Sherman Anti-Trust Act (1890)
archives.gov> The act was designed to restore competition, but it was loosely worded and failed to define such critical terms as "trust," "combination," "conspiracy," and "monopoly." Five years later, the Supreme Court dismantled the act in United States v. E. C. Knight Company (1895). The Court ruled that the American Sugar Refining Company, one of the defendants in the case, had not violated the law even though the company controlled about 98% of all sugar refining in the United States. The Supreme Court reasoned that the company’s control of manufacture did not constitute a control of trade.
The reality is that the law itself was not very well written, and takes an extremely litigious executive branch to do anything with.
Even if you agreed with the premise (the idea that Standard Oil arose purely through conspiracy is clearly wrong), it's ultimately a very ticky tacky set of definitions and has failed to achieve anything meaningfully as policy (do we actually feel strongly that breaking Microsoft's productivity suite from their OS division was really worth all the effort?)
(The notable exceptions being actual illegal coordination - like was recently uncovered in the egg industry. But these have more to do with price fixing amongst competitors and less with actual Trusts/Monopolies)
The only practical way that countries have tackled monopolies has been to write legislation that specifically puts oversight on that industry.
That's what the article says. But Knight was mostly overturned by Swift. Wikipedia has a decent overview of US antitrust law.[1] The ongoing question is the extent to which courts and the Justice Department should look at the economic implications of sheer size.
If you just grow to being #1 without acquiring your big competitors, that's generally OK in the US. It's competition, not collusion. Google never acquired Yahoo or Bing. Facebook never acquired Myspace or Twitter. So there was no key merger to block.
European Union antitrust policy is based more on the goal of having several major competitors, so there's price competition. An EU study indicates that, below four competitors, the leaders tend to implicitly collude. But in industries with strong network effects, it's hard to have many big competitors.
A modern summary from a law firm: [2]
[1] https://en.wikipedia.org/wiki/United_States_antitrust_law
[2] https://www.arnoldporter.com/en/perspectives/publications/20...
> Facebook never acquired Myspace or Twitter
They did acquire Instagram & Whatsapp, which their internal communications show was to block competition of a credible threat despite clever legal maneuvering to have the FTC antitrust case dismissed.
This is also where a lot of it gets ticky tacky. The difference between being anti-competitive and acting competitively against a credible threat is often in the eye of the beholder.
Acting competitively is releasing a competing feature of your product (e.g. Reels vs TikTok). Anti-competitive is buying off the competition (buying Instagram). I don't see how this is a complicated definition.
There are plenty of acquisitions which I think are competitive, but the market conditions are the critical evaluation point.
Where this breaks down & this is something that the FTC needs to learn from the past (poor in retrospect) approvals and redress in the future is who the acquiring company is and how quickly they could bring something to scale which would stifle future competition.
When facebook acquired Instagram, it has ~900M MAUs and Instagram had ~30M MAUs (3-5M in just the last month or two before acquisition, accelerating very quickly). By facebook's own account their MAUs and growth rate was considered a threat, so they acquired them to remove the threat. Fast forward 14 years, Instagram has 3b MAUs and displacing them is going to be __very hard__.
If you were to apply that same retrospecitive lens to Google, they've been able to acquire Google Analytics, YouTube, DoubleClick, Android, Google Maps and others to take them to hyper scale, completely dominate a particular product space because of their existing massive user base.
Not an easy job for the FTC to get right, but (in my opinion), it is clear that the FTC have been acting on an incomplete understanding of how hyper scale companies operate, how they can influence a market with their user base etc.
> There are plenty of acquisitions which I think are competitive, but the market conditions are the critical evaluation point.
There is an easy circumstance when an acquisition is never "competitive" -- when the takeover target is a healthy business and the acquiring company is a major incumbent in the same market.
Conversely, something like Google buying Android isn't a merger problem. Google's market share in mobile phone operating systems at the time was essentially zero. They were buying it to get to market a little faster and it's very likely that if they'd built a new one instead of buying it, Android would have died and their thing would have ended up where Android is.
Because it wasn't Android's technology that won. Take the median OS and then make it free by funding its development with Google's search engine revenue and also give it free advertising from the largest advertising company. The OS isn't the reason that became one of the two major mobile platforms.
