the reality that gold has much value, but you're wrong. It may not be the official currency anywhere, but it is worth a lot everywhere in the world. Furthermore, central banks hoard gold. If it was as worthless and detached from money as you propose, they would sell it all immediately. But they're buying instead. BRICS countries are launching a gold-backed currency any time now. Get your head out of the sand.
>Gold has always been far more volatile than goods, which is precisely the reason the world saw the incontrovertible evidence during the Great Depression, which is why all of the 200+ countries on the planet left it as a standard. If it were some magical better way, then surely at least one of the 200+ countries would do it and win, but it has been clear nearly a century what a bad idea it is.
Every country wants fiat because they can print it, plain and simple. The way these governments win is not by having a functional currency so much as by having something they can manipulate in their favor. Gold is not easy to manipulate because you can't print it. The avoidance of gold-backed money has nothing to do with "volatility" of gold. The supply of gold is steady, growing at perhaps 2% per year. There can be localized shortages of gold in a gold-based system that create volatility, but that is nothing compared to reckless printing of fiat. A responsibly managed fiat system could potentially be more stable than gold, but artificial manipulations of the money supply are not free and represent unfair enrichment or impoverishment of different players in the economy.
>Tying your economy growth to the rate you can dig up gold is also a reason that those thousands of years saw economic growth at a fraction of a percent. When mankind realized the stupidity of limiting growth in all areas to the growth in digging up gold, and detached the two, economic growth greatly increased. You cannot invest in new production, inventions, and goods, without having capital. Limiting capital, i.e., gold supply, results in lower growth.
You have really muddled things here. For thousands of years, debasement of currency and fiat currencies were also real threats. What really stimulated progress in the last 200 years was the discovery of steam power and oil. The Industrial Revolution. You can certainly invest in new innovations with a gold-based system, and that is indeed what happened. The Great Depression was caused by nothing other than reckless expansion of credit. Some people even think it was deliberately planned so that gold could be stolen by the government in the aftermath. I'm not saying gold is perfect or fiat absolutely can't work, but gold is real and it is the one thing that has truly lasted. Fiat currencies go to zero. Gold has not and will never go to zero, unless a solid gold asteroid hits the planet and we end up with an explosion in supply.
> I didn't say they were getting great returns
Ah, the old voodoo defense of the missing money.
> You can keep denying the reality that gold has much value, but you're wrong
I never said it didn't have value. It's bad as a currency. It's volatile and it hamstrings economies. Please read what is actually written and don't argue with your own strawmen.
> Every country wants fiat because they can print it, plain and simple
Then you have very little understanding of modern economies. Explain carefully the difference between the Fed and Treasury and how new money actually enters the system. You seem to know nothing of this.
The actual reason countries switched is the evidence against gold is overwhelming. Simply google the papers with the empirical data across several hundred years and ~100 countries that make it clear.
You can continue to argue from ignorance, or you can simply educate yourself and read the literature and evidence.
> The supply of gold is steady, growing at perhaps 2% per year.
This too is simply untrue, e.g., https://en.m.wikipedia.org/wiki/File:World_Gold_Production_1....
The mismatch between gold discovery and production versus needs of an economy is simply bad for a currency, which should expand and contract as the needs of the economy demands, otherwise there is more inflation/deflation cycles.
For example. around Christmas, as people increase spending, there is a need for more actual money, otherwise prices will fluctuate as too little money chases more goods. Every January this spending decreases, so, if the money supply didn't contract, then again prices will fluctuate as too much monry chases less sales. As a result, the Fed every xmas releases more money as demand increases, and every Jan pulls money back to prevent these cycles. You cannot do that with gold, or BTC, or any of the "too ignorant to understand money" crowd that has claims of "better" solutions without first understanding the current solutions, and in particular why those solutions exist and what previous problems they solved.
I suspect you're completely unaware of these things, yet continue to argue from ignorance.
> The Great Depression was caused....
You might want to read some literature on the topic first https://scholar.google.com/scholar?hl=en&as_sdt=0%2C15&q=gol...
Question: how many courses on economics have you formally taken and what were they?
Nice rebuttal bro.
>>You can keep denying the reality that gold has much value, but you're wrong
>I never said it didn't have value. It's bad as a currency. It's volatile and it hamstrings economies. Please read what is actually written and don't argue with your own strawmen.
It's not bad as a currency. It is portable, divisible, not perishable, and intrinsically valuable. It also does not vary a lot in supply. It's not as volatile as you keep asserting, in real terms. If there is ever a shortage of gold, it merely becomes more valuable and that encourages people to spend the gold they have and hold it for less time when they get it. If it gets more valuable, that also encourages people to dig it up. So any natural shortage is a self-correcting problem for many reasons.
