domain:nature.com
DaseindustriesLtd
late version of a small language model
11m ago
In regard to Zhilin's words, if I may psychologize a little, I think that it's very natural for a Chinese person with close knowledge of and experience with Western ideas and societies - but also an attachment to an identity as Chinese - to conceptualize things in terms of a dichotomy between East and West
If that is how the Chinese actors themselves conceptualize this, does it matter if we can object to such thinking as historically reductionist and stereotypical? Yes, obviously both types exist in both societies. (At least Zhilins exist in the US; if there are Liangs, I'd be happy to see them. Lambert is an academic, not a hedge fund CEO who also somehow happens to be a great engineer and an open source fanatic. The closest we had was Emad Mostaque, neither exactly Western nor very technical or good at being a CEO). But it is clear that the Chinese discourse, particularly in the VC sphere, maps pragmatic and idealistic archetypes onto the East-West dichotomy. Half of Liang's interview is the journalist saying “but this is madness, nobody does it, nobody will give the money for it” and Liang saying “and yet we must learn to do it, because crazy Westerners do and that is how they create things we've been imitating all this time” .
But since it's not a matter of "deep roots," it makes sense that a single breakout success like DeepSeek could precipitate a shift in orientation. … To go far afield of my knowledge, it seems as though these extrinsic factors might end up being better for China than for the US.
I agree this is a possibility, and I think it's one of the more interesting cultural trends to track, which is why I'm writing these updates. Deep roots or not, Chinese fast-following is more than a subtly racist trope, it really is the backbone of their economic ascendance. If they start similarly rewarding high-risk innovation, it'll change the gameboard a great deal.
DaseindustriesLtd
late version of a small language model
22m ago
Well, I don't really understand American law but it seems to me that Anthropic has set the precedent of LLM pretraining corpora being essentially immune to copyright claims. Anthropic's models are, ironically, the most paranoid about reproducing copyrighted material.
OliveTapenade
24m ago
I'm not actually sure we are seeing a rise in more traditional, or 'high' forms of Christianity. There are some links here, but as far as I'm aware Orthodoxy is not growing in America in any particularly significant way, and the supposed trend of people converting to Catholicism is mostly a few high-profile examples, rather than a larger statistical trend. If you take away migration, the Catholic Church in America looks like a mainline Protestant church, like Episcopalians. Their retention is quite dismal, and to the extent that they've managed to hide that decline and retain political or social force, it's on the back of Catholic migrants. Evangelical Christianity remains the 'stickiest' form of Christianity in the US.
Now that said, raw numbers don't tell the whole story - the church that most successfully cultivates elites is not necessarily the one that will have the most social or political influence. The most visible example of this is probably the Supreme Court. Catholics utterly dominate the Supreme Court. At the moment it's six Catholics, two Protestants, and a Jew, and notably it has zero evangelicals. Congress also has slight Catholic overrepresentation, but it's much more marginal - 28% versus 20%. At any rate, it is possible that Catholics will become the de facto representatives of Christianity in the halls of elite power in America - the mainlines are collapsing, evangelicals are too plain and uncultured to ever get in there, and Orthodox are, with apologies, a rounding error.
There is potentially a discussion to be had about how Catholics got into that position, and I'd guess it has to do with the quite large and influential Catholic education system. (I also have a pet theory that religions that place a strong emphasis on the interpretation of law are naturally going to do better in terms of producing lawyers and judges; hence Catholic and Jewish overrepresentation on the Supreme Court, and I'd hazard a guess that Muslims will do pretty well too.) But that's something of a different subject.
Anyway, predictions...
I think Catholicism will not take over America or even necessarily grow that much from its current position, but I think it will get more politically influential, as it seizes ground that used to be held by mainlines.
Mainlines will continue to collapse. Some outward adherence to mainline churches will survive in places, among politicians, but the era of mainline dominance is over. In America more broadly I don't think mainlines will all die out, but they will need to reinvent themselves; I foresee conflicts like the like in the Methodist church, between theological progressives who see the church as handmaiden to preferred cultural causes, and cranky traditionalists, which will probably end with the former withering away and fading into culture, and the latter declining in numbers and turning into a small but devout rump.