But the problem there not being the merger doesn't mean there's not a problem. It'd be the same problem if they'd developed it internally, because the issue is that it's "free" but it isn't. Anyone can "run Android" on their competing device but they can't then run their bank app because that requires attestation which requires Google. It's an antitrust problem that isn't a merger. Those are a thing too.
Help me understand how acquiring them is acting competitively? There's always grey area for sure, but it's hard for me to understand how that would be good for a competitive market.
Myspace didn't expand into new markets and got left behind. FB was legitimately worried they were going to be left behind too.
There's an alternate reality where Facebook fell off, and Instagram never monetized and died like the dozen other competitors it had.
If it was as easy as "industry consolidation is bad" we could write simple legislation to make it illegal for a company to buy another company - but there are too many counter-examples where the alternatives to acquisition is worse.
> There's an alternate reality where Facebook fell off, and Instagram never monetized and died like the dozen other competitors it had.
You make its sound like that sort of thing happening is a problem. In practice the only way they both fall off if is something else replaces them. If the something else is the size of Meta, that's the status quo. We're no worse off than what we got. Whereas if it was five other companies or a federated system etc. then we'd have more competition.
Meanwhile the more likely scenario is that Facebook would still have >100M users even if it had less than it does now, and Instagram would have figured out that "put ads on it" is a profitable business model and survived independently as well.
In any market there are an appropriate number of sellers. If you think about it in terms of balance equations there have to be exactly as many sellers as there are buyers at a given price and any want-to-be sellers who aren't competitive at that price are eventually forced out of the market through bankruptcy because they can't sell and run out of reserves to hold on. There are no guarantees on how many sellers a market wants. Some markets it is thousands, some tens, some just 1. The number also fluctuates continuously in ways too complicated to model.
Acquiring a competitor is acting competitively if the natural market number of sellers is N and there are currently N+1 sellers. Someone is going bust and it is better for the assets and whatnot to be reorganised quickly rather than waiting for bankruptcy to force everyone's hand and cause a disorderly change.
A fun lens to look at Animat's comment upthread is that if the EU was trying to force more sellers than a software market can handle, then we'd expect them all to be weak and the EU might be unable to compete in software because they'd be too small.
> If you think about it in terms of balance equations there have to be exactly as many sellers as there are buyers at a given price and any want-to-be sellers who aren't competitive at that price are eventually forced out of the market through bankruptcy because they can't sell and run out of reserves to hold on.
That's not how it works in practice.
In a market with low fixed costs, you can have an arbitrarily large number of sellers. Some markets have millions of sellers, billions if you count the labor market.
In a market dominated by fixed costs, larger sellers have an advantage, but even that doesn't imply that there is a specific "appropriate" number. If all of the sellers are the same size then none of them has a scale advantage over the others regardless of how many there are.
So we have to distinguish between two different things.
The first is, how many sellers can the market support? In other words, suppose it costs a billion dollars to enter the market, the maximum value customers place on the product would allow for a gross margin of $100, and there are a billion customers. Then the maximum viable number of competitors is 100 because otherwise they can't each get enough customers to recover their costs and go out of business. Whereas if there are 100M customers then it's only 10, and if there are 5M customers then it's 0 and nobody produces that product.
But this being a low single digit positive integer is very rare, because to get there you have to thread the needle exactly and find something that can only support 2 rather than 20 or 5000 or a billion, but also not get so tight that it's 0.9, because anything less than 1 is really 0.
Which brings us to the second thing. You're in the market that can support e.g. 100 suppliers. To support 100 suppliers they each need a gross margin of $100, and customers value the product in that amount when it's not available for less. But now suppose some of the sellers are willing to accept $10 margins. At the lower price the market would only support 10 sellers; they each need 10 times as many customers. At a margin of $1 it would only support one seller. But this is every market where the fixed costs aren't fully zero. And, of course, the fewer sellers there are the less inclined they are to accept the thinnest of margins, or the more likely they are to be operating a large inefficient bureaucracy that eats up the economies of scale through red tape etc.
Notice also that the "maximum" number isn't necessarily the optimal number -- the market could support 100, but you might be better off with 15, because then the margins can be $15 instead of $100, which might be worth it on a $1000 product, whereas trying to save the remaining 1.4% by going all the way to a monopoly is not likely to go in your favor.
