I think I’d rather my phone be a little “dumber” than my laptop or desktop, though. Or I want it to be powerful enough to be the brains of both, but that would make it expensive enough that I would worry about losing it. Making it just a browser gives it enough utility to be broadly useful, but also enough friction that I won’t get sucked into it.

Also, I think a low-cost, low-power, mass-market B2G-type phone (a la the Chromebook) is way more likely than a mass-market Linux phone. Maybe that’s just me being cynical, though.

As for Google, yeah. I agree that they don’t have the users’ best interest in mind. But there’s currently enough of a pull from mobile Safari that they’re willing to play by the rules for now. My understanding is that the Web Attestation API was basically dead in the water—though maybe that’s me being too optimistic, ha.

Anyway, I asked where you’re coming from so thanks for sharing.

Same to you! Good conversation. I appreciate it.

Phones are terrible anti-user devices, so I can’t do the things I’d like to do with it that I can’t also accomplish on a website. Wasn’t that kinda the problem that was initially stated in the OP?

Maybe I phrased it poorly. I meant, what things do you do on your phone that wouldn’t be possible on a website if you were on another platform?

Actually, I’ve been actively trying to use Firefox Mobile for everything I reasonably can on my phone, and it’s way more possible than you might think.

I actually think a browser model limits a lot of what you say here,

I think you misunderstand me here. I’m not asking for a browser model to increase the number of things that app developers can do, I want to increase the number of things that end-users can safely do, and running web apps in a browser are currently the easiest way to do that.

and browsers definitely have ecosystem lock-in problems: what Google says essentially goes these days. The browser isn’t the great liberator of phones imo.

That’s absolutely a huge problem, yes; but it’s a different one. And in the faintest praise possible, Google does at least maintain fairly solid web standards.

I do however think it’s a weird way to try to fix the phone ecosystem by replacing a restrictive sandbox with a restrictive sandbox that also ties you to a really terrible development ecosystem.

It would be a replacing a sandbox that’s restrictive for the user and developer with one that’s only restrictive for the developer. And I don’t think it’s a particularly terrible development ecosystem; in a lot of ways, the front-end dev ecosystem is the most mature ecosystem. We’re absolutely spoiled for choice in IDEs, in linting tools, in packages…I mean, I used to work in email development years ago. THAT is a terrible development ecosystem, let me tell you.

In fairness I’ve not tried it, but their homepage has been all about apps for a while.

As for KaiOS, I don’t think that’s really a good successor of Boot2Gecko; from what I’ve seen they went the app route, which kind of fundamentally violates the spirit of what B2G was supposed to be.

Yes, the problem is that Android is talking about requiring developer verification at install time; as I understand it, without allowing an override.

Let me answer your question with a question: How many things do you do with your phone that aren’t also able to be accomplished with a website already? I’d be willing to bet that the answer is in the single digits. And for most of those, that limitation is likely to be entirely arbitrary, instituted by a developer as an anti-consumer form of lock-in.

Delivering application-like experiences via the web allows users to make accessibility changes to that experience without the developer needing to support it explicitly. It also allows users to implement plugins that extend and improve their experience, by removing undesirable content or adding functionality that you haven’t provided. And because browsers are built on open standards, there’s no longer any device ecosystem lock-in; I should be able to access all of the websites I want to from any browser on any device. Users could even build their own bespoke applications, without the need to enable a developer mode on their phone or get a certification from a megacorp.

And because downloadable and cacheable progressive web apps are a thing, as well as local storage options for browsers, the experience for an end-user of a browser-only phone wouldn’t need to be any different in low-signal or high-latency situations.

The web is a mature and proven platform for delivering arbitrary code and data, plugins make the web more accessible and easier to use, and web standards make the world more open. It’s not a perfect platform, of course, but it’s the one we’ve got; I think making it the default rather than the fallback for the devices most people use more than any other would be a great boon for the world at large.

I think that, with the current state of OSes like Windows and Android, there should be some minimal amount of friction to enabling installation of non-vetted apps. Maybe some switch that can’t be enabled accidentally, or without understanding that there’s risk involved (or at least a switch that can be disabled and password protected) for the sake of children or the elderly.

On the other hand, though, an OS should be built with enough security and sandboxing that no single application can brick your entire device without at least tapping through and giving it a ton of permissions; which means that the only remaining risk to the end user would be access to disinformation or other harmful content, or the risk of personal information exfiltration (i.e. phishing). At that point, a simple block list (or even just an allow list) maintained by a trusted guardian or third party would be sufficient to keep children or the elderly from harmful content, and whoops we’ve just invented the internet again.

I am once again begging for Boot2Gecko to become a thing.

Yes, that’s the classic solution to the riddle.

