Google is the reason for the current AI boom. Without the transformer architecture they invented by funding basic research, there would be no modern LLMs. YC is arguing that their incentive for funding that basic research should be taken away in order to spur innovation?
But when AT&T had a monopoly it funded Bell Labs which was responsible for much innovation.
Then AT&T was shut down and Bell Labs went away.
If we take your argument seriously then AT&T shouldn’t have been dismantled. But it was a good thing AT&T was dismantled. It helped lead to the modern internet.
By your logic all Rockefeller had to do in the early 20th century was set up a lab to do basic research and then Standard Oil wouldn’t have been broken up.
Monopolies should be broken up. This is true regardless of any basic research that they fund.
And it was an antitrust action that unlocked a lot of that value. The consent decree required Bell Labs to license its patents (e.g. transistors) for reasonable royalties. The same consent decree also forbid AT&T from entering new industries like computing. So after they built UNIX, they sold the source code 'as-is' to universities for $200 ($20k for businesses).
Ask anyone who was alive back then and they will tell you stories of how legendarily awful AT&T was to deal with. My father has told me several. The antitrust action made things better for regular people by allowing them to do things like buy their own handsets or haggle over price.
All of them, in regions where they don't have a vertical monopoly. You can negotiate away installation fees and monthly package pricing on DSL, TV, internet phone... also sometimes get a no-contract deal instead of locking in for 24/28 mths with the dreaded ETF which is a large part of Comcast's profitability.
Citation needed. I hear this repeated, but the consumer experience was it was split into regional monopolies, and consumers now had to deal with both local and long distance, and both were still monopolies. It only got better with competition from mobile providers.
YMMV. I don’t agree with the narrowness of this analysis, and would like to see some links to academic studies in economics and the study of innovation tbh.
I'm pretty sympathetic to both sides of this. I don't really know the history well enough to say whether you're right that breaking up AT&T "helped lead to the modern internet". But even stipulating that it did, the loss of monopoly era Bell Labs was tragic.
Both things can be true! It's entirely possible (probable even) that breaking up monopolies has both positive and negative impact.
And I would be a lot more sympathetic if we had a lot more public investment in technology. But we don't. What I see is both public and private research investment under major attack. I think that's a recipe for disaster.
There's actually a pretty solid argument to be made that the railways were never more effective than during the days of Rockefeller, when he could throw his money around to force otherwise-competing railways to optimize their work in an industry that has some fundamental strong incentives to be non-competitive (it rarely makes any sense to run two rail lines purely for competition reasons and leads to a race to the bottom on pricing).
... but that's more a story of the failure of the US government to go far enough and nationalize the rail network and its operations. The most efficient era of US rail was during World War II, when the military took it over and prioritized schedules by optimal throughput over profit concerns.
Right, if you look back through the history of ideas, every breakthrough builds on prior research and inventions. In the realm of patents and copyrights, this is acknowledged formally: they expire after a certain time and enter the public domain. This also supports the view that the current state of the world, for better or worse, owes much to the past and those who came before (living or not living elements).
Bell Labs being defunded by a deregulated/competetive AT&T was precisely what led to the attempted commercialization of Unix and the near death of what would eventually be called "open source", though. In history as it stands, we had GNU and Linux and all that we lost was a few years.
But it's easy to imagine a world where that didn't happen and BSD was just killed dead. So no OS X, no iOS, no Android, no ChromeOS, and the only vendor able to stand on its own is the one we all agree had the worst product.
Ironically the world where the Bell monopoly was left in place seems to me to be one where we're all stuck running Microsoft Windows on everything, no?
I mean, fine, there's nuance to everything but the idea that "well, open research isn't so important" seems frankly batshit to me. Monopolies fall on their own all the time (Microsoft's did too!). You can't get stuff into the public space that wasn't ever there to begin with.
You’re making the assumption that only corporations can fund or perform basic research. But the transistor was actually the culmination of decades of research by materials scientists and physicists in university and other labs into semiconductors before anyone realized there were applications.
No, I'm making the observation that corporations do fund or perform basic research; and the straightforward inference that removing the incentive for them to do so is bad.
