site banner
PaperclipPerfector  5d ago (text post)   18342 thread views

Culture War Roundup for the week of June 8, 2026

This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.

Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.

We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:

  • Shaming.

  • Attempting to 'build consensus' or enforce ideological conformity.

  • Making sweeping generalizations to vilify a group you dislike.

  • Recruiting for a cause.

  • Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.

In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:

  • Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.

  • Be as precise and charitable as you can. Don't paraphrase unflatteringly.

  • Don't imply that someone said something they did not say, even if you think it follows from what they said.

  • Write like everyone is reading and you want them to be included in the discussion.

On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.

4
Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

Statement on the US government directive to suspend access to Fable 5 and Mythos 5.

A few immediate thoughts:

For one strain of doomer, this is a good thing: the USG showing willingness to directly intervene and cut down the addressable market for capability-pushing models reduces economic incentives to push the frontier at all costs and slows down the race dynamic. Correspondingly, it's going to be a nightmare for the markets: the AI boom is driven by the idea that frontier models are going to be replacing a significant chunk of global labor, and obviously now this is significantly less likely now that frontier models are going to be stuck in Uncle Sam's basement.

For another strain of doomer, this is a horrible thing: really this was fairly clear even with the DOD conflict, but now control of AI is now very firmly a direct White House concern - the fate of the planet and/or the universe is in the end going to be decided by DC bureaucrats and not SF tech nerds.

Companies, especially non-US companies, are going to be rushing to the doors to move their AI workflows off closed-source and onto self-hosted models, another blow for the AI market thesis - the business continuity risks of Anthropic's demonstrated willingness to silently cripple models and USG's demonstrated willingness to now arbitrarily cut off access are simply going to be too much.

I have nothing new to say about the ban itself, or to be more accurate, I can't be arsed to.

I did use Fable from the moment it was available, quite intensively, and I can promise you that it's a feel-the-AGI moment. Is it an AGI? Nope. But it only took a little use to realize that it was clearly ahead of the pack, and we're only a small n number of iterations away. The only reason I'm not updating my timelines harder is because something like this is priced in.

Speaking of price, Fable tempts me to buy a Max plan. That is not something even a large amount of tokens for Opus 4.6 and later managed. I'll do so if this kerfuffle sorts itself out, but until then, using Opus 4.8 feels like a Flowers for Algernon moment.

I'm curious to hear your Fable use cases. In the brief time it was available, I didn't get to do much. The only project I had on my plate was a simple web application, which it pumped it in a way that felt no better/worse than if I had Opus. What sort of task would you recommend for Fable that you wouldn't for Opus?

I set it to the hardest task there is: understanding me. It did very well. As well as me. I want so badly to be the psychiatrist that Anthropic hired to analyze Mythos, as detailed in the system card. Probably can't be that person, not enough time. But Fable is a very good psychiatrist. Once the regulatory and legal barriers fall, I'm hoping I have enough money to enjoy the fireworks from a safe distance.

Opus is very good. But I wouldn't use it for anything at all, if Fable was available.

Battles like this are a good test of AI's power, since if and when AI ever became sufficiently intelligent and powerful, one could tell it "{model name}, I would like you to run a campaign to politically destroy {list of politician names}. Assume that you have {number} dollars of budget and the ability to delegate tasks to humans." and have a reasonable hope of success, at least assuming that those politicians weren't running their own AI to keep themselves in power.

the business continuity risks of Anthropic's demonstrated willingness to silently cripple models and USG's demonstrated willingness to now arbitrarily cut off access are simply going to be too much.

In the long term, maybe, but in the short term, I don't know that this is bad for Anthropic. Mythos first entered the conversation with the press release about all the vulnerabilities it found and how they're too scared to release the model to the public. This prompted more criticism than these companies normally receive, with several commentators pointing out how the claims were exaggerated and that they were just repeats of similar claims about products that were released and didn't end up being dangerous at all. So now they have the US government saying that this product is too dangerous for the public. Never mind that this is the same government that unsuccessfully tried to tag Anthropic as security risk because they didn't like purported limitations on the use of their software, but never mind that. Also, these names are a little too on the nose. Mythos suggests a product that's meant to inspire but doesn't actually exist. Fable suggests something mean to impart a moral lesson but is likewise nonexistent. The only similarly named product I can think of is the Mitsubishi Mirage, which feels like a similar thought exercise in how much you can cut from a vehicle and still be permitted to import it into the US.

I haven’t been able to figure out what authority the feds are using here, but if it’s ITAR it gets rough quick.

