Skip Navigation

Posts
30
Comments
608
Joined
3 yr. ago

  • It is not spam, and you miss-read it. Prose is an open-source XMPP client. They can set you up (host on your behalf) for free, up to a certain point. You can pay for it (there is a commercial offering), or you can use it unlimited and with no extra costs than your own server's if you self-host. It's all being developed there in the open in case you don't want to take my word for it: https://github.com/prose-im

  • In terms of tech and implementation details, it's been years since everyone has been converging towards the same WebRTC architecture (with everyone bundling/linking the same set of basic components and libs as found in chrome, android, ...). As such, a call between two participants (or as a group with less than a dozen participants) should be as good on XMPP as anywhere else (including the commercial options like Google Meet, Zoom, Matrix, ...).

    Of course there are caveats like relying on TURN where direct connection is impossible, but that's the gist of it. Regarding XMPP group calls,

    Where things start getting spicier is in large group calls (dozens of participants or more) requiring the stream to be brokered by a central server (SFU), with stream re-compression and optimisation. Standard-XMPP isn't great for that yet (non-standard XMPP, like Jitsi, on which it is based, is pretty damn good, but unavailable from your regular XMPP setup). Work is going on to improve that (on two fronts, with some XMPP servers turning into SFUs, and with a protocol being designed for offloading AV streams to any willing existing SFU).

    The problem with large group calls essentially boils down to how much bandwidth and CPU you want to throw at it, and that's not cheap (unless, of course, you are the product, i.e. Google Meet, Discord & al). The same applies to self-hosted Matrix/Galene/Jitsi: you probably won't want to hold a large conference call on a home-server, and the server admins are bearing some costs, so get to know them and how sustainable that is. In the case of Matrix.org, it is not.

    No idea what prose is.

    Prose is an open-source XMPP client with a focus on large rooms/banquet-style conversations (like IRC, slack, …). It is still in its early stages but already quite usable and possibly a good fit for a subset of Skype refugees.

  • None of those (except Jitsi to a small extent) qualify as replacements if we ever want to evolve out of the silos we let megalomaniac CEOs build to better control us. So I'll add to the list: prose.org , movim.eu (or anything based on XMPP) and matrix.org (though this one is rapidly falling into obsolescence). The keyword here is federation.

  • It has already started. Microsoft and Google hiking prices with AI bundled in is both a way to inflate the "demand" artificially, keeping the show going (covering up the fact that nobody really wants that, and even less so wants to pay a premium for it: there just is no miracle AI product/application to sell), and to mitigate some of the absurd imminent losses.

    You wouldn't see that in an "optimistic" and sound market.

  • Count me as a fervent critic of Hollywood, but the world isn't binary and (unfortunately) Hollywood hating it doesn't automatically make it a good thing for the rest of us. Essentially OpenAI, Google and the rest of the pack of thieves are lobbying to establish themselves as the rulers of a lawless world, and everything you already hate about Hollywood (its inordinate amount of power, the bullying of the weaker that ensue, the corruption and politics around it, ...) is meant to get back to us, in worse, with new names at the top.

    Indeed that would be the end of the copyright law, but only for the oligarchs.

  • Binnen schoenen aan

    🤢

  • That is exactly the point, and I wouldn't be surprised if soon there is more money to be made "certifying works made without AI" than there is selling API tokens for LLMs, i.e. the OpenAI business model (although I have no idea of what the technical implementation would look like, perhaps a mix of secure enclave computing offering only a predefined set of capabilities barred from AI, combined with a blockchain to persist and distribute the reference and hash of the works done? More to the tally of GenAI being a net loss for humanity).

  • That's why it should be made compulsory to indicate when something was produced, even partially, by an AI. This time of your life you spent reading some low-effort no-value bullshit you will never get back, and neither will the hundreds/thousands others who read it. This is a net loss for humanity.

  • Matrix has gone open core, XMPP is safer

  • You don't have to reinvent the wheel, open standards with similar or better capabilities already exist. Don't create another silo, contribute to making e.g. XMPP clients better.

  • but it taught us that you always want more than one method of contact, as a a rugpull can happen at any time off any whim.

    Being on the internet long enough taught me instead (by having seen countless providers rise and fall since the early 00's) to self-host my comms and prefer open federated protocols. I switched to XMPP, I have no regret, everyone that matters made the move painlessly a decade ago or so.