>>Every country wants fiat because they can print it, plain and simple
>Then you have very little understanding of modern economies. Explain carefully the difference between the Fed and Treasury and how new money actually enters the system. You seem to know nothing of this.
I don't care to get into too many details (I know you will be a pedantic jerk about anything I say). But in short, the Fed is a pseudo-private entity intended to give an illusion of independence so that the government does not appear to have the power to directly print money. The Treasury borrows money from the Fed and the public. That is one way that fiat money gets created. There are other types of money too, such as fiat money created through commercial bank lending.
If you're not satisfied with my explanation above, let me refer you to this excellent presentation by Joe Biden's economic advisor Jared Bernstein: https://www.msn.com/en-us/money/markets/biden-economic-advis...
>The actual reason countries switched is the evidence against gold is overwhelming. Simply google the papers with the empirical data across several hundred years and ~100 countries that make it clear.
There are some plausible downsides to gold, but the academics are practically owned by banking institutions. Do you think you'll get to be a famous economist with accepted papers by going against what the cartel wants?
One reason we moved off of the gold standard was because in the wake of World War 2, the US held a disproportionate amount of the world's gold. Being in such a favorable position, the US pushed for a fiat system based on the US dollar with the promise of eventual convertibility of dollars to gold. From there, the US stopped converting dollars into gold directly in 1971, and since then everyone has been using fiat. Who in their right mind would set up a gold-based system to trade with the dominant power who has a fiat system? You'd be trading your precious gold for units that can be rendered worthless at any time. That is exactly why the world is dumping the dollar now and there is so much interest in the BRICS system.
>You can continue to argue from ignorance, or you can simply educate yourself and read the literature and evidence.
You can continue thinking you know everything just because you heard some "expert" apologist for the current system tell you, or you can think for yourself and read up on what I'm telling you.
>>The supply of gold is steady, growing at perhaps 2% per year.
>This too is simply untrue, e.g., https://en.m.wikipedia.org/wiki/File:World_Gold_Production_1....
You don't know what you're looking at. That's production. The current supply of gold is about 212k tons (I think metric tons): https://www.gold.org/goldhub/data/how-much-gold . According to your own link, the production in 2014 was 2860 tons. That's less than 2% of the total supply. I would link you to a chart of the supply over time, but it seems incredibly hard to find one.
>The mismatch between gold discovery and production versus needs of an economy is simply bad for a currency, which should expand and contract as the needs of the economy demands, otherwise there is more inflation/deflation cycles.
What happens with gold is that the price of gold just changes, credit is issued, and/or it circulates faster as needed. Printing currency or issuing credit creates an illusion of abundance when the economy is not productive enough to support the spending that's taking place. Basing the currency on a real thing discourages abuses of the financial system, because you can only keep up a charade so long before people see through it. You can increase the money supply with a gold-based system, just like with a fractional reserve system, so long as people don't call you out on your mismanagement of the supply and exit the system with their gold.
>You cannot do that with gold, or BTC, or any of the "too ignorant to understand money" crowd that has claims of "better" solutions without first understanding the current solutions, and in particular why those solutions exist and what previous problems they solved.
>I suspect you're completely unaware of these things, yet continue to argue from ignorance.
Oh I'm aware of it. You seem to not be though. I would never endorse BTC but gold is in another ballpark entirely.
>>The Great Depression was caused....
>You might want to read some literature on the topic first [condescending link here]
I've read books on the subject, on top of being told in every economics class that it was due to a credit bubble. There were other factors such as a series of crop failures, the Dust Bowl, and so on. But the real crash was in 1929, before the Dust Bowl. By the way, the Federal Reserve was in place before that crash. So I guess their fancy theory didn't fucking work now did it? Or maybe the directors of that system accomplished exactly what they set out to do, which is to seize the country's gold and enrich their friends, and to start the process of getting more power through a fiat money system.
>Question: how many courses on economics have you formally taken and what were they?
I've been out of school so long I don't remember. It doesn't especially matter. You can look up everything I said if you want to, but I don't think you do because you're too conceited. But really, I think I've taken at least 2 general macroeconomics courses. I have received the official story about the world monetary system and the Great Depression formally like 3-5 times in school, starting as a juvenile. It is all plausible, by design, and of course academics cling to that story as nothing else will be accepted by the people who run the system. They cannot tolerate people in positions of authority raising doubts about the financial system or the prudence of printing money.