Evangelicals will not advance much in terms of political power, but they are disorganised and in constant ferment and will remain a powerful voting base for politicians canny enough to appeal to them. That said, what appeals to them is somewhat unpredictable, as they are a fickle demographic that is highly responsive to charismatic leaders. Right now they are more-or-less solidly behind Trump, but they didn't come to support him for theological or doctrinal reasons, and I think Trump's successors may not necessarily inherit evangelical support. I'm really not sure which way they will go.
Orthodox are irrelevant. Again, apologies for being so blunt, but there are just far too few of them and I don't see any signs that will change.
Mormons are one that I predict will grow and increase in power. I think they have the most gravely mistaken theology of any of these groups, but even so, they are demographically healthy, expanding, and confident. They are currently adjacent to the big evangelical coalition and can sometimes be counted with it, but not consistently, and when you look under the surface there's a lot of submerged evangelical dislike of Mormons, so that may not be stable. I think they will grow in influence unless there is some kind of concerted effort to declare Mormons 'uncool', the same way that evangelicals are uncool, and keep them out of power that way.
As for other religions...
There aren't enough Muslims to be a very significant electoral demographic nationally, but there are towns and potentially states where the Muslim vote matters, so I expect to see local gains in influence for them without making a huge impact nationally. The big question I have with Muslims is whether American Muslims as a community hold on to traditional doctrines or become secularised; there are plenty of people for whom 'Muslim' is an ethnocultural identity but doesn't make moral demands or shape their moral or political thought. (Think e.g. Zohran Mamdani.) I expect a significant number to hold on and continue to practice. As mentioned above, I expect Muslims to do reasonably well in terms of elite roles, especially those to do with law.
Jews are, well, an invitation for certain people to come out of the woodwork and declare them the secret puppet masters of the US. I don't know the future of Jews in America. Until recently I would have said that America has been a very good home for Jews, and I expect American Jews to continue to prosper, but we have yet to see how much Israel/Gaza causes a realignment. This is definitely one to keep an eye on.
Hindus mostly can't be disentangled from Indian ethnic politics. (Sorry, ISKCON, you tried but there aren't enough of you.) I'll skip over that because it's much more to do with ethnicity and multiculturalism than it is about Hinduism as religious belief. Sikhs are in roughly the same camp.
Buddhists are a group that I expect to continue to grow, partly from immigration and partly from conversions, but to have practically zero political impact. Buddhist organisations, at least in the US, rarely mobilise for politics, and most converts practice on their own or in small groups without necessarily applying Buddhism to politics more widely. There aren't many of them anyway; Buddhists as a constituency is not worth pandering to. Maybe in some local contexts where there are heavily Buddhist migrant groups, but I doubt you'll get much more than politicians visiting a temple or dharma centre and saying they appreciate this group.
That's probably most of what matters. Scientologists are few and don't matter, Unitarian Universalists are few and don't matter...
DaseindustriesLtd
late version of a small language model
24m ago
I am skeptical about diffusion even for images and video, the whole subfield is a giant nerd snipe for mathcels. Autoregression is strictly more expressive and more suitable for continual generation, sadly we pay the price of worse parallelization. If anything, I'd be more enthusiastic about honest-to-God Energy-based LLMs. There have been a series of breakthroughs in making DLLMs that don't totally suck and offer extreme speeds in low batch size regime, but eh. I think sparse attention Transformers will crush them.
DaseindustriesLtd
late version of a small language model
30m ago
I literally cite Kimi's own arguments for open source:
[…] 3. Why Open Source
#1: Reputation. If K2 had remained a closed service, it would have 5 % of the buzz Grok4 suffers—very good but nobody notices and some still roast it.
#2: Community velocity. Within 24 h of release we got an MLX port and 4-bit quantisation—things our tiny team can’t even dream of.
#3: It sets a higher technical bar. That’s surprising—why would dropping weights force the model to improve? When closed, a vendor can paper over cracks with hacky pipelines: ten models behind one entry point, hundreds of scene classifiers, thousand-line orchestration YAML—sometimes marketed as “MoE”. Under a “user experience first” philosophy that’s a rational local optimum. But it’s not AGI. Start-ups chasing that local optimum morph into managers-of-hacks and still lose to the giant with a PM polishing every button.
Kimi the start-up cannot win that game. Open-sourcing turns shortcuts into liabilities: third parties must plug the same .safetensors into run_py() and get the paper numbers. You’re forced to make the model itself solid; the gimmicks die. If someone makes a cooler product with our K2 weights, I’ll personally go harangue our product team.