It's still a real hassle getting something as simple as a scope for a hunting rifle from the US to Canada -- which IIRC is ITAR (the hassle is on the US side, and there were some big fines for exporters before people started paying attention), and has been the case for a number of years.

Certainly an actual AGI would be a bigger threat to US national security than a nice cheap scope from SFRC, so I guess it makes sense?

a scope for a hunting rifle

This is more a problem with US exporters not bothering to keep up to date on ITAR reforms (there have been a few that reclassified most of the stuff into a much laxer category) than anything else, though.

Perhaps something an AI could solve for them, but their market is large enough that they don't have to care.

I don't think the full info is out yet. Axios has one news article with a bit more (found via Zvi), but no specific law is named.

I’m not saying that I have faith in the government here, but it is shocking that anyone might find “self regulating” private SF company of weird tech elitists to be a superior path.

I recall once Rush on a monologue saying that global warming can’t possibly be real the way the left says because if it was true, it would require government intervention in a way that undermined capitalism and he couldn’t accept that.

I remember being just shocked at the backwards reasoning. Obviously there’s a true point about backwards incentives reducing trust in the claim. But Rush himself was just openly engaging in the reverse mindset.

Here we have the same people saying that this threat exists also not wanting to have to overturn their grip on power over it.

I remember being just shocked at the backwards reasoning. Obviously there’s a true point about backwards incentives reducing trust in the claim. But Rush himself was just openly engaging in the reverse mindset.

There's also a slight utility maximizing thing here, similar to what I usually use to dismiss claims of us being simulations. Information only matters in-so-far as it can increase your expected utility.

Suppose we might live in two possible world: with probability p we live in world A which is normal and you have agency. With probability q we live in world B which temporarily looks like world A but in a couple years will collapse and nothing you do matters. Either everyone dies, or AI takes over and does whatever it wants good or bad in a way that you personally can't control, or it's a simulation and the matrix lords do whatever they want.

For any strategy you employ now before knowing which world you live in, the probably q does not matter. If your utility in world B is approximately constant with respect to your strategy, then you might as well act as if you're in world A and maximize your utility within that world. If there's a 40% chance we're all going to die in a few years and it either happens or doesn't outside your control, then might as well live for that 60%. I think this is a steelmanned version of the backwards reasoning. If you believe in capitalism so strongly that you think communism is equivalent to the end of civilization, or would kill you personally, or just be so bad that life is not worth living anymore, or would be equally incapable of solving global warming, then you might as well act as if you live in the world where it's not too bad for capitalism to solve.

Note that epistemically this is different from it actually affecting the probabilities p and q. It doesn't actually change how bad global warming is or how bad you ought to literally expect it to be. Just how bad it would be rational to treat it. If you are not a hero with a very high locus of control and are not going to step up and save the world from existential threats, then you might as well go about your life as if they don't exist and not worry too much about them.

Apropos of that old cyotography algorithm as a book first amendment case, absolutely nothing prevents Anthropic from publishing and open sourcing Mythos 5 weights as a book available world over to strike back at the government. Would be a totally glorious act of self immolation, think Denethor's pyre etc. as China and co are brought up to the exact same level of AI tech as the US in one single, simple, beautiful act of spite.

For the size of models we're talking, I don't think "books" is the right metric. The PGP code in the book was 350kB or so, and 950 pages. Models here are many GB. Even with some optimization, in human readable form it'd be somewhere between a bookshelf and a whole library.

Plus the weights aren't really human readable to begin with, and it's unclear if it'd be a "creative work" as much as C source code. I suspect models, the result of a purely mechanical, if expensive, optimization process aren't "creative" enough to be copyrightable, although that wouldn't stop printing books necessarily.

That is based on the assumption that Anthropic have principles. Which the real world experience doesn't show it is true.

What leads you to believe that Anthropic's stated principles are compatible with speeding up Chinese progress to AGI?

This is honestly hilarious in the context of Anthropic's own writing from a few weeks ago.

When a model poses risks of this kind, the government should have the legal authority to block or deter its deployment—beyond what exists in current law or in existing proposals in Congress.

It looks like they're getting what they asked for, and they're getting it good and hard. The admin simply took them at their word.

Did you read to the end?

As we have stated publicly, we believe the government should have the ability to block unsafe deployments, as part of a statutory process that is transparent, fair, clear, and grounded in technical facts. This action does not adhere to those principles.

Lots to unpack here.

At the philosophical level, if we don't have the state capacity to implement a regulatory regime that meets Anthropic's specifications, then why is Anthropic training and releasing frontier models? It reads like they are saying, "here are our demands for submitting to the authority of your government". That's not how it works. You submit to the government or you get shut down, like the rest of us.

