deaux

> we have rule of law here

You're in the lead for funniest comment of the year award!

How's Ghislaine Maxwell doing nowadays? Last I heard, living a very comfy VIP "prison" life. How are her mates? Right, running free. Now let's look at how people jailed for marijuana possession are doing.

Everyone commits crimes, yet prosecution is hilariously selective. Everyone knows this. White collar crimes have a prosecution rate of <1%. Three felonies per day.

There is absolutely no meaningful rule of law in the US. The absurd degree of selective prosecution is antithetical to this.

That's a cop out. You can do both. Do you really think they haven't tried to fix the issues? There's no reason why they wouldn't. They try to fix the issues and rightly jail the CEO. To act as if he was just an unknowing cog in the system is nonsensical.

He was definitely planning on PII abuse at scale for personal gain from day one. Probably not planning mass surveillance, but from all we know of him, he wouldn't have disliked the idea.

There are dozens of these.

There are definitely other things that could happen, that would not make OP scream. Not by the justice department though, that's true.

It would take 1 week for everyone to move over to something else.

> but the world still needs the company for the time being.

You have been caught hook, line and sinker. Flush the koolaid out of your system. I can absolutely guarantee you that if all of Meta disappears tomorrow, everyone will move over to a like-for-like replacement within weeks. I'd be willing to bet everything I have on it.

The world absolutely does not need Meta, period.

The inability to personally jail megacorp CEOs who cause untold amounts of societal damage is not common "everywhere" ("many places", maybe). And that is the actual thing the person you're responding to cares about, and rightly so.

Flock doesn't just sell ALPRs. They bundle it with CCTV, audio sensors, sells to private parties, analyze visual data. That's why they're everywhere. Pointed at pools, walking paths, indoors. In countries where ALPRs are widespread, they're exclusively on roads, because.. they read license plates.

> and like I already said punishment is hard specifically because how do you fine someone with infinite money?

This isn't true though, if they had infinite money they wouldn't be spending every single second of their waking existence trying to get more of it and move higher on the leaderboard. If your opinion is that Meta/Zuck wouldn't care about being fined a year's worth of global revenue, or in his personal case, 90% of his net worth, then I'd really like to hear why. Because it contradicts all of their actual actions showing that all they care about is hoarding as much money as possible, meaning they'd very much care about it being taken away. Maybe I'm missing something.

My counterexamples prove exactly the point; TBTF is fugazi, arbitrary, something in your head.

I see this parroted a lot, yet have never seen a case where saying "not" to do something makes it more likely to do it, which is what you're implying by saying `It's also worth noting that saying "Don't cheat" just added "cheat" to the context`.

At worst, it gets ignored some of the time, it may even degrade output quality, but I've seen no evidence that it makes it more likely to do it.

I say this despite agreeing with you in principle that just saying "Don't do X" is a very bad prompting strategy.

China is indeed "bad" for the gestapo shit. The difference is that China's gestapo shit comes with benefits for the common man too, whereas in the US the gestapo shit only serves the inner circle at the top of the regime with zero benefits to 99% of the populace.

I do this and did it in this case, but I've never seen it make a difference.

It closes the development loop. Without it a model can't check if the stuff it made actually visually renders like it's supposed to. It can only guess/assume.

Sometimes you intentionally want to verbatim keep "mistakes", sometimes you don't and want them to be "fixed". OCR-only models tend to only do one of those two, in VLM cases often the latter. With multi-modal LLMs you can just tell them (adherence of course needing evals/differs per model).

The obvious use case, especially on HN, is frontend dev of any kind at all. The second most obvious one is OCR of paper documents.

In the sense that asparagus is an alternative to bananas, if that's what you want.

> Quite. And quite the opposite - I hear a lot of messaging encouraging us to consume more. To the point that most Westerners (and no doubt others) have a strong sense of entitlement when it comes to consumption. If you ever bring this sort of thing up with people the entitlement becomes obvious. We feel entitled to be transported thousands of miles in a matter of hours for 'a break'. We feel entitled to some of the most inconsiderate vehicles ever produced because "how else am I going to tow my boat/horse float?" As if a lifestyle that involves moving massive things around is a god-given right. And all of this within a matter of two or three generations. We truly live in the most entitled times.

