serbuvlad

Obesity is trending back down and will continue to do so.

Machines have removed our need to work, this stunts our body, and the solution is synthetic work (gym).

AI removes our need to work with our minds (at least partly), this stunts our mind, and the solution is synthetic mind-work.

Sure it will take some time to get used to. But we will get used to it.

If you look in history this is perfectly normal.

Nothing new under the sun.

You are missing my point so much...

1. I am sure there are historical reasons for the way things are.

2. Qt is not a platform. XDG/Wayland/Vulkan/PipeWire/D-Bus is the platform. Qt is a library. The platform is (relatively) lean. And the library is more complex. Whether an app is written in Qt or GTK or anything else it should be transparent for the end user. Whether an app is an executable or a website is not transparent, hence different platforms.

3. Rendering in terms of HTML/CSS is certainly valuable, but it can be reduced to a small subset of what it is + libraries.

4. This reduction would be very useful.

5. In this counter-factual reduction, one can just invent a better method for library deliverables, although I suspect a long TTL chached GET from a cannonical source would be much more than sufficient.

> That's a luxury not afforded to browsers.

That's a luxury not afforded to platforms. You cannot remove core Wayland protocols or core Vulkan functions or Linux syscalls.

This generally isn't an issue if you don't bloat them in the first place with things that don't belong in a platform and that belong in a library.

I'm not arguing against Jev. Jev is very cool.

I'm arguing against the idea that Jev fixes a fundamental deficiency in LLMs by showing it's confidence.

It doesn't.

Jev can do a strict subset of what LLMs can do, but incredibly quickly and incredibly cheaply, which is certainly valuable.

This feels more like mythology though.

Claude and GPT tell me all the time what they are confident and what they're not confident in. This is partially useful, but it's not a guarantee of anything, since they can be confidently wrong or unsure and correct.

Same with Jav.

Arguing that Jav in some way more useful than current LLMs because it includes a confidence score smells of someone who hasn't updated their priors since GPT-4/4o and it's particular issues.

Jav is cool because it's very fast, very cheap, and "good enough". That's it.

How would a GET that hits cache 99% of the time track users?

Sure, but I meant more in the sense of asking about the current status of the page.

One can just add APIs to the Linux kernel, and this would be a pretty straightforward one. (though it might be a little more difficult to avoid a syscall here)

I'm sure an agent can work on this and get some numbers with a day's worth of tokens.

> Further (and this one is entirely on you) the real comparision isn't "unixes" it's a graphical env. So compare the complexity and attitude to win32 dialog apis, QT, etc...

Yes these entirely fit the platform/library hard distinction model.

The platform is the small API to create windows, recieve input events and get an OpenGL/Vulkan/DirectX/Metal surface to draw to, as well as the graphics library itself, which is similarly based on primitives.

Everything above that is a library.

> Because the users of those systems don't really want a canvas they have to render everything on... rendering is hard.

Qt isn't a "system". libQt5 Core + Gui + Widgets + Qml is roughly 20MB on a Linux system I can readily check. This is way more than lean enough to ship with every app and even to download-on-first-use and cache for a web app (assuming it was ubiquitous and many websites would want to pull it from cache).

What the system gives is exactly a canvas to render stuff on.

If you know anything about non-technical people, "how an app looks", is by far the most important thing about it.

The web is uniquely capable (aside from raw OpenGL et. al) at rendering any UX designer's vision.

Almost all alternatives to the web that I've seen make aesthetic impositions, and are therefore non-starters.

Almost every website I visit has massive pop-ups I have to dismiss. Clearly a cached fat library GET would be less annoying than what I already put up with.

The browser wars are long gone, and there is no technical reason why the standards each browser added had to be complex per-use-case builtins instead of composable primitives.

In any case, that is what happened.

And now new projects like Ladybird take a decade to arrive, even with AI, because of the immense amount of stuff in the platform.

> standard that's the product of feature competition between pre-existing nonstandardized behaviors is doomed to be huge.

None of the Unixes are this bloated.

You are the one misunderstanding my point.

There ARE ONLY THREE implementations of the client platform.

I would like for there to be a hundred, and there would be if the platform was simpler, but its not, so there won't.

Yeah exactly. So does native. So what? Why is this a bad thing?

But libraries can be GET'ed from common URLs and cached!

Coming from a more general programming view, I find web development extremely odd. In general programming we tend to find a small set of abstractions which we can use in a composable way to cover our problem space.

