> So in that way, the Deepseek models are heavily subsidized as well, with the possibility of them being served by a different company that paid nothing to develop them.

Has anyone seen the neo-cloud profit margin on deepseek? I wonder if the deepseek served API prices account for the training cost? Because they don't / have not raised the money to fund their future operation / training- and they depend on that cashflow for now? That would suggest a really high profit margin on inference only.

I actually don't think this is so deep- you couldn't curse on broadcast TV and basically because Instagram is an ad platform you aren't supposed to say "bad" things on there either.

There's been endless debates here about moderation vs. censorship- on the spectrum of Hackernews to 4chan. Calling the Hackernews moderation rules "censorship" would also not be reasonable- nor do people want to visit 4chan because it's "free speech".

I think it's actually a really nice workaround that human language and english is flexible enough that we can still say exactly what we mean in a different way and everyone knows what we are talking about.

I fundamentally don't think the platforms want to get rid of the engagement these topics bring, (the last round of criticism being that they want divisive topics on the platform) and if and when they do it won't look like changing "kill" to "unalive"- the algorithm will simply be directed to silence the topic. Many such examples in China. We should be angry about the censorship we don't see and can't detect as easily instead.

I agree. The censorship I'm way more concerned about is the government flexing on platforms, since that's actually about suppressing ideas. Also think the author is missing that a lot of kids using internet slang irl are doing it to be funny.

It states in the article *[our lawyer] advised us that [...] the judges would be extremely deferential to the Rodin Museum and would grant it the benefit of any doubts* but doesn't dare to say why they would be deferential.... It's probably because these high level cultural institutions basically get to do whatever they want... France definitely loves their institutions.

I think the part that stands out is how at least one court went out of its way to invent a new argument on behalf of the museum. I was surprised to learn this is actually somewhat common in French law. The court itself, at least with cases involving administrative actions, has every right to step in like this. It is not a neutral arbiter.

After reading the article, my interpretation isn't that they put an enormous legal effort into combating the release of the scans, it's that they put very little work into fighting against the OP, made a lot of bad arguments and knew from the start they would get away with it.

It comes off a little bit like they made a lazy decision at the outset, didn't want to reverse course to save face and knew they didn't need to back it up with anything....

I would go as far as to say the museum was negligent (by their own admission) and lazy, assuming (rightly so) the judicial system would protect them. Wild stuff.

IMO, their collected filings, both in the Paris tribunal and in the Conseil d'État, are a true trainwreck. I don't discount the possibility that they were lazy (it seems likely) but I also think they may be truly incompetent and genuinely indifferent to the law, and without a doubt they behaved and argued in bad faith throughout.

Well, a broader interpretation of this idea is that paradoxically there needs to be a certain level of authoritarianism in order to have a civil functioning republic.

In the 1900s this was communism. A lot of people (either disingenuously or not) interpreted this as anti-democratic. But if you look at what actually happened where conditions were right, it turned out to work well- China and Vietnam, and if you just look at authoritarianism, not just communist authoritarianism, then Singapore, S. Korea, Taiwan as well. Some would also argue that the USA falls into this category- that the same arm of the government suppressing corruption is also going after marginalized groups.

There aren't that many examples in history where the ideal balance is actually struck.

How long is the long run though? China doesn't have the chips, and probably won't have them for a while.

I agree in principle, but it could be more than 5 years, maybe 10. Who knows what things will look like then.

In a recent Dwarkesh podcast Dylan Patel breaks down how little compute the chinese labs actually have- not even the fact that they don't have access to new Nvidia chips and they're stealing them through shell companies- just that, even if they have cheap electricity, the compute just doesn't compare. Maybe even two orders of magnitude less. They couldn't get it even if they had the money. And if you look at how much more efficient newer chips are, that cuts the effective compute in half again. The conclusion was that they are at least 2-3 years behind.

For frontier labs the current compute seems to be driving model progress (in training) at least to some degree, even without true RSI, and this seems like it'll continue to keep any chinese model from drawing even with the frontier labs, at least for the foreseeable future.

Inevitably the chinese government will drive more funding in chip fab technology and the money will come around to build chinese data centers, but who knows how far off that is. A few different things in the tech tree need to fall into place. It doesn't seem like it'll be next year.

The counterpoint to that, though, is that the Chinese companies have to figure out how to be competitive, regardless of their significant compute deficit. And, as far as I can tell, they're actually doing that. They're trailing the frontiers in model effectiveness, but not by years. It's single digit months.

If there is no upper bound how how these things scale with compute, and if China does really begin to catch up to Nvidia (and they're probably not going to feel encumbered by US patents for domestic AI hardware, given how important AI seems to be to the Chinese government), there will come a day when China leapfrogs the US on AI.

I think on the timescale of 10 years, that's a super likely scenario. But will it be any sooner?

For instance a Chinese EUV machine seems like it's very far away. Even if they have (steal/borrow) the necessary IP.

One of the optimistic aspects of what's happening is that it doesn't seem like the leading model capabilities are outpacing the more open ones by that much- that just as a general principal there are pretty ok models relatively close behind the best ones.

This could mean the long term viability of my open claw bot to shop for me- that openai / google etc won't control AI, and I'll still just be able to use a model via API to do what's actually in my own self-interest.

What's crazy is that the ToS is basically, "allow us, Amazon, to give you algorithmically higher prices- and if you try to circumvent that, you can't buy from us anymore".

I think the arguments for these kinds of near-monopoly companies being true market-driven capitalism seem more and more false.

The idea that Amazon will tell you that you don't even own the right to your own shopping experience tells you everything you need to know. Apple owns your device and Amazon owns the right to show you whatever price/product they want, regarless of your search intent.

I sometimes look around at people proposing technical solutions for these political/legal/economic problems, and think: "Was I that naive X decades ago, or were things really different then?"

I didn't realize how much money Uber makes from ads.... Why does every business devolve into an ad platform?

Might be because most hit their maximum growth but need to keep growing indefinitely or risk becoming a “mature” company?

Because consumers don't care much about ads and prefer them to even minor cost differences.

Vomit. We should ban 90% of ads. They are mostly net negative on society.