That is, if you only look at wage inflation, it would mean that you can never factor in productivity improvements to how much more people can buy with their money.
(There's of course a big opportunity cost in employing the best physicists and biggest r&d investments in nukes - eg solar energy and grid storage could have been here decades earlier in an alternative scenario.)
A Manhattan project should be much more achievable today than it was in the WWII era.
But this is not inflation per se, it's what's counted as “economic growth”: the median family in 1940 had a much lower standard of living than the one in 2024.
But of course sorting what's inflation and what's relevant economic growth is highly subjective: some things like cars and housing have literally inflated (they are enormous now compared the what they looked like even 40 years ago) and that's usually counted as economic growth, but is it really? In some way it is: having a bigger house is nice. But in other way it's not: do you really need that much space when you're older and the children left long ago? And sometime it's actually a net loss: living 30 kilometers away from work and depending on a car for everything is not a gain.
Same for electronics: computers are much, much faster than they used to be in 2000, so the statistics counts that as economic growth. And it is, you couldn't do modern AI on 2000 hardware, or physics simulations for industry and all. But at the same time, for the average consumer, the gain in speed has been eaten by software consuming more and more resources. Sometimes from sheer laziness (there's literally no reason for Windows to be this slow) but sometimes it's a bit more ambiguous: today's video games couldn't run on hardware from 20 years ago, and I remember being very happy to see the graphics improve during my teenage years so it's not entirely pointless. But at the same time, it's not as if video games were more fun or enjoyable now than they were 20 years ago: the consumer surplus from video games is basically the same no matter the quality of the graphics, so maybe it shouldn't be counted as growth at all.
The measure of “real” economic growth (and the measure of inflation) depends a lot on how you weight these things (it's called “hedonic adjustment”) and as we've seen, it's very difficult, and fundamentally subjective. And I wish statistics agency provided a measure of CPI without hedonic adjustment in addition to regular CPI (like they already do multiple variants of CPI depending on the basket of goods that's taken into account).
Inflation is not underestimated by a factor of 3.5 over that time, as can be checked by nominal prices of a suitable basket of goods, which is the definition of inflation.
Ideally, inflation only observes the change in value of the currency. Easier said than done, of course.
There is no "except." This is central to inflation definition and econometrics.
As a result the average budget spends less on this component. Inflation tracks a basket of goods of what people currently buy, then compares that to neighboring adjusted baskets. For example, we don't buy as many horses and buggies any more, so that component of the basket has phased out. Electronics has phased in.
> Ideally, inflation only observes the change in value of the currency.
Not really, since different goods do not always correlate in price, so there is no "value of the currency". There are values of goods priced in a currency, which do not move in lockstep.
Inflation is defined as the general rise in prices, and in this case is precisely measured relative to (several) baskets of goods obtained by statistical sampling of the US population. All the precise methodology, data, sampling methods, etc., are all easily found on the BLS website.
Exactly. If there was no except, you could just peg inflation to an apple and be done with it. You need the basket exactly to filter out the noise that occurs when the goods in the basket see an independent change in value. Food is unquestionably less valuable today as compared to most of the past, irrespective of the change in value of the currency.
> Not really, since different goods do not always correlate in price
Exactly. That is because everything has its own independent value, including currency, which change over time independently. Let's say you can buy an apple for $1 today. But tomorrow a damaging storm decimated the apple crop. Now it takes $2 to buy an apple. Did the price rise because apples are now more valuable, or because the dollar is less valuable?
Who knows? It is impossible to know. But if you look at the change across many items...
> Inflation is defined as the general rise in prices
Exactly. The general rise in price is the value of the currency declining. It is statistically improbable for all the items in a basket of goods to gain in value at a similar rate all at the same time. As such, you can conclude with near certainty that the rise in price is a result of the decline in value of currency.
As it pertains to the above, if a general rise in price over a given period is 25¢, while apples are up $1 over the same period, then you can reasonably surmise that the storm caused apples to become 75¢ more valuable than the previous state with the other 25¢ being a decline in the value of the currency.
There is no perfect way to measure the change in value of currency, but that's a really good way to do it. Which is why we do it.
Contrast this with the US where a service like a massage would be >100x more expensive than a liter of clean drinking water. The same pattern holds true with other services such as shaves/haircuts etc.
“Clean up” is clearly a euphemism for some sort of money funnel.
Hanford.gov even readily admits this!
“In April 2009, the Hanford Site received a nearly $2 billion in funding from the American Recovery and Reinvestment Act. Contractors quickly hired thousands of new employees for temporary stimulus jobs in environmental cleanup.”