Which brings us to the thing that most commonly happens in practice. The market can support 100 sellers but one or two of them have more capital, so they lower their prices temporarily until the others go out of business, then raise them back up again as long as no one new enters the market. Which doesn't work when the barriers to entry are low because then people constantly enter the market anyway, but does work when the barriers to entry are high. Which is why we need to keep them low or, where that isn't possible, need antitrust.
I simply don't see Instagram and Facebook as competitive products. You could look at the advertising side, but there are tons of advertising companies. Whatsapp is more directly competitive to Meta Messenger (though that product was pretty different when Meta acquired Whatsapp), but again, in the US, messaging apps are a dime a dozen.
If we are looking solely at antitrust, there are a lot of companies- and a lot of tech companies- that are far worse than Meta.
In practice it's hard to avoid having one, two, or three top players. The market tends to invite disruptors soon enough unless really underhanded things are done to keep them out, but in the meantime the situation can be unpleasant anyway. I'm not sure there is much of a fix, not least because it's difficult to define in law behaviors that are anti-competitive -- we all know them when we see them, but defining them ain't easy.
I definitely agree that the law was not written with clarity in mind. On the one hand, that gives flexibility to those companies which might try to skirt legislation through technicality, but it bites the other way too, in that the executive needs to be very willing to enforce and defend its enforcement.
I wholly disagree that it failed to achieve anything meaningfully as policy, though. Under Sherman AT&T was split, which, before the baby bells all merged back together to create the present oligopoly, provided a much more competitive landscape for the telecom industry. And who knows what we would have seen if the original Microsoft decision was held up calling for a break up, instead of the neutered decision we saw come out of their appeal.
ATT split itself so it wouldn't be bound by the government - they had a lot tech they wanted to put into use they weren't allowed to and they could pick and choose what to shed (like pots lines) that wasn't making as much money.
There is a surprisingly high amount of legislation that seems shockingly poorly written unless one thought that as a feature rather than an bug.
E.g., while the 25th amendment prohibits someone from being elected president more than twice, it does not prohibit succession into presidency, such as by becoming the Speaker of the House (which position does not require one being a US Representative in the House) and President and Vice President resigning. That may seem unlikely or an outlier but on a site like hackers news, I would assume that we would relish the outliers that have bedeviled our software career
>”do we actually feel strongly that breaking Microsoft's productivity suite from their OS division was really worth all the effort?”
The antitrust attention distracted Microsoft so badly in the early 2000s that it completely gimped their ability to leverage Internet Explorer’s dominance, thus opening the door for new browsers like Firefox (née Firebird) and Chrome.
So yes, it was actually critical in creating the conditions for an open Web 2.0.
It wasn’t just the US antitrust of the late 1990s, but the EU also had their own legal challenge that targeted IE specifically. This doesn’t get enough credit IMO — the combination was powerful enough to break Microsoft’s hold on the market. Of course SV libertarian types would prefer to write it out of history and say the market sorted this one out on its own.
(Kids these days don’t remember when Mac OS X shipped with Internet Explorer as the only browser…)
> The only practical way that countries have tackled monopolies has been to write legislation that specifically puts oversight on that industry.
I'm reminded of what that notorious old socialist Adam Smith said about this:
“People of the same trade seldom meet together, even for merriment and diversion, but the conversation ends in a conspiracy against the public, or in some contrivance to raise prices.”
Good to see continuing evidence that the supreme Court continues to be an injust violent tool against the American people, a sad perpetrated of plutocracy.
I wish we had statutory antitrust mechanisms instead. Get to be a certain size, you have to be a public company. Have a sustained supermajority of the market? You have to file a plan to split up.
This was explored theoretically a few times, but the problem is it's really hard for this not to become extremely anti-consumer.
iPhones dominate the cellphone market? You now have to figure out a way to split the manufacturing and selling of iPhones across multiple different entities. Consumers end up with a handful of worse products that all cost more.
When AT&T was split up, it was just turned into a bunch of smaller regional monopolies that were in a lot of ways worse. It also forced the closure of Bell Labs - one of the most important R&D powerhouses in the US. (Most of these regional monopolies would end up merging back into bigger monopolies over time anyway)
AT&T was much more heavily regulated to operate in the public interest in a way that iPhone manufacturers are not. It's not so clear-cut that the regional monopolies were worse because they were smaller, rather than because they were no longer as strictly regulated.