No, it would certainly be ironic (and raise lots of questions), but I don’t think I’d call it “silly.”

I meant “if the goal was also to find out which one is which,” sorry; meaning that you’ve already asked the classic question. But yes, that would also be a great way to figure it out.

Correct. If he’s the lying guard and something happened, or the truthful guard and nothing did, then in either case he would answer “no” to all the questions.

It’s interesting, because it makes me think about the solution to the original puzzle. In that one, you don’t care about the individuals’ identities, you just care about the answer, so you actually don’t know which one lies and which one tells the truth at the end. If the goal was to find out which one is which, you would need one more question: “and which one do you think I should take?” If he gives the same answer, he’s the liar; if he gives the opposite answer, he’s the truthful one. (Or just ask a question with a known answer first.)

1. Something cannot be silly if it is deadly.

2. Not every conversation has to go back to Trump. I know he’s on our minds a lot, and what he’s doing is awful, but being able to retreat into spaces where he doesn’t dominate the conversation is important for mental health, and also for our ability to recognize that he’s not a god-king. He wants to dominate your every waking thought. Don’t let him.

Who said anything about foresight? This is entirely historical. Thanks for actually reading anything I’ve said.

I’m saying they can only do it because the big innovation was “throw more money at it.” Yes, given a functionally infinite amount of hardware, electricity, legal free reign, and publicity, I could invent a machine that does at least one (1) impressive thing, too.

Remember, these models weren’t created to identify cancer in patients better than humans. They were created to do everything better than humans. And the fact that they are mediocre at everything except identifying cancer in patients (and a handful of other things) means that they’re failing at 99.997% of their goal.

That doesn’t mean that it’s innovative, or a breakthrough technology that deserves time to mature. It just means that you get more swings at the law of averages if you have a lot of money.

First of all, because it doesn’t matter whether it’s actually real or not, investment doesn’t actually follow innovation. The actual value of a company or idea has almost nothing to do with its valuation.

But more importantly, why do you think that’s the important part of this conversation? I’m not talking about its long term viability. Neither were you. You were just saying that it was a new innovation and still had to mature. I was saying that it was actually a much older technology that already matured, and which is being given an artificial new round of funding because of good marketing.

Are you trying to claim that the fact that there’s lots of money flowing to these AI companies is proof that AI isn’t just a bubble caused by money flowing to these AI companies?

Please let me know what major breakthrough has happened recently in the machine leaning field, since you’re such an expert. Throwing more GPUs at it? Throwing even more GPUs at it? About the best thing I can come up with is “using approximately the full text of the Internet as training data,” but that’s not a technical advancement, it’s a financial one.

Applying tensors to ML happened in 2001. Switching to GPUs for deep learning happened in 2004. RNNs/CNNs was 2010-ish. Seq2seq and GAN were in 2014. “Attention is All You Need” came out in 2017; that’s the absolute closest to a breakthrough that I can think of, but even that was just an architecture from 2014 with some comparatively minor tweaks.

No, the only major new breakthrough I can see over the past decade or so has been the influx of money.

AI isn’t “emerging.” The industry is new, but we’ve had neural networks for decades. They’ve been regularly in use for things like autocorrect and image classification since before the iPhone. Google upgraded Google Translate to use a GPT in 2016 (9 years ago). What’s “emerging” now is just marketing and branding, and trying to shove it into form factors and workloads that it’s not well suited to. Maybe some slightly quicker iteration due to the unreasonable amount of money being thrown at it.

It’s kind of like if a band made a huge deal out of their new album and the crazy new sound it had, but then you listened to it and it was just, like…disco? And disco is fine, but…by itself it’s definitely not anything to write home about in 2025. And then a whole bunch of other bands were like, “yeah, we do disco too!” And some of them were ok at it, and most were definitely not, but they were all trying to fit disco into songs that really shouldn’t have been disco. And every time someone was like, “I kinda don’t want to listen to disco right now,” a band manager said “shut up yes you do.”

The mayor of NYC has a lot more power than the governor of, for instance, Wyoming. More power than the prime ministers of a lot of sovereign nations, even.

I did a bunch of research into second brain/zettelkasten apps (that is to say, apps that support note taking with note interlinking and rich text) earlier this year, and I couldn’t find a single app in the category that’s (1) FOSS, (2) stores notes as .md files natively (Logseq will import/export to .md, but it’s not native), and (3) is cross-platform in some way (for my purposes, I need it to be on Linux, Android, and Mac OS, or have a usable web app). Even the ones that get close all have some kind of gimmick to them, or are super ugly or slow or otherwise hard to use.

If Void can get those three nailed, and do it in a usable way, it will fill a very particular and exciting niche.