There's probably dozens of mechanisms at work to explain "why??!!", but the root cause is - like so many other damn things in America - financialization, from its birth post-1972 with vehicles like EUROBOND, to its "pedal to the metal" moment with all the liquidated commie national assets, to the craptacular commodity runup crisis courtesy of Bush II wrecking the Levant and West Asia. Oh, right, and the financial system ending in 2008. When we seamlessly blended up - legally - private with public risk. Oh yeah. Sixteen years ago.
Anyway, financialization. We don't account for money supply accurately, they're dynamic with price indices now, blabbity blabbity blah blah, 1962 dollars were probably worth waaaayyyyyyyyy more than we think they were.
(Since you can't time travel from 1962 to a point where those existed. If you just waited for them, then presumably you've kept them in an interest-bearing account in the meantime, and so the value of the dollar doesn't apply.)
Less so after decades of unchallenged safety and prosperity where it switches to nepotism and rent seeking as the default MO.
It's all cyclic.
It is a bit difficult to compare, because NASA shuffles around cost, and the plan is that Artemis will do way more (moon base) than Apollo. But Apollo seems to have been not only a technological marvel, but exceptionally well managed?
Regarding Manhattan project it has this mythical quality (the name alone is cool), but the Russians had the a-bomb only 4 years later and smaller countries (South Africa) are able to achieve it too.
Or did it simply benefit being done in a time where the critical components were less valuable? After all, projects like Apollo reveal to us new uses for materials, people, etc. so they tend to become more valuable afterwards.
What's bizarre is how few people understand how currency has become devalued since its separation from the gold standard, and even more so in the last 5 years. These are the same fools who put their money into "high yield" savings accounts giving them 3% when official inflation is >5%.
(Not to mention, one side (meta) has little to do with government funding to begin with.)
What Facebook does with the money they earn is up to them.
Would you have asked the same question if Mr Zuckerberg had spent it all on yachts?
That's still an insane amount of money.
That's just incredible, and not a good thing.
It seems to me to be far better than a company which simply gives 0.1% of US GDP back to shareholders in the form of dividends or stock buybacks, which is a very common occurrence given the size of companies nowadays. Or worse, a company which spends 0.1% of US GDP manufacturing things that are wholly useless or even actively damaging to consumers, the environment, etc.
Jury's still out on this one
> manufacturing things that are wholly useless
Companies manufacture things that people want to buy. Presumably those people find the things useful. Companies that manufacture things people don't want go out of business.
Do they, though? When was the last a time a really big company went bankrupt?
https://en.wikipedia.org/wiki/Bankruptcy_in_the_United_State...
Silicon Valley Bank was actually the third largest bankruptcy in US history, which is news to me. And apparently Signature Bank was fourth.
https://www.statista.com/statistics/1096794/largest-bankrupt...
The point is that's what would happen, and because of that, companies focus on making things people actually do want.
Look at what happened the Facebook stock over the last few years as they rebranded to Meta and started spending wildly. The market whipped them into shape, and rewarded them handsomely.
So any compliance with the 'wishes of market' was somewhat voluntary.
Revlon
Avaya
Talen
Hertz
Alcatel-Lucent - 2016 (Bell Labs/AT&T Technologies/Western Electric)
Nortel is a good example though. That was a big one.
https://en.m.wikipedia.org/wiki/2023_United_States_banking_c...
The whole point of investing in companies is for those companies to eventually give the shareholders more resources back than they put in. [0] Those shareholders can then consume those resources, or invest them in something else; like eg startups that pursue new technologies.
Returning capital to shareholders is exactly how investing works and how it has to work. Otherwise it would be charity, which is also fine, but it's something different.
Are you also against companies paying back debt, instead of endlessly rolling over one loan into another loan and paying interest only with money from new loans? (If not, how is that different?)
[0] On average, and in expectation. Investing in risky endeavours means that sometimes they fail, and you get less or no money back.
Facebook could do a stock buyback with all of that H100 money. It's a significant amount of money. As a human, complete non facebook shareholder except in ETFs, I would rather they buy H100's and train huge neural nets and then release them for free. Maybe I've bought the kool-aid but I think I believe that Zuckerberg genuinely wants to move humanity forward with this one (of course, it's also a business move). A lesser chairman of the board would have not spent huge capex on GPUs, and instead simply issued a fat dividend or what have you to shareholders who eventually would buy a newer nicer house in Marin or whatever. And indeed, it's only Zuck's unique position with preferential shares that allows him to do this; at a different company, he'd have been shot down and dividends it would be.