At the operational level, I don't know how possible it would be to implement a regime that is "transparent, fair, clear, and grounded in technical facts," for regulating capabilities that don't exist yet. How can it be fair if we don't know yet what the relevant metrics are? How can it be clear if we don't know what red flags to look for? How can it be grounded in technical facts if nobody knows how these models work?

It reads like they are saying, "here are our demands for submitting to the authority of your government".

Did you read to the start?

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.

They are submitting to the authority. We can't get anything perfect (and it remains to be seen whether we can do anything good enough), but surely we can do better than this??

They have submitted to the authority of the government. They’re arguing that the government’s demands should be reasonable, untargeted, and not sprung on us out of the blue.

this is significantly less likely now that frontier models are going to be stuck in Uncle Sam's basement

I think that Anthropic is being targeted, not AI generally. The Trump administration does not like Anthropic for obvious cultural reasons and a reasonable suspicion that if Anthropic got hard power via AI, they'd use it to crush Trump and gang. Trump faction wouldn't treat Google, Microsoft or Amazon this way, there are lots of cozy interrelations there. The employees might not like Trump but they'll still help fund his ballroom. Recall the OpenAI, Facebook, Palantir executives getting sworn in as lieutenant colonels in the US military. Or Nvidia getting away with far more grievous harm to US interests via looking the other way with smuggling of their chips to China, lobbying against export controls, failing to . But Jensen knows how to court Trump and Anthropic doesn't. Jensen also has stocks (the biggest and the best!), so it's be costly for the avaricious elements of the US government to harm Nvidia's interests no matter how strategically obvious it is. You just don't sell high tech capital goods to strategic rivals, nevermind if it's 2 years behind or not! Anthropic stock is owned by VCs, not senators and elites or anyone who enjoys an index fund. Anthropic have no such financial pull factor.

But also, I'm reminded of how the big players in nuclear research stopped publishing openly in the 1940s...

Dario is just a very autistic guy who seems to have some California / Obama libleft views on account of having grown up in SF etc, and he can’t mediate them around Trump. If you read / watch interviews with him this is immediately obvious. Musk and even Altman historically were very scared of AGI but around Trump they’ll say it’s amazing and will create millions of jobs and secure American dominance forever. Amodei is the kind of guy who says he’s really afraid of ASI and bores the president by trying to explain foom over a Mar a Lago table.

Altman is clearly a mostly skilled operater who will say what he needs to, Musk is often foolish with his words but the right likes Twitter and Trump respects his wealth, and the other big tech companies have been around long enough to play Washington like you said.

I think that Musk and Altman have just convinced themselves that throwing the dice on ASI is worth it for them personally. They might not even be wrong there, putting the creators of ASI into the torment nexus even if ASI remains human-aligned as an acausal disincentive for recklessly betting humanity for personal gains is not widely explored as a strategy, sadly.

I think the USG is acting against Anthropic because Amodei is less of an idiot whisperer than is colleagues, but because his company pissed off the DoD by having limitations on use of their LLMs.

AI labs: "we are building a scary robot god who will destroy humanity and/or give its owner immense power"

US government: "ok, you are building a scary robot god who will destroy humanity and/or give its owner immense power"

AI labs: *shocked Pikachu face* *begins to cry and stammer incoherently*

This is just the obvious expected outcome. If frontier labs 1) continue to work towards the same goals that they have been loudly and publicly proclaiming for the last 4 years, and 2) get anywhere close to achieving those goals, then the traditional centers of kinetic power are going to take notice and step in and ensure that their interests aren't being violated. They're probably not going to just watch from the sidelines and twiddle their thumbs while Sam Altman merges with Skynet and becomes lord of the observable universe. They're especially not going to sit idly by while based Chinese hackers create an open source version so that you too can become a cyborg warrior who fights Sam Altman for control of the observable universe (which seems to be the tacit underlying assumption driving a lot of AI boosters). That's just fantasy. Other people (or machines) will have immense power, but it won't be you. You will be subservient to them. That's how it's always been and that's how it always will be.

the fate of the planet and/or the universe is in the end going to be decided by DC bureaucrats and not SF tech nerds.

SF tech nerds have desperately been in need of a wedgie for the last 7-8+ years. Regardless, this was always going to happen once the nerds got their hands on anything that was more than a plaything. I remember that meme about how the US was full of innovators who make things and Europe is full of lawyers and bureaucrats who only have it in them to regulate. In the end the "regulating" lawyers are going to win out over the "creating" "innovators", first internally and then world over.

