Wowfunhappy

Haiku, Sonnet, and Opus also made sense to me, but Fable ruined it. It isn't at all clear that a fable is bigger/grander than someone's magnum opus—actually, I'd think probably the opposite.

I don't think it gets better if you swap "fable" for "myth" either, although when you add "-os" to the end it suddenly sounds grander so there's that. Mythos isn't a model I can use, though.

You originally said iOS 16. The recent iOS 16 update was only for older devices.

Apple typically continues releasing updates for n-1 iOS for a few months. After that limited window, updates are only released for the latest iOS, except for devices which are too old to run the latest iOS, which continue receiving updates on the last-compatible iOS for a couple of years.

Only devices that can't run anything newer got that update. There is no choice available.

...Unless I'm misunderstanding, I don't think any promises were broken here. Google said they would provide ten years of software updates. Now they're saying existing ChromeOS devices will be upgraded to the new "GoogleOS" within that ten year window, they will continue to receive said updates.

This is annoying if you don't like GoogleOS, but also par for the course. You can't stay on iOS 26 and continue to receive security updates, you have to upgrade to iOS 27.

You clearly aren't using subagents!

I really think they were just concerned about Wi-Fi congestion.

Do they block individual wired connections too, or only Wi-Fi?

Asking residents to refrain from setting up their own wireless access points actually seems like it could be reasonable? It might lead to less congestion and better performance for everyone.

I'm reminded of my college dorm. Every room had an ethernet jack, and there was a communal Wi-Fi network, but students were asked to not create their own Wi-Fi hotspots.

> You’ve got a vendor who could kill switch critical military technology if it’s used in a way they decide falls outside the scope of what they want.

If the DoD hires ACME to build some software to make widgets, and ACME uses Claude to build that software, where is the kill switch?

I guess the Claude model could hide a kill switch in the widget manufacturing code, but so could ACME's human subcontractor. Why is Anthropic being considered a supply chain risk here?

The model can refuse to build something, but it can't take back what has already been built.

So we have a pen that may refuse to sign certain orders, and is clearly labeled as such.

If the DoD needs to sign such orders, they won't buy this pen.

If a supplier needs to sign such orders, the supplier also won't buy the pen.

If a supplier needs to sign other orders, and this pen is the best one available, the supplier may decide to buy this pen. Since the orders are within the range of what the pen will sign, they get signed, and the supplier fulfills its contract with the DoD.

Where is the supply chain risk? The ink isn't going to erase itself post-hoc.

---

Now, if the DoD suspects that Anthropic will train its models to try to actively sabotage military operations, that's a very different story. Does anyone actually believe that?

> The individual numbers are easy to skim past, so it is worth putting them side by side.

I just don't understand why people who clearly spent time doing interesting work, which I would like to read about, feel the need to run their findings through an LLM like this.

Please just talk about what you found! Why do you find it interesting or notable? That's what I want to read!

It's not a bad idea, although at that point it probably shouldn't be a .md file anymore. You'd want to use YAML or something.

> [The Manager] continued:

>> I know that if we get into the details, the reasons will be perfectly reasonable. You'll explain what happened, I'll understand why everyone made the decisions they made, and I'll empathise with you.

>> Then it'll happen again.

>> So I don't want the details. I want to know what we're changing.

Even if the manager's approach was correct, this is just a hurtful way to frame it to your team. It's great that the author could reverse-engineer what the manager was doing, but it shouldn't come to that.

Why couldn't the manager say something like:

> You are all great engineers, so I already know that everyone made the best decision available with the information they had at the time. What are we going to do differently moving forward?

I'm still not sure I agree with the strategy, because I don't see how you can understand what happens next without understanding what went wrong. But I can see how the framing might help steer the discussion.

> somebody, somewhere, was taking advantage of that to give one set of instructions to Claude, and a different set of instructions to some other provider.

I'm this person, I am actively doing this! I have some instructions which the Claude models need and OpenAI models will do by themselves. If the OpenAI models see the Claude instructions, they will go way overboard in a way I don't want.

Mind, the change does not affect me at all, because when both instructions are present Claude continues to read CLAUDE.md and Codex continues to read AGENTS.md. But if Claude started preferring AGENTS.md, that would be mildly annoying.