DeepSeek's arguments are more ideological and cultural:
For technologists, being followed is a great sense of accomplishment. In fact, open source is more of a cultural behavior than a commercial one. To give is to receive glory. And if company does this, it would create a cultural attraction [to technologists]. […]
plus stuff about accelerating the development of Chinese ecosystem.
High-level researchers are not slaves or menial workers, they have massive pride, they want to publish and gain clout. You can pay them hundreds of millions to get over that, or you can let them publish. Open sourcing is the ultimate form of publishing your work.
FtSoA
30m ago
the sock-puppeting of civil society, the media manipulation of public opinion in the interest of stability ?
I don't know how much of this I'd sign on for as you've characterized it. As a classic liberal/gray tribe rationalist type, I very much would prefer to see a return to limited government and greater economic freedom on many fronts. But most things aren't plots, and I usually take a very dim view of the MAGA approach, even when I agree there is a problem with Progressive Ideological Capture in any given institution. I'm a bit to Scott's right on several fronts in the Culture War, and I'm less EA-pilled. I generally agree with Garrett Jones about democracy and immigration.
In terms of foreign policy, I also have many gripes with both the Left and MAGA Right. For example, in my view we should use free trade agreements against China. And we should arm Ukraine and Israel to the teeth to unleash on Russian and Iran.
All of this is to say that, even with the faults of "the way liberal democracy functions in practice," we still have it pretty good in the US.
OliveTapenade
52m ago
It's possible he thinks that way, or even that he just thinks that owning the character and account is what matters. I suppose one could make a comparison here to his companies: he presumably thinks of himself as designing or making cars or rockets, even though almost all of that is done by lower-level employees. Likewise he may think of himself as playing PoE2 even though almost all of it is done by a lower-level contractor.
To be honest the impression I've gotten with regard to Musk and gaming is that he just doesn't understand how gaming works. That PoE2 YouTuber, as I recall, pointed out that what Musk claims to have done is mathematically impossible - he could not have reached that level in the game in the amount of time available. It's not doable. But I would not be surprised if Musk believes that sheer skill can accelerate one's progress in the game. Is it possible that he just doesn't understand how grinding works?
I suppose I think that he has very surface takes on games. I remember when he claimed that chess was too simple and Polytopia was better. Not only does that tell me that he doesn't know much about chess, it also tells me that he doesn't really know much about Polytopia, which is a quite simple 4X that can be mastered relatively quickly, and which did not hold much interest for serious 4X players. On the surface Polytopia looks more complex than chess, because it has more widgets to manipulate, but in terms of strategy it has less depth. What this tells me is that Musk probably played Polytopia for a few hours, maybe even a few dozen hours, but has never deeply familiarised himself with the genre.
I suspect that Musk finds the idea of gaming interesting, and is enchanted by the idea of being a hardcore gamer, but he is what we used to call a casual.
There's nothing wrong with being a casual. Casual gaming is a great way to spend your time. But a casual who wants to be seen as hardcore, doesn't have the skill, but does have the money... well, that's just cringeworthy.
gattsuru
1hr ago
Technically, two furry versions, though you have to go into settings for Bad Rudi. Tbf, they're both obnoxiously monofocused and pretty lackluster when it comes to animations or gameification; Rudi on telling 'cool' stories, and Bad Rudi just trying to swear at you (cw: exactly what I said, loud sound).
But, yeah, it probably says a lot of strategic things.
OliveTapenade
1hr ago
I can understand that case - part of what makes Musk willing to be daring and innovative in business is also what makes him willing to do bizarre and eccentric things in other fields. Having enough ego to disregard the advice of everybody else in terms of what's possible for rockets or electric cars probably goes with having enough ego to, well, do these other absurd things. So you've got to tolerate a bit of weird gamer nonsense as the price for all these other benefits.
I suspect that overall we disagree about the net value of Musk's contributions to society, or about the desirability of things like AI girlfriends or artificial companions more generally. I'm quite pessimistic about AI in general, so I consider it preferable to maintain as large a taboo as possible against using AI for social purposes. If there is a respectability cascade that results in the public considering AI girlfriends/boyfriends to be legitimate or healthy ways to spend one's time, I would consider that a negative development. But we may have different high-level generators of disagreement on this issue.
Hieronymus
1hr ago
The Reader’s Digest condensed version: The Old Testament ritual for purifying Jewish priests to serve in the temple requires the ashes of a spotless red heifer. Rabbinical tradition adds a bunch of criteria to the biblical law (as rabbinical tradition is wont to do) such that qualifying cows are absurdly rare.