Is that wrong? Possibly. But if you're hanging your argument on the idea that the transistor would have been invented without Bell Labs (or BSD without AT&T's open research, which was my example), I think it's on you to provide the evidence.
>Monopolies fall on their own all the time (Microsoft's did too!)
What in the heck are you talking about??? After Microsoft was convicted, even though they never received any actual punishment, the were very internally cautions about any behaviors that could be perceived as monopolistic. This is like a total misinterpretation of what actually occurred on your part.
And yes, monopolies do fall, after very long periods of time. Some monopoly sitting around 25 years may not seem like much, but that's half an average persons working life.
You responded to a parenthetical. The point was about AT&T, not Microsoft. And yes, they effectively killed BSD, relenting only in the late 90's once it became clear that (post-Linux) there was little value left in Unix. In a different universe, they go to the mattresses with SysV against Windows NT and lose, and we have nothing.
Yet empirically, the biggest funders of basic research have historically been monopolies. The US government was, at least up until the last few months, the largest funder of basic research globally and it obviously maintains multiple different monopolies, a monopoly on legal use of force, a functional monopoly on financial transactions as the global reserve currency, and I'm sure others. Excluding national governments, Bell Labs and IBM were both probably the biggest funders of basic research in the last century during their respective heydays. Bell was obviously a monopoly and while IBM might never have faced anti-trust penalties, they did at one point control 70% of the mainframe market and the DoJ did bring a case against them (that was eventually dismissed by the Reagan administration.)
I think there are many good reasons to pursue anti-trust action against companies that are in a dominant market position, but we should be honest about the tradeoffs. Businesses that have to aggressively compete to maintain market share don't have the slack to fund basic research.
Is this not a bit "tail wagging the dog" thinking? There wasn't much innovation in telecommunications once Bell's monopoly was entrenched. Once it dissipated, innovation was everywhere in the space. Similarly, computers had less innovation while IBM was a monopoly than they've had since its monopoly dissipated.
Though both companies created novel & useful inventions, the biggest shifts in those industries during those monopoly eras were from outside those organisations by competitive startups. As an example, IBM should have produced Microsoft, but they didn't. They missed out on a multi-trillion dollar value creation opportunity as a result.
There's a difference between conducting basic research and bringing new inventions to the market in the form of consumer goods or services. Monopolies are much, much worse at doing the latter than normal businesses because they have no competition pressuring them to improve their offerings.
> There wasn't much innovation in telecommunications once Bell's monopoly was entrenched. Once it dissipated, innovation was everywhere in the space.
I don't think that's accurate at all. If we take say 1920 or so as the date when the monopoly was entrenched and 1984 as the break-up, there was tons of innovation in telecom in that time period. Novel telecom technologies introduced in that time period include television, microwave relays, satellite communications, submarine telephone cables, cellular telephones, fiber optics, electronic telephone switching, packet switching, the Internet.
Which is why, like the 'monopoly on violence' the government should also be funding a _lot more research_.
It should be at, or partnered with, higher learning institutions and since it's public funded all of the results should be free to use*. I'm willing to entertain the idea of: Free use for people and corporations within the country/countries that funded research, everyone else pays compulsory license fees.
But public funded research isn’t “free to use.” In many cases, you can’t even read it without paying a scientific journal for a subscription. See the Bayh-Dole Act as well: universities can patent discoveries from federally funded research.
The intent of the Bayh-Dole Act was to deal with a perceived problem of government-owned patents being investor-unfriendly. At the time the government would only grant non-exclusive licenses, and investors generally want exclusivity. That may have been the actual problem, moreso than who owned the patent. On the other hand, giving the actual inventors an incentive to commercialize their work should increase their productivity and the chance that the inventions actually get used.
> I think there are many good reasons to pursue anti-trust action against companies that are in a dominant market position, but we should be honest about the tradeoffs.
Okay. I’ll take the monopolostic government over the monopolistic corporation. Thanks.
This line of argument to defend monopolies is the same line of argument against progressive taxation of high incomes. Just because those with trends excess use some small fraction of that excess to do good does not justify the means.
I have been close enough to billionaires and how they spend their money to not be fully impressed by such arguments.