For example, to interact with the VFS, we have read()/write(). When we add more APIs, it can be to enable a new paradigm, like epoll(), or for performance, like readv()/writev(). For a functionality that can be composed out of existing APIs in a performant way, we do not add it in the platform, but leave it instead to the domain of libraries. This separation has immense values, as it makes platforms easy to implement. A Linux filesystem, for example, needs to implement only 10 or so functions.

The web seems to be perfectly composable too, out of <div>, <span>, <p> and a small subset of CSS, but doing this is considered an anti-pattern, and you're supposed to reach for the platform to find the closest thing to what you need, whereas reaching for a library or composing yourself is frowned upon.

The result is that there are only three platforms: Blink/V8; WebKit/JavaScriptCore; and Geko/SpiderMonkey. With many things only working or working well on Blink/V8.

And implementing a new platform is a titanic task.

Yeah that's what I meant

Afaik tab groups are a much weaker idea because they are non-recursive.

> The best two examples of this are the Arc browser and Chrome OS. But I think Arc is actually the more interesting of the two experiments. Their first innovation was to break the idea of tabs at the top and instead move it to be a sidebar model.

I will forever hate Arc for popularizing vertical tabs the way they did.

Vertical tabs were a thing since at least the Tree Style Tabs extension (2007), except the Tree Style Tabs were infinitely cooler, because they were a tree, not a list, and right click->open in new tab, would open the new tab as a child of the current tab.

So you could have a hacker news tab during your break, start opening a bunch of articles, read through them, simply un-expand the HN sub-tree, go to work, and then re-expand it during your next break.

Because of Arc, everyone is now implementing vertical tabs, but no one is implementing Tree Style Tabs, and, probably no one ever will again.

I have had way more issues provoked on RHEL-compatibles by RHEL's Frankenstein backporty kernel than I ever have on Arch or Nix by the latest stable kernel.

Linux LTS is much more about proprietary drivers targeting a stable internal kernel API/ABI than about anything else.

1. I highly suspect a pretrained-only model could not do this, yes.

2. Most problems anyone has are solved problems or variations on solved problems. This type of distillation is extremely useful.

I don't think that Linus is dismissive of security, it is that he is very much a proponent of always rolling to the latest stable release.

Linux only ever wanted to promise support for the latest release and even Linux LTS is a concession.

And CVEs are basically a useless concept if you roll. (or at least not any more useful than any other bug tracker which supports tags)

I think this is focusing entirely on the wrong thing.

Non-Mac computers are still called "PC's", but IBM has nothing to gain from this. Similarly, copy machines at least in my country are called "Xerox's" but it's very rarely a machine from the Xerox company.

If you want to de-throne Google, you cannot do so by policing language. You have to build a better search (not more private! better!).

"The Brothers Karamazov" is my favorite book of all time, and I always find it quite an easy and enjoyable read. Though this may be a culture thing, as I am Romanian and the cultures are closer to each other.

The thing I love most about Dostoievski is how real everything feels.

A lot of fiction nowadays is based on: how the character portrays himself/how he thinks about himself/who he really is.

And a lot of resolution is about finding out who he really is, or about other characters finding that out and accepting that (usually immediately). Followed by emotional chatharsis and resolution.

In Dostoievski it feels like a lot of the character are chasing that idea and simply finding out it's not true. They will give long speeches about who they really are, expectijg emotional chatharsis, only for none to come. And they react by feeling resentful towards their confidants, or towards themselves. Or shameful. And then they also react to their reaction.

There is no gold character core waiting to be resolved, and a reveleation is usually just a very awkward monologue.

My model is not a value judgement on quality. It simply is.

It's not different, it's just what Apple is doing (Safari uses pills).

In fact, I propose that all computer User Interface design proceeds as:

    D(s) = (1 / (c * s + 1)) * (A(s) + N(s))

Where c is some constant, A is what Apple is doing and N is random noise.

Turns out humans are not at all hard to persuade. :)

I haven't heard about them, I was going to comment about how these reaserchers will get bitter lesson'd, but after looking at what they do, their "world models" are actually transformer models.

This means that the conversation has shifted to the point where we can assume that transforms hold emergent world models (already proven for tiny toy worlds). This is good, because it's probably true.

In any case, this looks awesome and I wish them all the best!

I'll give this a shot just as a cheaper Morgen.

Sounds like all the features I need and none of the ones I don't.

Cool!

> That is, if you move your git hosting to GitLab then you have to change your code!

Not to be cynical about this but I fail to see how running sed s///g after a very rare event qualifies as a serious problem.

Now, sure, given that the solution is so simple, it's a nice recommendation, but still...