SquareWheel

Doesn't that just mean that local inference runs on the CPU?

The original source was posted yesterday.

https://news.ycombinator.com/item?id=35928000

They said they were using a personal Workspace account, not a paid business account. They're likely on a legacy Google Apps account, as am I, for having signed up using a custom domain name 15 years ago.

Just making a new account isn't a realistic option when you can't move over purchases, YouTube uploads, and the millions of other pieces of data tied to that account.

I wonder the same. I bought an extra code back in alpha because Notch promised it'd grant all future versions of the game. I wonder if it will still redeem as a valid code after two different system migrations (Minecraft account > Mojang account > Microsoft account).

That's an odd choice as WordPress, which is by far the most popular CMS, uses ?s= as a search query.

I would expect those pages to be included in the data. They could offer some sort of segmentation if they think they they can separate out searches, though.

At the risk of bikeshedding, what's with the syntax? Would it not make sense to stick closer to HTML tags? eg.

    <script type="importmap" name="browser-fs-access" src="https://unpkg.com/browser-fs-access@0.33.0/dist/index.modern.js">

The current implementation seems like something I'd have to look up every time I wanted to use it.

You can't block fingerprinting completely without breaking a ton of useful features. But the sandbox has a concept called the privacy budget which tries to determine if a site is collecting too much information. It should allow sites that actually use some of these features to continue to work.

The idea is that if sites that query fonts, engage canvas, read the user agent information, etc, they are likely trying to build a fingerprint, so the browser will start to return generic data.

Presumably - hopefully - it would allow users to set their own privacy budgets. Even better if it supports granular per-site control, which may be needed for certain specialized websites.

https://github.com/mikewest/privacy-budget

Correct. Also it was said by a lone developer a couple months after starting a hobby project. To expect that throwaway comment to mean anything after fifteen years and a 2.5 billion dollar acquisition is simply unreasonable.

I'd say Veloren is closer to Cube World, while Minetest is closer to Minecraft. Both are cool projects, though.

This is a weirdly contentious issue by web devs, but in my experience, system-ui is fine. It's used as the default in popular frameworks like Bootstrap[1], and most of the teething issues have been solved for a while now.

[1] https://github.com/twbs/bootstrap/blob/adf7b8dc4083b6ddc318e...

I can't agree. The comment above is extremely constructive, and I'd be glad to receive that much helpful feedback on a project I'd submitted. The advice is both well-informed and actionable. That's not something you usually get for free!

In the context of scientific skepticism[1], denier or denialism[2] is the correct label.

[1] https://en.wikipedia.org/wiki/Scientific_skepticism

[2] https://en.wikipedia.org/wiki/Denialism

>it's literally "GPT for all"

It wouldn't have even occurred to me to read it any other way.

That might be true in concept, but I don't think it's true in reality. When people search for something online, they want the first result to be whatever they had in mind. If I search for Python, I want the homepage for the programming language. When somebody else searches for Python, they might be looking to learn about the animal.

You could argue that we should just ask for exactly what we want, but that puts more work on the user and reduces the effectiveness of the tool. I don't want to type more than "python", I just want the link!

Search engines provide very little personalization even today. I'm almost surprised a competitor hasn't popped up with a product that tries to fill that niche.

> but removed it initially for unknown reasons

Just an FYI, they gave their reasons in comment #84 of the thread you linked.

> Nobody is going to adopt something without being natively supported by browsers without ugly js hacks

That's fair, but it is worth pointing out that <picture> tags support multiple source formats. I serve some logos and other important images as both .png and .webp. Adding additional lines for avif for jxl seem reasonable - especially if browsers are testing new formats.

I tend to feel that way mainly towards new features that are very similar to ones that already exist. eg. WebGPU being a more advanced form of WebGL. Yes they work differently, but they generally solve the same problem.

Eventually I'd like to see WebGL deprecated, and perhaps an extension released to reimplement it. Or it could be adapted to translate calls to WebGPU, rather than having a completely separate implementation.

Is that realistic? No, of course not. Browser vendors take backwards compatibility pretty seriously, and WebGL is used on a ton of websites. Still, that'd be my preferred solution for avoiding said feature creep.

This is a really great question. Learning more can open you up to new ideas, new topics, and new ways of thinking. At the same time, it can bias you towards specific ways of thinking, or to adopting others' models rather than creating your own.

All I can say is that learning more will change your creativity in some way. And often, I think that's a good thing.

He's been heavily researching machine learning for the last few years, and formed an AGI startup last year.

He's not the world's leading researcher or anything, but he's far from green in this space.

At the very least, it has its own privacy policy.

https://docs.github.com/en/site-policy/privacy-policies/gith...

>It's one of the most inefficient, slow, resource hungry app platforms that humanity has ever devised.

Is it? JavaScript interpreters are crazy fast, almost comparable to compiled languages. The layout engines are powerful and efficient. They support extremely-optimized networking protocols and media formats.

It's not (currently) a great platform for high-poly 3D rendering and such, but who knows what the future holds with wasm and WebGPU.

Browsers do use a lot of memory, but so did Java, Flash, Shockwave, Silverlight, etc.

The trouble with spammy jokes like this is it discourages companies from bothering with open-source in the future. I know I'd be less likely to champion an initiative like this if I thought it might blow up in my face.

That seems pretty extreme. Why not just add nofollow to links? That's what websites like Wikipedia do.

Pretty much every platform with notifications supports a "do not disturb" mode for just such occasions.

Most browser vendors already do, or will soon require a user gesture to initiate. That doesn't mean that abuse will disappear, but it should help.

https://developer.mozilla.org/en-US/docs/Web/API/Notificatio...

Usually the dynamic data would be generated by the server. eg. if you have an eCommerce site, it might update prices, stock remaining, applicable regions, etc. That would then be pulled into Google Shopping or whatever other service might consume it.

Much closer to what I had in mind! Thanks, this one is new to me. It would've saved me time in my previous testing.

Hopefully one day the existing CSS properties like letter-spacing can be supported in @font-face to allow for even better matching. That would make techniques like this even more powerful.

Then your default font stack should look something like this.

https://github.com/twbs/bootstrap/blob/adf7b8dc4083b6ddc318e...

Thanks for the link. I've actually used this one before, but it's not quite what I was hoping for.

These settings (font-size, line-height) are defined in the content CSS, not the @font-face rule, so they won't be automatically applied when the font is swapped. Properties like size-adjust or ascent-override are better for matching like-fonts as it's all automatic.

Also, while it is very useful to have a tool to come up with your own pairings, I think a list that covers some of the most popular fonts and creates perfect settings would be very useful for quick implementations. Something like the submitted site above that has done the legwork for you.

Great website. There's really no concept of "web safe" fonts today when you mix in mobile operating systems. At best you can choose similar-looking fonts and hope for the best, and this site seems useful in that aim.

One tool I'd love, though would clearly be out of scope here, would be a way to find safe fallbacks for popular webfonts. For example I recently created a site in Montserrat. After some testing I found a close fallback font was Verdana, with a size-adjust[1] of about 99.5%. That resulted in minimal document reflows when the font was slow to load.

Picking the top 10 or 20 popular Google Fonts and finding nearby fonts with good scaling tweaks would be very useful. I could see a sister project to this site offering something like that.

[1] https://developer.mozilla.org/en-US/docs/Web/CSS/@font-face/...