Not really because American AI can’t be walled off from Europe without (at the very least) shutting down the internet, and probably world trade.

I think it can be pretty easily, Anthropic has been using Mythos internally for some time now.

I don’t mean defensively, I mean aggressively. European (and other) platforms can’t easily block Mythos from accessing them, which means true or x type risks remain global.

Ah, gotcha.

I...sort of hope the internet gets regionalized, to be quite honest.

Do you, though? Look at Chinese open-source AI. It's un-cucked (along the lines most of us here care about) because what the Chinese government really cares about you don't, and vice versa. Likewise Japanese manga and anime provided an escape hatch for every man sick of relentlessly feminist media. Things like Sci-Hub and Pirate Bay are based in the 'stans and the Netherlands respectively. Do you really want your countrymen to get final see on everything you're allowed to access?

Regionalizing the internet wouldn't shut down access to Chinese open-source AI or Japanese manga and anime.

Chef's kiss. The subtle deviousness of the USG demonstrated by taking Anthropic at their word. From what I see on HN - people are quite entertained by it. Seems that Dario has ground a lot of gears with his safetysm. And in a way - the USG is absolutely right. If the technology is as potent and dangerous as they claim - then it is not up to Anthropic to decide who can use it and for what.

Unfortunately the timing could be better - it should have been given just before their IPO is irreversible, but enough to tank them hard.

Is this related to the fact that it was Anthropic (at least I think it was them) who took a principled stand against the US gov using their models to automate kill chains and the like?

Yes, but it wasn't principled. It was to get cheap shot at openai and lure the antitrump SV to use Claude when Codex started seriously catching up with them.

This is nonsensical. Anthropic could have avoided signing up to supply the DoD altogether if they wanted to attract anti trump people. And OpenAI could have avoided walking into a pretty obvious trap.

Claude has been crushing GPT for enterprise use for way longer than this case as well

Almost certainly. More conspiratorially, it’s related to the discussions around government investment in OpenAI. But I think it’s punishment beatings.

Some Thoughts in a Conversational Style

  1. “It’s actually quite easy to block or pause AI development because the compute required (as seen in NVIDIA/TSMC/memory manufacturers/etc’ share prices) is so big that you can’t hide training a major frontier model”.

  2. “So what?”

  3. “This is actually bullish for humanity because it means some guy in a lab probably can’t build an AI that can destroy the human race without anyone noticing”

  4. “But wait, Deepseek and other models have shown that within just a few months of a frontier model’s release, vastly cheaper models can be trained to show similar performance. And even if inference is blocked, local models (which will always leak eventually even if they’re trained with a lot of compute) are probably no more than a year or two behind frontier models. So at best this is holding back the tide by what, months?”

  5. “Yeah, and in the current geopolitical situation, if US businesses complain that Chinese firms have access to more powerful models capable of more efficient and effective work than American businesses, even though America has better frontier labs, because those labs’ models are being blocked from public use by the US government, that will be quickly reversed”.

What is this place?

This website is a place for people who want to move past shady thinking and test their ideas in a court of people who don't all share the same biases. Our goal is to optimize for light, not heat; this is a group effort, and all commentators are asked to do their part.

The weekly Culture War threads host the most controversial topics and are the most visible aspect of The Motte. However, many other topics are appropriate here. We encourage people to post anything related to science, politics, or philosophy; if in doubt, post!

Check out The Vault for an archive of old quality posts. You are encouraged to crosspost these elsewhere.


Why are you called The Motte?

A motte is a stone keep on a raised earthwork common in early medieval fortifications. More pertinently, it's an element in a rhetorical move called a "Motte-and-Bailey", originally identified by philosopher Nicholas Shackel. It describes the tendency in discourse for people to move from a controversial but high value claim to a defensible but less exciting one upon any resistance to the former. He likens this to the medieval fortification, where a desirable land (the bailey) is abandoned when in danger for the more easily defended motte. In Shackel's words, "The Motte represents the defensible but undesired propositions to which one retreats when hard pressed."

On The Motte, always attempt to remain inside your defensible territory, even if you are not being pressed.


New post guidelines

If you're posting something that isn't related to the culture war, we encourage you to post a thread for it. A submission statement is highly appreciated, but isn't necessary for text posts or links to largely-text posts such as blogs or news articles; if we're unsure of the value of your post, we might remove it until you add a submission statement. A submission statement is required for non-text sources (videos, podcasts, images).

Culture war posts go in the culture war thread; all links must either include a submission statement or significant commentary. Bare links without those will be removed.

If in doubt, please post it!


Rules


Recommended Posts And Communities

Recommended Realtime Chats