Wowfunhappy

> Have you tried this recently?

Yes, constantly. My default search engine is Startpage, which uses Google's API behind the scenes, so I have Google's results with no frills.

No, because the bottleneck isn't just model speed. The model has to literally compile the engine and run the massive test suites. Some of that could be sped up if you had a real build farm instead of my little computer, but there's a limit to how much you can parallelize things. And remember, I was porting WebKit, not remaking it, which would be orders of magnitude more work.

So you're not going to get to "build me an original web browser from scratch, make no mistakes" and get a new app the next morning or even next week, I don't think.

I've actually used AI to create a WebKit port (not an original browser). The amazing thing is that it worked--but it still took months of issue triage. Even when the AI was working on its own, it had to spend enormous amounts of time just on compiling and testing. There are so many moving pieces, and each individual issue could mean the AI spinning for days.

Even if a future AI can drive web browser development completely autonomously, I suspect it will take a huge amount of real-world time because that's just the shape of the work no matter how "smart" you are.

This really isn't complicated. Students should be able to search for something and get back the "ten blue links" which they are forced to actually click through and read, without an AI overview or any other kind of conversational chatbot.

Approximately no one cares if Google is using an LLM internally to select those ten links. Maybe it creates some complications for lawyers around the definition of "no AI", but if the parties are aligned in spirit this should be easy to work out.

Schools should just be able to disable the AI overview.

> They have made a pragmatic decision already to cut loose a whole swathe of users so they can pursue areas higher up the value chain. that was essentially the original decision they made when they went after software engineering instead of general users.

...I don't think I agree that this happened. Claude Cowork exists, as does Claude Design.

My read of Anthropic's strategy is to focus on coding as a means of achieving capability in other domains, via writing code. I would say that's paying off: see Claude Opus 5.5 creating a music video, via building animations in Javascript. https://www.youtube.com/watch?v=CS8ro03rJOM

By writing bits of code, Claude can organize your calendar, create vector graphics, organize your tasks, and so on.

If what you mean is that newer Claude models are worse at creative writing—okay, but to what extent were LLMs ever seriously useful for this, beyond generating slop no one actually wants to read? Opus 3 is still available in claude.ai if you want to use it.

---

Now, separately, I do think that Anthropic is internally obsessed with "existential risk", to the point where they will cut off users. I do think this is genuine based on everything we've heard from Anthropic employees.

And of course, this safety culture likely persists in part because it's also useful to the company's overall strategy, in the same way Apple's focus on user privacy, while likely genuine, conveniently lets them maintain control of the App Store.

Right, but that means there must be some PCs that take > 8 seconds. Has... anyone ever seen one? In the past two decades? (I'm time bounding the question only because I know nascent PCs did lots of weird stuff.)

It's not like there are that many PC SKUs out there (or motherboard SKUs for custom stuff). And even if there are two weird models, how useful is only that data?

But that doesn't explain where the dumps are coming from.

> What further complicates matters is that Chen details how the kernel dumps created after pressing the power button for 8 seconds do, in fact, get sent to Microsoft once the computer turns on again. In fact, he is one of the people who gets to see these dumps and respond to them, so obviously they are indeed rolling in.

Appropriately for halloween, this seems somewhat spooky.

Yes, the article mentions this:

> This glaze is used on jelly beans, Milk Duds, Junior Mints, Whoppers, Lemonheads, and Sugar Babies, to name a few.

I think you're seriously underestimating how expensive software decoding AV1 is!

I'm not even entirely sure whether a NUC could decode 4K AV1 in software in real-time.

> but I guess because no one at Apple was willing to maintain it and it apparently is a frozen API that doesn't support newer than TLS 1.2 (!)

Well yeah, so not really curl's fault here.

More importantly, IIRC Apple's curl actually stopped using SecureTransport before upstream dropped it.

> CLI tools are not to be feared or considered somehow hacky.

I agree, you're on a *nix OS, calling a CLI tool should be a perfectly valid way to do something.

IMO the real issue is that Apple's built-in CLI tools aren't properly integrated with the rest of the system. Apple's curl used to use the same certificate store as the rest of the OS, then Apple changed it.

No. But I'd certainly put in the cost of a router.