On the whole, which is better? What moves the needle in terms of bringing humanity forward?
And this is kind of true of companies in general. "Returning money to shareholders" is not created equal, even though it's all money in the end. Sports betting companies are also returning basis points of GDP to shareholders. Is that a worthwhile thing to do? Net good? (Please, dear reader, do not make the argument that sports betting is actually a good thing because hey we all make free market choices.)
Though to come back to the main argument: you seem to be comparing Apples to Oranges here?
So the first thing is: how are those companies making money? That could be via Facebook ads, or via sports betting, etc. The second question is: what are those companies doing with the money. That could be investing in eg training neural nets, or returning it to shareholders.
When you invest money, that's the same as spending resources. That's only worthwhile, when in expectation you get sufficiently more resources out later than you put in now.
In contrast, when you return money to shareholders, no resources have been spend. It's up to the shareholders what they want to do with their 'resource-coupons' (ie their money).
From the point of view of Facebook, dividends vs investment both look like spending money. From the point of view of the economy, the former is a mere transfer of resources, the latter is spending of resources.
(You could argue that it's better in some sense for Mr Zuckerberg to control these resources than for the average investor to have that control. I'm sympathetic to that argument, especially if you carefully state what sense you have in mind.)
> On the whole, which is better? What moves the needle in terms of bringing humanity forward?
Dividends only move money around. They don't move the needle one way or another. It depends on what the shareholders are doing with the money.
Btw, Amazon and Tesla show that 'the markets' can be extraordinarily patient with companies that they believe in. It's just that most managers don't have nearly as much charisma. Anyone can (and does!) say they have great long term plans, but it's hard to tell the charlatans from the true visionaries. So as a safe default, markets often prefer hard-to-fake signs of progress, like dividends.
Sports betting is non-productive, money goes into a sink and gets absorbed without production. The production of Titanic had to hire hundreds to thousands of professionals through multiple companies to produce something, those people and businesses get paid and spend it back in the general economy (both to produce their product and the surplus that gets generated goes back into investments for the next project).
Productive activities are always much better for society than non-productive ones like sports betting.
(Of course, your central bank doesn't check in your garden whether you buried some money. They follow price and spending statistics in aggregate.)
If you dig up your money, the central bank will print less money or even remove money from circulation by selling assets from their balance sheet.
It does make some difference whether you consume or invest. But that's because of a difference in how resources are used.
If you burn your money in a big fire, that's approximately equivalent to a donation to your central bank. If you bury it, and dig it up ten years later, that's approximately equivalent to an interest free loan to your central bank.
---
Production isn't useful by itself. Production is what we do to enable consumption. We don't need to worship production nor put it on a pedestal.
Both sports betting and producing the Titanic spends some resources to entertain people. If sports betting spends less resources for each unit of entertainment, that would be great. (In a competitive market, that would drive down the cost of sports betting, eg bookies would compete by offering tighter spreads or more exciting bets or whatever. Alas, I don't know enough about sports betting to tell what people find enticing about it.)
But the resources spend on both running the sports betting systems and on producing Titanic are gone. Those work hours of those thousands of professionals that worked on Titanic don't come back. They are a cost, not a benefit.
Isn't velocity of money a question in this though? The spending on producing Titanic pours the money into pockets (workers and businesses) who will spend it faster into the larger economy (consumption/hiring others to do projects) than what the sports betting company will, the spending from the wages of those workers will, on average, be more diluted into other productive means which will support other businesses and workers. In my view this is a more beneficial way for society than the accumulation of the same capital into a sports betting company, while betting has many more negative impact to society (addiction, for example) than watching Titanic has.
No, that's pretty much irrelevant. I was giving an example of someone digging a hole to dump their money in. That money's velocity would be pretty much zero. The central bank will just print more money to make up the shortfall.
It's the same, but less extreme, if velocity falls to eg half of what it was instead of zero: the central bank notices that inflation is below target, and prints more money.
The net result is that the overall nominal (!) spending in the economy is whatever the central bank wants it to be. For that total nominal (!) spending figure, decisions by individual economic actors are pretty much irrelevant; because there's a control system with a negative feedback loop.
It is better for Zuckerberg to control those resources, indeed that's what I'm saying. He famously has special shares that give him outsized control over how facebook spends its money. And he was famously punished by the street -- that disastrous Meta earnings in 2021 when zuck decided to invest huge capex in GPUs. A normal company would not have made a move like that, they would have issued dividends with the money instead.
What I'm saying is, that's a good thing. Markets aren't perfect (they are actually pretty f!@#@$ dumb, especially about what a GPU was worth in 2021, and about what sports betting is worth today). I'm making a value judgement.