Where do the people who understand this congregate?

> (and no doubt others)

While the top 30 on this entitlement ranking will contain a lot of Western countries, I can unfortunately confirm that they're far from alone, and in fact none of them will even take the top spot.

> It’s difficult to actually punish a machine that acts almost like infrastructure.

It's not at all. There is zero reason this couldn't be applied to Zuck. [0] There's also no reason why fines couldn't be 10% of global revenue, or more.

[0] https://www.nytimes.com/2026/08/20/business/evergrande-found...

TBTF is not a thing. It's all in your mind. Evergrande and Country Garden would've been classified TBTF in the minds of many USians, yet they failed.

The dead comment below by AnthonyMouse already pointed out that even in your scenario there's no difference with the status quo anyway. It's insane that it's dead and reeks of astroturfing, as nothing about it is worth flagging.

> Isn't that the other way around? The law already can't be enforced against the TBTF corporation, so what do they care if it stays on the books? But now you can't enforce it against someone else or they can point to you not enforcing it against them.

> Except that's the point. The government would now have to do that or they lose the law. If it's a good law then they don't want to lose it, but if it's a good law then they should be enforcing it against them.

> And, of course, the correct way to deal with this is to break the TBTF corporations into many smaller pieces so that stops being a problem.

Only EU providers, seemingly huge latency, >2x pricing is just in no way an alternative.

https://cortecs.ai/detailedServerlessView/deepseek-v4-flash-...

https://openrouter.ai/deepseek/deepseek-v4-flash-0731#provid...

Every other coding harness (i.e. 9 out of the top 10) automatically read AGENTS.md. It is a universal standard. Claude Code does not.

What's even weirder is that the "bcherny closed this as completed 1d ago" is missing after his comment, even though all other issues have it, e.g. https://github.com/anthropics/claude-code/issues/66955#event...

A permalink actually exists for the event in the post's issue too, but it leads to nothing: https://github.com/anthropics/claude-code/issues/6235#event-...

It was closed yesterday as "completed" by Anthropic's bcherny, despite not being implemented/completed, and being by far the most requested feature that would take all of 10 seconds to implement. So yes, something happened yesterday.

> They wont do it.

That's the point of this post, Boris Cherny closed it as "completed", of course without implementing it.

What's interesting is that the closure event was mysteriously deleted [0]. Compare with others like [1] which do keep the "bcherny closed this as completed 1m ago"; for some reason it's missing here, even though Boris closed it as completed in the same manner as confirmed by GitHub notifications. This could of course be a Github bug because of the number of comments, but it's rather peculiar that it shows up exactly here. Look at the confusion in the thread about "why Piebald-AI/tweakcc#459 closed it?". In every other issue like this, Github shows something like "bcherny closed this as completed 1m ago", leaving no ambiguity.

Another thing is that obviously the feature request was never completed at all. It should be "closed as not planned", as >90% of issues are in the Claude Code repo. Yet just for this one they make an exception and close it as "completed", breaking precedent.

It's so chickenshit. Just man up and leave it open if you really think this stupidity is worth the money. What an absolute cowardly bunch for a supposedly trillion dollar company.

[0] https://github.com/anthropics/claude-code/issues/6235#event-...

[1] https://github.com/anthropics/claude-code/issues/66955#event...

Here, filtered down for you. [0] Look at the individual benchmarks, not the combined one. You can tell that this model is much more benchmaxxed as its relative ranking swings between benchmarks is much larger. This is a hallmark.

[0] https://artificialanalysis.ai/models/qwen3-8-27b?models=deep...

This one is very benchmaxxed, and you can tell from this page alone. Look at the huge variance in ranking per benchmark. Most models, including at that size, are much more consistent.

> It is on by default for free sites fr the reasons we wrote about in the blog post below.

Nowhere in the blog post does it give the reasons why it's A. on by default for free sites B. off by default for paid sites.

No support was added, and the closure event was mysteriously deleted [0]. Compare with others like [1] which do keep the "bcherny closed this as completed 1m ago"; for some reason it's missing here, even though Boris closed it as completed in the same manner as confirmed by GitHub notifications.

[0] https://github.com/anthropics/claude-code/issues/6235#event-...

[1] https://github.com/anthropics/claude-code/issues/66955#event...