• 0 Posts
  • 14 Comments
Joined 5 months ago
cake
Cake day: July 15th, 2024

help-circle

  • Intermediary platforms are like this, yes. They take place of what should be infrastructure.

    I hope everybody understands that if some standard, easy to get into payment and catalogue system were in place, nobody would need these platforms. If you could pay to an IP address as easily as you can ping it. I mean, I think identities should be cryptographic in that, but you get the idea. It should be lower level functionality.






    1. Not just that, but also it’s small in description. If you read their papers, they are very easy to understand. I suppose that’s intentional, clarity and simplicity are among the main criteria of anything intended for security.

    2. “A lot of eyes” is overvalued. There are a lot of eyes on every nation-state in history too, you tell me how that works.

    3. It doesn’t matter because of protocol design. They’ve solved very complex problems and have not stopped doing that. E2EE is the wrong buzzword, zero-knowledge is the right one. No, I’m not remotely qualified enough to explain what that is.

    4. Still supply chain attack on clients is the most probable, but not much they can do with it. It’s similar to fearing trojans on user devices. Yes, 3-letter agencies and such most likely will do that, not bother with pressuring Signal developers. And no, there’s not much you can do to defend against a targeted attack, if it’s targeted, then you’ve already bothered people you shouldn’t have.

    5. Well, it’s not as if one could avoid that. It all lies in the area of smart contracts and distributed computing then, and see point 1, right now Signal’s protocol can be in general strokes understood by someone like me. If they make something like that, it won’t be. Everything is a compromise.

    There’s functionally less “trust” here than any messaging application on the planet.

    I think Wire and maybe Session use slightly modified Signal protocol. But Signal itself is the thing, made by people with clear vision of the whole architecture, model, which is not limited to protocols, but also to sociology, human psychology, politics. And they’ve explained literally every architectural decision of theirs in articles.


  • E2E is only available in one on one chats and is disabled by default.

    Considering that there’s no technical problem with enabling it for all one-on-one chats, this tells a lot.

    Also no E2EE on desktops.

    I hate TG’s UX. It’s atrocious. WhatsApp is the closest to something normal, but imperfect too.

    At least it has an open-source client.

    Chromium is an open-source browser.

    OK, more specifically - what matters is that TG’s protocol is a big ugly target moving fast. So its official client with released sources is in practice the only one. There are things like libpurple plugin and some python TUI client and an emacs one, but they are all lagging behind. And I think they are all using official tdlib.

    This tells something too, that their talk about possibility of alternative clients is of the same kind as their talk about privacy.

    About the network effect - bring your family and friends to Signal one by one. Of course it won’t happen overnight.


  • It’s not just tech bros, it’s the whole approach - weird names, version numbers turning into marketing tool instead of just numbers, attempts to hype up things that shouldn’t be hyped up.

    When I was a kid in Russia in year 2003 (suppose), it was associated with everything Chinese. But then Windows Vista and iPhone and what not … came into normality. And now everything, not just toys produced in China, is something made of plastic and intended to break next day and be unfixable.

    I’m torn between two things - one is to accept life as it is, because that’s truth, and another is that in future of my dreams we’d have good, reliable things, their price and availability helped by scientific and industrial development.

    I guess what one can wish is for the developing world to finally develop in all its parts sufficiently to make the current paradigm of a few manufacturing countries making everything for the rest of the world, but using IP of a few designing countries, unworkable.

    Decentralization and competitiveness help everyone.

    I think IP and patent laws have been a tool to create stagnation. You won’t make Spectrum-like machines for kids in school, when you can have something from the Intel+AMD/ARM-ASML-TSMC ecosystem. And if you don’t accept US and EU and in general European world’s IP and patent laws, you’ll get practically embargoed. And those are close to legalized monopoly. And without breaking a lot of patents, even trying to build a competition to ASML and TSMC in like 40 years is going to be a few orders of magnitude less possible than with breaking them (still not very likely).

    So what I’m trying to say - Speccy is probably not something to aim for now, it’s not problematic, just no demand. But aiming for something like Sun equipment of year 1997 would be a good idea. If hardware of that level were produced on scale in a few bigger countries, like Brazil or India or even China, it would make a lot of difference. I know China has Loongson. On scale.






  • Small notes to be answered rarely.

    I’ve looked at the early Usenet archives, and typical posts there resembled this format quite a lot. It’s later that Usenet became a place where you write long considerate posts, and also expect rather quick answers.

    It’s actually interesting to communicate in a rare terse format.

    The reason I don’t use Twitter, BlueSky, anything like that is - I don’t have a scenario of it being useful for me.