And certainly the robber barons of the 2000s spend far less on the public good than when tax rates were higher and they used to fund universities, libraries, hospitals and the like.
The biggest funders of basic research are those with the most resources. This is your insight? I don't think anyone disagrees. Then you conflate correlation with causation and move it to _monopolies_ fund basic research. Bravo.
That's true, but if it weren't for Google's monopolist position, I doubt they'd have the money to throw at the wall for random research. For every AI transformer they revolutionized, there's a self-driving car project that's dragging on for decades.
Had Google operated like a normal company, the risk/reward of this kind of research would've looked completely differently.
Google's monopoly helped research along in the same way totalitarian countries like China are developing infrastructure at break-neck speed: if you don't need to care about pesky rights and regulations, you can do things that would otherwise be impossible.
I don't think Google's monopoly is worth having the current generation of lie generator bots around, but I don't think generative AI would be where it is right now had Google been forced to comply with antitrust regulations ten years ago.
It's funny that you use self driving cars as a negative example. We have a perfect natural experiment to look at to compare the slower research-driven approach (Waymo) to the "normal company" short term profit driven approach (Tesla).
From where I'm sitting it's pretty clear which approach has been more successful.
My first thought was Bell Labs here. The lack of time pressure for research results sounds like a big reason why so much innovation happened—people could pursue projects that may not have immediately benefitted the company's bottom line, because Bell had money to throw. I think UNIX was an example of this, because MULTICS was a failure and Bell was wary of similar projects, but I might be wrong.
According to Kernighan in UNIX: A History and a Memoir, the Unix team was constrained with regards to hardware capital at the time of Unix's creation. The Multics team got to use a fancy GE-645 (36-bits!) while the Unix team had to beg for "cheaper" systems like the PDP-7 and eventually an 11. Thompson has a great quote about how ultimately he was thankful they didn't have as much money to play with as the Multics team did but at the time he was annoyed he had to beg for a PDP-11. Fun book!
Most of the Bell Labs fundamental breakthroughs had no path to commercialization at the time of their discovery. The transistor was discovered 12 years into Shockleys research on fundamental properties of semiconductors.
If commercialization had been considered before funding, the project would have never been approved.
Instead, Bell management took the view that they could find everything, and some of it might become a new market, maybe.
> You certainly have a point. Places like google and bell labs have pushed innovation, apparently enabled by monopolies.
I've heard this argument before (and recognize that you aren't defending it), but telecommunications, network and technology innovation has hardly suffered since Bell was dismantled in 1982.
Telecommunications is still led by small groups of companies. Giant backbone providers stitch the internet together. Telecoms providers within a country, the ones that actually have hardware in the field, can often be counted on one hand, and often in one hand after a fireworks accident. 5G/6G/7G research is led by a small handful of companies that actually build the switches, transmitters, and modems. Cell tower frequencies are sold to a tiny group of carriers that sublet their network equipment to smaller companies they eventually buy up (if succesful) or disappear from the market (if not succesful).
Fiber rollout in countries where a government funded phone line rollout has already succeeded is laughably slow, taking decades and many billions with little to show for it. Even in countries where no phone lines were rolled out back in the day, fiber is more and more being skipped as 5G allows for cheaper (though less reliable and less capable) deployment.
I'm not saying monopolies are good or anything, and I think our problems would be even worse had we stuck to the monopolist systems that brought us telecoms as we know it, but I wouldn't consider the industry one where there's enough competition to drive innovation, especially since at least half of the entire sector is competing against Chinese government-controlled companies with seemingly endless coffers.
Right, but there other, real, negative impacts of monopoly, whereas the positive impact of R&D funding seems to be at best a maybe as to if it's better than the alternative.
This is debatable, I think. What I've read is that, whereas Bell Labs did foundational, groundbreaking research that radically altered the course of human history, technological innovation since then has more often followed the lines Bell Labs, and their ilk, laid down. Giving the world a faster computer or faster network is fine and nice but pales in comparison to the consequences of giving the world modern computers, Unix, and C.
I might not have explained my position adequately. I am not commenting on how Google became immensely successful, but just looking at it today in its current state.
It seems like the end result of unchecked capitalism is monopolistic practices. Companies want to make as much $$$ as possible. The most effective way to do that is to become a monopoly and avoid/destroy/inhibit any amount of competition as possible.