FB shareholders would have spent that 2021 dividend on ice cream or a remodel in Marin, and it's better for all of us that in 2021 they were forced by zuck to go without the ice cream and the remodel. Because we all got a free 400B param neural net. Humanity did.
Oh you mean like Facebook, Instagram, and friends?
When I see the best in people and then see the silly loops and squiggles we get ourselves into it is like playing Chopsticks on Mozart's piano. We know it could do so much more. I mean, we are free to play chopsticks and that is fine, but ultimately in focusing on the short term despite being aware of the long term, it eventually gets taken out of our control and the base ecosystem has its way with us. It is better to show restraint now rather than have it forced.
Society has (and should) continue to become wealthier. The laws that govern us has improved from the medieval days where money is everything, so just because a private entity is able to spend a lot doesn't imply there's anything untoward.
I don't know enough about Facebook to say whether the expected value of that bet was positive. But that's not my judgement to make in the first place.
> The laws that govern us has improved from the medieval days where money is everything, so just because a private entity is able to spend a lot doesn't imply there's anything untoward.
Yes, I agree that it's a good thing that people and companies can spend their money how they see fit.
Facebook tossing us a pseudo-open source kinda-free LLM will make some of the nerds on this site happy though (like how decades of evil from Microsoft were erased with the release of a free text editor lol). I guess that's all that really matters at the end of the day.
Why? It shows that they were willing to spend lots of resources on something risky. That's how you get progress. Risky things don't always pan out (almost by definition).
(And I have sympathies for that! We should have more smaller countries, like Singapore or Switzerland.)
Here in EU I saw many people that basically just clock in/out everyday and then mind their own business because they are un-fireable due to labor laws/unions. A colleague was calling these people "institutionally pre-retired", which I believe it's a great description of the phenomenon. To some extent that happens in the US also in the Public Sector, but I believe it's rarer.
- There's no way Meta is paying retail.
As aa part of GDP from 1970 (1.073T) the total cost of Apollo would be like 4% of that single year's GDP. Not a good comparison because when Apollo started GDP was 563B and was 1.3T when it ended.
[0] https://www.researchgate.net/figure/Total-NASA-Appropriation...
[1] https://www.statista.com/statistics/580087/energy-use-of-fac...
Companies returning money to shareholders is pretty much equivalent to companies returning economic resources to shareholders.
Compared to the process of producing a 3nm chip with all those billions of transistors, the Manhattan project isn't huge. Groundbreaking, sure. Huge, compared to the development of EUV wafers, no.
90% margin is likely the low end for Nvidia's AI offerings.
I dont work in hardware, so i dont know what is involved. I do however work in software, where people are constantly saying things are easy or devs are just lazy. Until they put their money where their mouth is and try it themselves. Usually they are in for a rude awekening once it becomes obvious they are full of shit.
Indeed. Sustained nuclear chain reaction was achieved in Chicago Pile-1 beneath the stands of campus football stadium at the end of 1942. That's when, scientifically, it was proven that the bomb is possible.
If we consider just ASML, (the main) supplier of only the EUV machines (and accessories) for manufacturing chips (they don't run the fabs, they just build and maintain the equipment), they also directly employ 42,000 people and utilize another 5,000 tier 1 suppliers. https://en.m.wikipedia.org/wiki/ASML_Holding
That's just the ONE business to make the machines to make the chips, that doesnt include all the folks who make the chips! If we count just one of their customers (TSMC) that's another 70,000+ employees!
National projects are impressive for different reasons than whole industries.
"At the Hanford Site in south-central Washington state, 177 giant tanks sit below the sandy soil, brimming with the radioactive remnants of 44 years of nuclear-materials production. From World War II through the Cold War, Hanford churned out plutonium for more than 60,000 nuclear weapons. The sprawling enterprise eventually contaminated the soil and groundwater and left behind 212 million liters of toxic waste—enough to fill 85 Olympic-size swimming pools. Decades after the site stopped producing plutonium, the U.S. government is still grappling with how to clean it all up. "
[edit] anecdotally, as related by people who watched the transition happen, one of the things this new management culture screwed up was management itself, especially lower and middle management. There used to be a gradient of how much time managers spent in meetings with peers and higher-ups, with the lowest ones doing very little of that and mostly focusing on the needs of their subordinates and keeping things running well—after the shift, all levels of management got sucked into a culture of meetings, meetings, meetings, most of them very low-value. Posturing and politics soaked down into lower levels of organizations so-afflicted to a much more extreme degree than before.