Some Jews who want to restore the temple would like to breed qualifying cattle. A few eccentric dispensationalist Christians, who believe that the rebuilding of the Jewish temple is part of the unfolding of biblical prophecy, want to help them. This isn’t a common thing, but it has geopolitical relevance, as rebuilding the Jewish temple would require tearing down the Al Aqsa mosque.
OliveTapenade
1hr ago
The tennis example strikes me as absurd and lacking in dignity for either Ackman or the tournament, but the presence of a substantial benefit to the tournament does change the calculation a bit for me. The tournament has traded part of its credibility for a large payment. Depending on the tournament's finances, that might have been a worthwhile trade for them, but it's still undoubtedly sordid.
Sunshine
1hr ago
If you observe a person's number of teeth over time, you will find that they start with 0, rise to a certain peak, wobble a bit as baby teeth are lost and replaced, and then stay mostly stable at a certain number for a long time (possibly losing a few to accidents). Then, at the end of their life, they gradually lose teeth their one at a time until they have none.
If you were to look at a snapshot of a community, you would find that some children are gaining teeth quite quickly, the adults have a stable number of teeth, and the elders are gradually losing teeth. In a community that is 50% children (not rare historically) you might take an average and find that the number of teeth in the community is rising rapidly. However, if you were to extrapolate that to assume that the community's children are mutating into shark-like creatures who constantly grow more teeth, you would be making a mistake. In the long run, everyone ends up with exactly 0 teeth.
The mature civilizations of this planet are becoming less religious. It would be a mistake to assume the immature civilizations will continue their current trend lines exactly. It is better to assume they will follow the same course of evolution the more-developed civilizations took. The more-developed civilizations are becoming less religious over time, and unlike with children there are no new undeveloped civilizations rising up. It is reasonable to assume that, if things continue on as they have for another 100 years, secularism will continue to rise.
ABigGuy4U
1hr ago
If the temple is going to be rebuilt, the site would need to be consecrated with the sacrifice of a red heifer. It has to be a heifer that has no other colored hairs at all. That is very rare. Ranchers in Texas are working with ultra-orthodox groups in Israel to breed a line of such cows to have it ready in case the Dome of the Rock suddenly....goes away.
RandomRanger
Just build nuclear plants!
1hr ago
They're invertebrates ping-ponging around inside this closed space of legal 'rules', penned in by judicial review and the ECHR. They don't realise that they have the power and the duty to write and rewrite the rules as necessary to achieve outcomes. They're too wedded to the 'rule of law' now, they don't realise that it's truly just a social construct.
Quantumfreakonomics
1hr ago
>He doesn't know about the red heifer
shoeonfoot
1hr ago
Sorry to resurrect this thread but I've only now gotten a chance to read this post and all the responses carefully. I'm curious what you make of this thread. I'll copy it below in case you don't want to click the link.
Why the Epstein story matters so deeply to the political right—and why sweeping it under the rug is not just offensive, but a civilizational betrayal:
This isn’t just about Epstein. It’s about what his case reveals: a nexus of unaccountable power, intelligence cover, institutional rot, and elite impunity. The story touches every nerve the American right has been warning about for a century.
Since FDR, the right has feared the unchecked expansion of the administrative state. But the real danger wasn’t just bureaucracy—it was the fusion of that bureaucracy with the intelligence community, financial elites, and transnational interests.
Epstein is the singular window into this world. A man with no clear source of wealth, deep ties to U.S. and foreign intelligence, and access to the most powerful people in the world—running a blackmail operation under institutional protection.
The CIA won’t talk. The FBI walked away. The media refused to dig. And Israel—whose alleged involvement through cutouts like Wexner is whispered about but never investigated—remains off limits. That silence says more than any report ever could.
For decades, the right has asked: Who really governs? Who watches the watchers? Epstein gave us a glimpse. And what we saw was not a “conspiracy theory”—it was conspiratorial governance: intelligence services operating with total impunity.
This isn’t just about criminal sexual behavior—though the abuse of underage girls is itself an unspeakable crime, and one that demands real justice. But the fact that such crimes were instrumentalized for power is what makes this even more sinister.
The use of sexual blackmail to compromise institutions and shield a network of elites is not a subplot—it’s the playbook. This was kompromat as statecraft, and it operated in the open, protected by the very agencies tasked with protecting us.