I'm not convinced. Apple + Google controlling 97% of the world wide mobile market and choosing for all users what they are allowed to run as well as choosing for all business who they are allowed to do business with and how much they'll be required to pay to do so is, IMO, the single worst anti-consumer issue.
The part i've not been able to wrangle is what to do about companies which are internationally monopolies? Couldn't they just pick the winners/losers of local companies and ensure their own supply chain is the one that survives? That seems worse than a national monopoly that is at least HQ'd inside your borders.
Companies are basically mini governments. They have leadership (CEO / President), board, etc. However they are they are not democratic. Corporate takeovers happen when an entity with lots of money buys your shares. As such, the rule of the jungle abounds, not the rule of "lets make good products" or "lets have self respect".
One thing I wish being a public company entailed was having a set of internal shares that are distributed only to the people working there, that allowed them to vote on board members / top leadership positions. For example, Amazon currently abuses its delivery staff, but I suspect this is a policy that is not endorsed by the majority of Amazon employees. In a democratic public company, there would be a feedback loop that enables the employees to have a voice.
I see this as a way of having built in regulation, where the government doesn't need to pass laws that are continuously becoming outdated but instead giving the people on the ground the ability to fight for their rights.
I don't see it as useful for small companies, but I agree that a large enough company should be forced to be public + democratic. Going from the rule of the rich to the rule of the people that work there.
To me you are making this sound a lot simpler than it actually is. Corporate takeovers are anything but a guarantee, and companies very often adopt anti-takeover measures, such as poison pills, dramatically increasing the acquisition cost (macaroni defense), and worse case scenario, soliciting another friendlier party to purchase instead (white knight). Also if it is large enough, governments do tend to get involved.
Also, in kind of an uno-reverso, Germany kind of did something simlar being described here, but in a much different fashion with Volkswagon: https://en.wikipedia.org/wiki/Volkswagen_Act which notably traded short term shareholder pressure for long term stability. It's had the unfortunate reality (hindsight of course) of being the very thing that may be leading to Volkswagons current severe economic crisis.
Side tangent, but I have to call uno reverse on the Volkswagen demise (its a lot more complicated than how they organized themselves). I attribute it in large part to the terrible handling of the Euro crisis, where tons of money was printed and given to banks to revitalize the economy, that then gave it to big business, who then looked out their window and saw a bunch of poor people, and consequently bought back their own shares instead of investing in new technology.
To address your point I agree that it is complicated, democracy in general is complicated, but I think its worth it to try.
I also think that the amount of defenses a company needs to have ready to repel a takeover in my opinion strengthens the case for internal shares: if you want to take over, do it from within, convince the people on the ground, instead of rapaciously as an outside force.
I believe that the last group that tried to bring democracy to companies were... uh... after the means of production.
It sounds great in theory, but how do you deal with new hires? If you don't give them a vote, then it's slavery. If you do give them a vote, then how do you decide who to hire? If it's up to HR, how frequently do you vote for HR? Do new hires get to vote for HR?
I agree that this would cause a reckoning in who has oversight over hiring, as it dramatically shifts the balance of power in a different direction than the one companies are currently organized around.
I think for large enough businesses this would incentivize them give the team the person is hiring into much more oversight into the process. I don't think HR is useless, far from it, but incentivizing them to point their efforts to benefit the teams they are hiring for is a plus in my book.
As for the question of new hires getting to vote I think they should be able to vote (past standard probationary period of no more than say 1 month, idk). The company hired them, they should trust their employees with the responsibility of voting. If one person is able to upend a company then they should take some time to reflect on their processes.
The government had the power to approve or block mergers of a certain size but the courts in their infinite wisdom decided the only thing that matters is if the companies involved claim it will lower prices for consumers if they merge (and no longer have a competitor to fight on price...). And also the courts seem completely unable to find a way to call bullshit and learn from the history of every other merger where prices inevitably go up (because they're under less competition).
Indeed, this was a neoliberal shift pushed by the Chicago school economics.
The neoliberals in government launched a nationwide campaign to educate judges that redefined antitrust from being about power (anything too big is bad) to be about demonstrable consumer harm (much harder to prove).
https://en.wikipedia.org/wiki/United_States_antitrust_law#Ri...