But it fades into history.
Nvidia: 26,000 people
TSMC: 73,000
Intel: 124,000
AMD: 25,000
Qualcomm: 50,000
ASML: 42,000
A more illuminating question might be "wtf happened in the 1940s and how can we get back there short of another total war?"
Wouldn't you know it, adding a little socialism to free market capitalism is insanely effective...
Another thing people don't seem to recognize, after both world wars the west basically dismantled all proprietary IP from Germany and gave it away for free. This affected everything and anything that was patented. It was as if the entirety of science development for several decades was open sourced.
And it was glorious and literally minted a golden age for about 30 years.
* Lack of injection-molded thermoplastic suitable for much more than toys.
* Relatively high labor costs may have reduced incentives to save money on materials.
* Consumers accustomed to paying higher prices for manufactured goods with the expectation that they would be repairable and last a long time.
* Tube-based electronics tended to require high voltages and heavy transformers or batteries. This dictated the form of the enclosure (metal or wood) and its usage (non-portable). For example, radios made for consumers would typically be incorporated into furniture.
Yeesh. This statement is not defensible.
The fifties-era cookie cutter houses were pretty bad (but at least they moved past using coal and gaslight and knob and tube wiring and so on). The quality of passenger cars has only ever trended upward, with a dip during the combined energy crunch and inflation of the seventies that was remedied pretty quickly. The sixties ushered in the era of solid-state consumer electronics.
Survivorship bias. I have a 19th century Eerie Company cast iron pan. It’s beautiful. It is not representative of most cast iron of the era. We can cast far better iron today, we just don’t devote those production processes to making pans.
Additionally the U.S. had probably better education and a stronger work culture. When only 5% of your country goes to college the classes can move really fast.
Quoting Dijkstra, https://cacm.acm.org/news/an-interview-with-edsger-w-dijkstr...
> I had no idea of the power of large companies. Only recently I learned that in constant dollars the development of the IBM 360 has been more expensive than the Manhattan Project.
Hee-hee! Immediately following that is:
> I was beginning to see American publications in the first issue of Communications of the ACM. I was shocked by the clumsy, immature way in which they talked about computing. There was a very heavy use of anthropomorphic terminology, the "electronic brain" or "machines that think." That is absolutely killing. The use of anthropomorphic terminology forces you linguistically to adopt an operational view. And it makes it practically impossible to argue about programs independently of their being executed.
Anyway, to confirm the main point, "IBM spent 5 billion dollars in mid-1960s money" says https://thehistoryofcomputing.net/the-ibm-system360 while "the Manhattan Project, which resulted in the first atomic bomb, cost $2 B". https://en.wikipedia.org/wiki/Manhattan_Project says that was 1945 dollars, so about $3.5 billion 1965 dollars using the CPI Inflation Calculator at https://www.bls.gov/data/inflation_calculator.htm .
Going backwards, this $30 billion on GPU training is only $3 billion in 1965 dollars, making it cheaper than the 360.
Meta knows where to compete, where to keep a foot in the door, and where to be a bystander.
"Meta AI isn't available yet in your country"
The Llama 3 70B model in English already beats all models from Google, from Anthropic and it's only bested by the last GPT4 version. (Based on the Open LLM Arena results)
It seems like there must be a ton of things other than the underwhelming “meta ai” chat box that they could be throwing H100s at - anything from content recommendation engines to anti-abuse to optimizing network bandwidth and traffic flows.
I think it would be clearer to express these costs as percentages of the global / societal (e.g. economy of Europe, of greater Chinese sphere, etc) / national economy.
Maybe I'm missing something.
Now, what could we have achieved in useful areas, with that kind of money?
When Google entered the telco space you would think they would easily take down the ancient parasites that are AT&T/Verizon/Comcast, but what happened?
Same story with Apple trying to get into auto/banking (credit cards)/content.
So the money is only going to be used in this way. They can't touch Exxon or Monsanto or Pfizer or Boeing. They can't do a thing about housing or education or health or banks.
So burning cash on novelty becomes their default path of least resistance. "Useful areas" have all been walled up.
If laws were changed so IP/data wasn't protectable, then next year there would be lots of people making cloned/modded copies of Facebook. Likewise, there would be lots of people taking Nvidias GPU design and making variations on it.
(this is assuming a world where everything currently closed source is made open source through law changes)
And in both cases, competition would push prices and profit margins to near zero.
Even if the code was all open sourced, the data was portable (e.g like open banking), and the messaging standards were open (like XMPP); the supermajority of users would keep using FB.