The reason the Epstein story haunts the right is that it confirms our deepest suspicions: —Our intelligence agencies are political actors. —Our elites are compromised. —Our allies are unaccountable. —And our institutions lie to preserve their power.
Worse still: every time the Epstein story is buried, the very institutions doing the burying destroy their own legitimacy. The cover-up corrodes the foundation they claim to defend—rule of law, transparency, democratic accountability.
This is what Eisenhower warned of—not just a “military-industrial complex,” but the seamless merger of state power, private capital, and foreign intelligence. Epstein is a grotesque fruit of that fusion. Ignoring it won’t make it go away.
The right sees Epstein not as an aberration, but a revelation. A moment when the mask slipped. When the postwar liberal order—underwritten by secrecy, mutual blackmail, and “strategic alliances”—showed its true face.
So no—we won’t move on. Not because we’re obsessed with scandal, but because the Epstein case is the Rosetta Stone for understanding the modern American regime. And the regime knows it.
That’s why it must be buried. That’s why we must never let it be.
Amadan
Enjoying my short-lived victory
1hr ago
I can already see the framework, the jewish tricks are practically manifesting before my eyes:
This is an odd and derogatory thing to drop into the middle of your post. I am not quite sure what you're getting at, or if this was meant as some kind of ironic joke that I missed, but you seem to be playing on the trope of "Jews are responsible for everything related to social degeneracy and porn." You're either failing to speak clearly, or if you really want to pin this, of all things, on the Jews, you need to provide some kind of evidence for the claim that "Jews" are behind this.
Amadan
Enjoying my short-lived victory
1hr ago
... is it more hilarious or disturbing that they are making the furry version available before the male human version?
Quantumfreakonomics
2hr ago
On the Protestant end, the number of things that are “demonic” are growing really fast.
This is in contrast to the Catholic view, in which everything is presumed demonic until exorcised (why do you think they bless things so much?)
shoeonfoot
2hr ago
Buddha statues you can put into your garden
Note that is frowned upon by most genuine Buddhists.
Trump was able to win over a lot more Catholics than he did in 2016 and 2020. In fact, a majority Catholics voted for Clinton and Biden, but then swung like +10 in favor of Trump. I'm not sure what effect this had on Trump winning as many states as he did, but if it was relevant, a Republican will have a harder time winning in 2028 if the Catholic vote swings back to D.
I'm curious to see if organized Christianity adopts a more hardline position on immigration. What I'm increasingly seeing among the Online Right, especially after Epstein, is that the only thing that will salvage the Trump administration is mass deportations, and their attempt to synthesize Christianity with this goal.
TIRM
2hr ago
It's not clear any company running this sort of thing can seriously prevent leaks over a long enough time for it to be relevant.
With sufficient will, they could do just this. This is a choice they actively make one way or another.
The_Nybbler
In the game of roller derby, women aren't just the opposing team; they're the ball.
2hr ago
There's real technical ability in lawyering and banking too, but it's always engineering that gets singled out for treating women wrong. Somehow or another (male) engineers are the worst of the worst in terms of oppressing women.
Or... the narrative is completely wrong. Engineering gets singled out because women prefer the bros to the nerds.
The fact that Grok is at all comparable (or indeed inferior) to Kimi on any metric, even the most obscure one, speaks to the deep cultural advantage of Moonshot. Grok 4's training compute is estimated to be 6.4e26 FLOPs; Kimi, like R1, is likely ≈4.0e24, fully 100 times less. They probably spent on scaling experiments for Grok 3/4 more than Moonshot has spent over their lifetime on everything. It's not really a fair competition, I admit Grok is a stronger model.
I think it wasn't designed with any specific focus in mind, it's an all around next-generation base+RL model.
You distort my argument. I was consistently skeptical that China can win this on HBD merits alone, after all the US also has plenty of talented people (very many of them Chinese, but also diverse global and domestic talent), in Nvidia and elsewhere, plus it has a giant and growing edge in compute. My thesis is that the gap in applied AI possibly won't be so profound as to allow some Pivotal Action To Secure Durable Strategic Advantage, that the hawks in DC and Dario Amodei fantasize about as they rail for more export controls. Nvidia will remain dominant, so will Western AI labs.
But so far China is doing better than I anticipated, both technically and ethically.
More options
Context Copy link