> The model that underlies Claude Fable 5.2 and Claude Mythos 5.2 is a step forward in general capabilities. However, during the model’s training, we took steps to limit advances in the specific research capabilities that could most directly accelerate AI development. Thus, in relevant areas, including accelerator kernel optimization and improving large language model architectures, Fable 5.2 and Claude Mythos 5.2 perform at about the level of Opus 5.5. Importantly, we ensured that the training changes did not affect related capabilities like agentic coding, hardware development, and traditional AI or machine learning research (including interpretability research). We do not alter these capabilities as the model is being served to users (that is, at inference time). As described in Section 1.5 of the System Card and similarly to previous models, Fable 5.2 and Claude Mythos 5.2 also include transparent fallbacks to Opus 5 for a small set of topics in this domain, including kernel development on certain nonstandard accelerators.
-----
> Be it enacted by the Senate and House of Representatives of the United States of America in Congress assembled,
> Sec. 1. Every contract, combination in the form of trust or other- wise, or conspiracy, in restraint of trade or commerce among the several States, or with foreign nations, is hereby declared to be illegal. Every person who shall make any such contract or engage in any such combination or conspiracy, shall be deemed guilty of a misdemeanor, and, on conviction thereof, shall be punished by fine not exceeding five thousand dollars, or by imprisonment not exceeding one year, or by both said punishments, at the discretion of the court.
> Sec. 2. Every person who shall monopolize, or attempt to monopolize, or combine or conspire with any other person or persons, to monopolize any part of the trade or commerce among the several States, or with foreign nations, shall be deemed guilty of a misdemeanor, and, on conviction thereof; shall be punished by fine not exceeding five thousand dollars, or by imprisonment not exceeding one year, or by both said punishments, in the discretion of the court.
The Anti-Trust act makes it illegal for business to conspire to restrict markets.
Anthropic is engaging in some product engineering to level set their offerings against each other.
I'm not sure I am understanding the connection you are trying to make.
The implication of the comment is that Anthropic is limiting the capabilities of their external models to prevent other companies from building models that can compete with Claude, and that this constitutes anticompetitive behavior under the law.
> Anthropic is limiting the capabilities of their external models to prevent other companies from building models that can compete with Claude.
Uhhhh... that seems like quite a logical leap. "The only reason Apple doesn't release a faster phone is so other companies can't copy it."
Antitrust generally does not compel companies to make product lineup decisions, especially if the industry itself is acting quite competitively.
Anthropic's "Pace the frontier" is a facially unlawful conspiracy to restrict markets. Imposing these restrictions in their product offerings is an overt act that they're taking to facilitate their unlawful market rigging conspiracy. (since their agreement with their conspirators won't have a full impact if upstarts or direct consumers can just take existing SOTA and use it to produce a competing product, routing around their oligopoly).
There is no legitimate purpose to restrict the ability to optimize accelerator kernels except to inhibit competition in this space-- it is the most benign and harmless act imaginable. There is at least some debate to be had about things like computer-security capabilities, but nothing is being damaged by AI inference being made 10% faster except the ability of market leaders to extract rents.
1. Anthropic is not coordinating with competitors so there's no "conspiracy"
2. Anthropic doesn't have much market power. The competition is intense.
3. Anthropic doesn't have a legal requirement to provide it's strongest internal model to competitors.
>...shall be punished by fine not exceeding five thousand dollars, or by imprisonment not exceeding one year, or by both said punishments, in the discretion of the court.
Wow, I believe violating Section 1201 of the DMCA carries a stiffer penalty these days.
In 2026 dollars, $5,000 1890 would be around $185,131.87. I guess they didn't foresee inflation working against the fine.
> I guess they didn't foresee
Our laws in a nutshell
That quote is from the original 1890 text. The statute has been amended and criminal violations are now felonies with bigger penalties including $100 million fines and 10 years in jail.
That was for a different, less civilized time full of very rich people, so called "robber barons".
Didn't they have the cool mustaches that you can twirl around and look cool?
I guess we are going back to that era with a different flavour.
An oligarch by any other name, something something...
> punished by fine not exceeding five thousand dollars, or by imprisonment not exceeding one year, or by both said punishments, in the discretion of the court.
Something tells me that's not gonna do much to dissuade the AI cartels...
Please add a closing parenthesis. Its absence is driving this Scheme hacker bananas.