See the example of gmail.
With that much money I'd imagine you could purchase a GPU manufacturer or something.
They didn't spend this cash for laughs, they're buying it for a purpose, and that purpose existed before ChatGPT was a big hit. Both Zuckerberg and Yann LeCun have stated in interviews that they're invested this money into AI for Facebook feed recommendations and for competing directly with TikTok in their Reels app. In other words, Meta aimed the digital equivalent of a weapon of mass-destruction at your kids' brains, with the express aim of getting them addicted to their products just like TikTok, but even more.
If you want a visualisation of the kind of raw power that your teenager is up against, I came up with this:
- A billion can be visualised as as 1m x 1m x 1m block subdivided into cubic-millimetre tiny little cubes. Think very coarse sand or very small pebbles. Just picture yourself trying to count out the grains in a block that size!
- A billion-billion is a cubic kilometre of the same sized grains.
- Facebook used two-trillion-trillion computations to train one of the smaller Llama 3 models! That's 2 million times the amount above, about the same as 1000 km x 1000 km covered 2 km deep. That's an area the size Egypt. Alternatively, Greenland covered 1km deep, or Texas 3 km (2 miles) deep.
That's what your kid is up against.
Good luck.
The interviews are:
"Mark Zuckerberg - Llama 3, $10B Models, Caesar Augustus, & 1 GW Datacenters": https://www.youtube.com/watch?v=bc6uFV9CJGg
"Yann Lecun: Meta AI, Open Source, Limits of LLMs, AGI & the Future of AI | Lex Fridman Podcast #416 ": https://www.youtube.com/watch?v=5t1vTLU7s40
The total FLOPS used in training is in bunch of places, I can't find a good link at the moment.
Also, most of their compute goes towards inference, not training.
What could go wrong.
https://www.cnbc.com/2023/04/14/nvidias-h100-ai-chips-sellin...
https://www.tomshardware.com/tech-industry/artificial-intell...
However, they are likely vastly cheaper in bulk at MSRP, that's true. And Meta apparently got them before the AI hype cycle.
My favorite part:
"When I went back to work on the calculation program, I found it in a mess: There were white cards, there were blue cards, there were yellow cards, and I started to say, "You're not supposed to do more than one problem -- only one problem!" They said, "Get out, get out, get out. Wait -- and we'll explain everything."
"So I waited, and what happened was this. As the cards went through, sometimes the machine made a mistake, or they put a wrong number in. What we used to have to do when that happened was to go back and do it over again. But they noticed that a mistake made at some point in one cycle only affects the nearby numbers, the next cycle affects the nearby numbers, and so on. It works its way through the pack of cards. If you have fifty cards and you make a mistake at card number thirty-nine, it affects thirty-seven, thirty-eight, and thirty-nine. The next, card thirty-six, thirty-seven, thirty-eight, thirty-nine, and forty. The next time it spreads like a disease.
"So they found an error back a way, and they got an idea. They would only compute a small deck of ten cards around the error. And because ten cards could be put through the machine faster than the deck of fifty cards, they would go rapidly through with this other deck while they continued with the fifty cards with the disease spreading. But the other thing was computing faster, and they would seal it all up and correct it. Very clever. That was the way those guys worked to get speed. There was no other way. If they had to stop to try to fix it, we'd have lost time. We couldn't have got it. That was what they were doing. Of course, you know what happened while they were doing that. They found an error in the blue deck. And so they had a yellow deck with a little fewer cards; it was going around faster than the blue deck. Just when they are going crazy because after they get this straightened out, they have to fix the white deck the boss comes walking in. "Leave us alone," they say. I left them alone and everything came out. We solved the problem in time and that's the way it was."
If I was the supervisor in the above situation, I wouldn't let them cut those corners, since it's far more important to get the right answer when $2B of project could be led down totally the wrong path if a calculation tells you you need 10x less/more uranium than reality because someone shuffled up some punch cards wrong...
Given the error rates described by Feynman, if they had to restart the entire computation from scratch after a single detected error they'd likely never have completed the task.
It's a bitflip. Just one that makes a surprisingly loud chunkk noise.
Who really knows, but I tend to think the nuclear physics were pretty involved. I enjoyed the What John Von Neumann Did At Los Alamos, from the evr excellent 3quarksdaily crew, which talks about how his brilliant mind combined with his explosives expertise was key to steering the Manhattan Project towards success. The article enhanced my appreciation for the challenge. https://3quarksdaily.com/3quarksdaily/2020/10/what-john-von-...
If he is not the boss, he might have got laid off couple of times long time ago.