> Exit servers (run by Mullvad) connect you to the internet but never see your personal info. Obscura masks your real IP address when relaying to the exit server.

How is this possible? If the exit server doesn't know your IP, how does it know where to send the traffic?

Do you know where these rumors are coming from? I keep seeing posts like yours on HN and Reddit, which, yes, literally confirms there are rumors, but I can't tell whether they're based on anything.

Well, yes, because both of your lawns got mowed!

Value was created!

The question isn't whether the penalty would be less than their profit, it's whether the penalty would be less than whatever they make by secretly downgrading the models (or whatever it is you suspect), which remember also causes consumers to get less value out of the product and more likely to cancel.

The reputational hit, if this was to be confirmed, would also be massive. And I do think it would leak! Some employee would say something.

> What is actually stopping these model companies from running a model at full capacity on release then once its name rings out, start serving users quantized garbage?

...I mean, if they were actually doing this despite saying that they don't—promising one product and delivering something else—I think that would be fraud, no?

And, maybe it's one thing to secretly defraud normies like us (although class action lawsuits do exist), but I don't think major enterprises or the US military would take too kindly to it.

The original source appears to be ArsTechnica: https://arstechnica.com/tech-policy/2026/09/republican-bill-...

When something is new, its capabilities feel incredible. Over time, those same capabilities become mundane, and you start to notice the flaws.

(Now, if TFA is actually measuring reasoning tokens, that's quite different! It's not entirely obvious to me how he is measuring.)

If it's circumventable, how does it prevent fraud? Seems like it would likely just hurt the enthusiasts.

You can already check whether the RAM has been swapped via https://geekworm.com/blogs/news/prevent-ram-swapped-raspberr....

Well, for a torrent to stay healthy, users have to seed after download, so it can't possibly have the same UX as a standard browser file download unless you want to either kill the ecosystem or hide from the user what is consuming upload bandwidth.

That said, for large files, I much prefer the UX of a well-designed torrent client like Transmission to my web browser. If nothing else, the downloads are reliably resumable.

Hi Chris Chibnall, great to see you on Hacker News. :)

In all seriousness we clearly have very different taste! I just found the Chibnall episodes boring. I wanted to like them!

I think GGP shares my taste though, based on their comment.

But that's from, like, floating point errors, right? If you used higher precision that wouldn't happen, it's just because we're cheap in how we do rounding.

I can see how you'd nitpick this, but to me this is a deterministic algorithm that just happens to be running on nondeterministic hardware.

The Chibnall episodes sucked, but then Russell T. Davies came back as showrunner. If you haven't seen the first three episodes he did, with David Tennant returning as the Doctor, you're missing out!

I liked the regular season that followed too. It had good and bad episodes, just like the last time RTD was showrunner, but the good ones were very good.

(I haven't watched the most recent season yet, but that's less because I'm unexcited and more because I want to save it.)

The randomness is something we add on purpose; you can set an LLM's "temperature" to 0 to get deterministic output. This tends to make the quality of its responses worse for reasons I don't think anyone really understands, but it's still functional.

I don't think the state of the art LLM providers let you do this anymore (?), but they certainly could if they wanted to, and you can do it yourself with a local model.

Trump won the popular vote once, not consistently, against an understudy candidate who was pushed on stage at the last minute. Also, opinions have shifted since that time as people have noticed how unhinged Trump is.

I had a TP Link wireless adapter (I think that was the name) for my HTC Vive. I was kind of skeptical going in, but it worked shockingly well--no detectable added latency. The only thing I lost was the ability to use the Vive's camera, which I think was just a software issue of some kind.

> To be clear, you don't have to run things on it, you can stream wirelessly from your PC.

But the hardware is adding weight and bulk to something that you really, desperately want to be as small and light and possible.

> What claude dosnt have is an off day, its just the same.

But it... does. In some sessions it will get particularly paranoid, apologetic, self-doubtful, or suspicious. I'd say these traits are always there but they can become more or less pronounced.

(By contrast, I'd say the GPT models are much more even keeled, although I maybe haven't used them enough.)