I really see it as an unstable system. Which is why we need society to put laws in place to keep the system in check, so it doesn’t turn sour.
Basically there’s a “healthiness” scale of capitalism. I want healthy capitalism.
This is debatable and probably a US centric POV, I think you could argue that much of the world is better at public / private partnerships and so no, their position is not due to capitalism, even if their continued existence is.
I’m not an expert on the economic history, but you could probably argue that the last big public / private partnerships in the US were in the railroad days. Defense spending I would not count in this category either. Although it builds whole ecosystems it is too insular and incestuous compared to something like transportation, or deliberately nurturing any other budding industry in a cooperative rather than competitive fashion
That's an interesting question. A lot of basic research is done in pursuit of creating entirely new markets. Being the first entrant to a new market makes you a monopolist in a way by default. It doesn't necessarily have to be anti-competitive.
I'd argue the fact that Google publishes much more of their research than their competitors do is a strong indicator that they're actually not the anti-competitive ones.
It's normal for mature industries to evolve into oligopolies. The filing goes far beyond fighting cartel-like behavior, into pretty ludicrous stuff (like "open access to Google’s datasets and search Index").
The greatest private sector basic research institute that has ever existed was the result of a government granted monopoly, so you have to admit, it helps.
I worked for a while at the company that used to be named Bellcore (Bell Communications Research), which was originally a research consortium for the Baby Bells. By the time I joined, it was a zombie cash cow shell that did no research, and it met the common fate of such businesses: acquired by a private equity firm.
There are many other search engine options, why should Google be punished for hiring the best people over decades to make the best one? Many consumers also switch their search to Google when presented with other options or defaults.
> why should Google be punished for hiring the best people over decades to make the best one?
Google should not be punished for having gained a near-monopoly by hiring the best people.
But perhaps their near-monopoly is due to other reasons like paying large amounts to ensure that they are the default option on various platforms. In this case, legal restrictions might be more appropriate.
How do we determine which of these proposed reasons for the near-monopoly is correct? We use a legal process where each side presents their evidence and a neutral party decides which is most credible.
> There are many other search engine options, why should Google be punished for hiring the best people over decades to make the best one?
I’m guessing this whole court case wouldn’t have been a thing if Google wasn’t bribing Apple, to the tune of $20 billion a year [1], to remain the default search engine on iOS.
Because the data they possess is the monopoly, they have better search results because it is too expensive for competitors to gather the data at this point as(I also think people who are unfamiliar with finance do not understand the quantum, Meta has a similar type of business but was under huge pressure with investing tens of billions into VR...it would cost hundreds of billions to get parity with Google, there is no way to finance this).
It is nothing to do with "hiring the best people". Google's exec-level leadership is extremely poor, Mr Magoo-tier management. This is largely due to their share classes, the people at the very top are not very good at business so Google largely isn't run like a business. They have one business that is probably worth $4-5tn, and the rest is worth -$3-4tn. The number of "best people" out there is usually under 500 in a country the size of the US, a country that has hundreds of thousands of employees is not hiring the best.
I would guess under 100 people at Google actually positively impact financial results in any way because the advertising business grows rapidly, uses no capital, and requires no staff. This is true of many of the tech companies do, you aren't getting the best if you pay an exec $50m because the best will always do their own thing and make more. Rather you get someone like Pichai or Cook who sounds good and will get shareholders to believe that setting fire to $200m/year to pay them is a good idea, they are indistinguishable from politicians.
Google is under-earning massively. Staff aren't a monopoly, you just pay someone to leave and they are yours now (you see this in other areas like HFT where staff actually know useful stuff, you don't see this in tech because most staff don't know anything, they are looking for drones).
It's incredibly naive to think Google's continuous growth is automatic. It happened because of all the work the employees do, not in spite of it. Are there employees who don't contribute? Sure. But that's different from only 100 contributing to positive growth.
Do you actually understand how and why they are growing?
They start with a page with zero ads, they add one ad to that page one year, ad sales pushes that placements, then a few years they do it again, etc. Meta are the same.