I don't think the limit changed per se, but Opus 5.5 consumes a lot less usage than 5.0 did, so it goes further.

> PDFs are executables, that's why it's hard. They support running code

I don't think Apple's PDF renderer supports e.g. running JS, you'd need real Acrobat for that. Or are you talking about something else?

> they support decoding images which themselves have their own custom compressors that are also basically code themselves.

But this is fundamentally true for anything that displays images, right? Or are the PDF formats particularly bad for some reason?

Is this not your work? If you didn't make it, you should email hn@ycombinator.com so the mods can remove Show HN from the title. This is a great submission, just not a Show HN.

One thing that occurs to me is this project is a way to get a fully jailbroken iOS. You can tweak the kernel and go nuts. Can't do that on an iPhone.

Apple's own design guidelines originally said not to do the thing they did (and which they are now once again saying not to do).

Couldn't Apple have just, like, talked to the people who originally wrote the guidelines? Why do they need to re-invent the wheel? Will Apple forget again in another 15 years and need to re-run the experiment, taking us all along as test subjects?

I can see how peas might solve the first and last problems on your list, but wouldn't the others remain?

> There is no form of surveillance that does more good than harm in the long run.

...how far do you take this maxim? Should we have any police on the street to enforce speed limits, since they're surveilling the speed of drivers?

Seems like a good time to get a third Claude to hear the cases of the first two Claudes and break the tie.

Again, this is really not the point I'm trying to make! If your prediction is that we'd end up in some apocalyptic dystopia, I'll concede the possibility.

But there still wouldn't be any jobs cleaning toilets, because the people who still had toilets to clean would have robots to do it.

That's not really the point I'm trying to make.

In the future, do bathrooms need to be cleaned? If yes, is it cheaper for that to be done by a person or by a robot?

Obviously, it is in the interest of the robot manufacturers to make the robot just a tiny bit cheaper than the person, so you'll buy the robot and the company makes the maximum amount of profit.

Now, if society breaks down and there's no one to consider buying the robot—well, that's not good for anybody, which is why I suspect society will find some way of working this out even if it's not ideal. But whatever, I don't particularly want to argue this point.

What I do want to argue is that there probably isn't a universe in which all white collar work goes away but blue collar work is still around.

Let's assume, for the sake of argument, that AI truly is able to automate away basically all software development. Any oversight currently performed by humans can itself be done by agents.

...don't you think the ability to create virtually unlimited amounts of software, at virtually no cost, is going to lead to major advances in robotics? I realize there's a lot of very hard problems involved in e.g. robotic hands (to take one example), but once software development is practically free, I'd expect a lot of seemingly-impossible problems to get resolved by the sheer quantity of what we can create and experiment with.

It's far from the only annoying false positive.

Here's a real world case I've run into a bunch of times. Parents send children to school with Airtags in their backpacks. (The children are young and don't themselves have phones.) Class goes on a field trip. Boom, tons of notifications.

If I had a pseudo random number generator that somehow kept outputting winning lottery numbers a reasonable percentage of the time, I would use it to enter the lottery. Even if it seemed completely impossible and despite studying the algorithm, I did not understand how it worked—I would still use it.

That's basically how I feel about LLMs right now. I have no idea how any of this is possible but somehow it seems to be working.

...if you are told in writing (say over email) to ignore a part of the contract, can you do that? Does that part no longer apply?

I don't understand how you're getting from "open weight model providers are forced to compete on price" → "nVidia is forced to compete on price".

Yes nVidia is not the only game in town, but that's no more or less true regardless of whether there are a bunch of competing providers.

> If it becomes a commodity business with frontier open models, NVIDIA won't be able to get away with such a crazy markup.

Of course they would. You need hardware to run the model. nVidia sets the floor.

Unfortunately, any "insight" it might contain is ruined by the fact it's clearly written by an LLM instead of a person.

> And it never comes in a single dramatic announcement. It comes in layers. A feature post with a price change inside it. A LinkedIn update nobody made a press release about. A values page that says something slightly different than it did last week. If you’re still on Bitwarden cloud and this is giving you pause — it should. [...] Whether self-hosting stays viable long-term is the real question worth sitting with.

This was great, but honestly, I would have enjoyed a non-joke take on this. Like, but with a cinematic story around it.