Google do sell at a higher price than offline ads because of targeting/intent but this is inherent to the product: search has inherently better intent and their targeting tech is no better than anyone else such that their prices are higher/grow faster.
That is their growth model: higher ad density, the CEO deciding to add another ad to a page and earning $100m/year. There is no value-added otherwise because you don't need many people, you don't need capital and you are making $300bn/year.
And, again, if you own the shares you understand that you are getting the ads business, and because of the dual class you are also getting this tech bureaucracy that sets fire to tens of billions every year employing people to do nothing. If these people are so productive...where is the revenue? It is all ads or ancillary business, they have GCP now but there is nothing else...because these people aren't doing anything.
It is like owning a business that turns lead into gold, the process is automatic, requires no capital...and then employing a bunch of monks to pray for the lead and saying the business couldn't exist without them. Lol.
> The number of "best people" out there is usually under 500 in a country the size of the US, a country that has hundreds of thousands of employees is not hiring the best.
> I would guess under 100 people at Google actually positively impact financial results in any way
It's really funny to me that people write comments like this and yet are incapable of basic arithmetic, like this being 0.0001% of the US population, or thinking that the entire ad business runs on just 100 people who are magically making the company $4 trillion dollars while everyone else drags it down.
> Without the transformer architecture they invented by funding basic research, there would be no modern LLMs.
Without Google the researchers who invented the Transformers model might have launched their own startup instead of sitting on the technology for 5 years while it's mismanaged at a big company. We would have had LLMs in 2018 not 2022.
A little bit unrelated, but the rise of Transformers in 2022 was because of the compute available - in 2018, it would have been almost impossible to make something like GPT-4.
Yes, large monopolistic corporations can spend large amounts of money on, presumably, completely unmonetizable research.
Nevertheless anti-Trust law exists because of the belief that monopolies should not exist and that it is the governments function to dismantle monopolies. The consequence of that is that corporations who can freely spend hundreds of millions on basic research will be dismantled as well, as happened with AT&T, and the funding for the basic research will cease.
>YC is arguing that their incentive for funding that basic research should be taken away in order to spur innovation?
No. That is the stance of the government. YC is arguing that the remedies the government is seeking are appropriate.
> YC is arguing that their incentive for funding that basic research should be taken away in order to spur innovation?
Ma Bell is arguing that Bell labs has been a fountain of knowledge everyone admires and has contributed tremendously to the advancement of telecommunication systems.
> Google is the reason for the current AI boom. Without the transformer architecture they invented by funding basic research, there would be no modern LLMs.
Sounds great to me! Wish they had been broken up before that happened.
Juergen Schmidhuber and his students came up with much of the basic NN elements, such as RNN, LSTM and others without behemoth resources (and several years earlier). Ttbomk, the one thing they DIDN’T come up with was the transformer; however it is very likely someone would have within a 5-10 year time frame.
I well aware of Schmidhuber, still the scaling of the compute was critical. The reason Schmidhuber didnt go all the way is still scaling/capital, which accrues to monopolies who can afford wildly speculative research. Also, LSTM, RNN, etc, while effective for their time, were dead ends.
I'm not saying that, I'm saying you need a company with huge amounts of cash on hand to spend on R&D. The monopoly provides a safe space for it, like it or not, this system has worked.
the argument seems to be basically that if Google hadn't had a dominant position they would not have had the money to make a new technology that was important for the current AI wave, and they should have the right to exploit this technology that they helped create - disregarding that in the American system this is what patents exist to enable.
OpenAI is the reason for the current AI boom. Google wasn't productizing anything and didn't put any of this stuff out in the open. Where was their productization of the transformer?
If anything, it should show that Google malinvests. Maybe none of it would have seen the light of day. Only now that they've been threatened are they building products.
OpenAi just decided they could jump the gun while Google CORRECTLY deemed the tech not ready yet. What you're breaking your neck to find fault with was Google being responsible.
They had AI assistants and machine vision aplenty. The current hype cycle seems to stem from the discovery that going big on these models was worth it.
That you are unaware of the applications of transformers is on you, not them. Search query understanding, search result ranking, translation, voice recognition, all other natural language applications, and generative applications like Gmail Smart Compose are all based on transformer architecture.