  • As someone who's been using ttrss for decades but would be open to trying something new, what would you say is FreshRSS' killer feature (and missing killer feature) compared to ttrss?

    (Not trying to start a flame war, ttrss feels like a finished project, which is not a bad thing, but I think it's healthy to wish for more innovation in this space)

  • How about nextcloud with the draw.io app/excalidraw integration?

  • XMPP had a kind of renaissance in the recent years (while Matrix only stagnated, and recently turned open-core, but that's besides the point), maybe time to give it a new look? :-)

  • I think it's steadily improving and should be fine for most projects nowadays. Moreover, there's now the option to let the Scala compiler do the error reporting instead of Idea's (haven't tried it, but in theory that should address the most vocal complaints).

  • Welcome on board! The pace is less hectic than on Reddit, tbh, but contributions are more than welcome :-)

  • And thanks to you as well for the cordial discussion! I'm hopeful that bystanders got an interesting read out of it :-)

    As of me, the "worst" I would wish upon you is to adventure into XMPP, via an easy-entry app like quicksy.im (android) or monal-im.org (iOS) and see for yourself that you can get something as secure and featured as Signal, without the captivity and monopolistic abuses.

  • I would go into the specific points, but really none of this invalidates my main point that Signal is a marked step forward

    We are going circles but I will repeat it: Signal isn't immutably better than WhatsApp, it only happens to be more politically-aligned with your beliefs (which we share in large parts, to be fair!) at this very instant (and we saw that this can change without notice).

    My threshold for justifying a mass-exodus out of a popular messaging system is that 1- it offers non-revocable privacy and security guarantees and 2-, that it doesn't lock its users in a single vendor/single service provider. Those two things combined are important, because they would finally give the chance of breaking away from the never-ending cycle of "enshittification → exodus → unsatisfactory explorations → painful rebuild(s) → monopoly consolidation → user captivity → enshittification". Anything else is a slight variation around the current disappointing status-quo. I don't think it's too far-fetched, and we really deserve this "luxury" for something as fundamental as instant messaging. I can only hope that you understand why I'm not willing to compromise on that.

    I'm also willing to bet that, with the rumbling going on in the USA at the moment, Signal might sooner or later become a target of/re-align itself with the new "administration". Maybe then you will sense more of that captivity I keep rambling about?

    My point was that you’ll be communicating with people each of whom chose their own service poviders, and thus you’re also trusting those.

    The worst thing the other server can do is drop your messages silently, which you will absolutely come to know. Think of XMPP with end-to-end encryption as essentially encrypted email. "What if I can't trust the other server at @bizarre_email_domain.org? Whatever."

  • Hey, at least thanks for having done your research on the topic :-)

    Re: "Signal technically cannot know your social graph" is more of "we, Signal, have got the information in our hands but we swear not to look at it". Essentially, your device is sending the data to Signal, and then the matching is done in a "secure enclave". One problem is that this step could totally be bypassed without your knowledge or consent. A second is that the technological underpinning of it (Intel SGX) has known unpatchable flaws. A third is that even if the build-up of your social-graph isn't stored initially, it can eventually be inferred from your usage patterns. A fourth is that even if you find good reasons to trust Signal today, they offer no definitive technological guarantee to enforce it in the future (the deal can change at any moment, being a non-profit isn't a guarantee either).

    I will also say that, in a decentralised communication system, you are reliant on every party you communicate with, and the tools they use, to not expose such data about you either. It’s not a panacea.

    No, in a decentralized system, you elevate your service provider to the same level of trust that you do today with Signal (with E2EE and maths taking care of the rest). The gotcha lies in the fact that you can be your own service provider in this case, or that you can establish other means to trust them (contractual, legal, moral, … obligations, that's up to you). And in the fact that changing service provider doesn't mean relinquishing all your contacts, histories, data, clients, etc…

    it’s utterly ridiculous to claim that this is not a way forward

    I don't disagree that Signal has some appeal over WhatsApp today. I only disagree that it represents a significant-enough step forward to justify having people massively migrate to it. From experience it is a doomed service that will deceive its users eventually (by design), and will cause more harm down the road (triggering another unorganized rush towards even worse services like Telegram) when it ultimately gets to this point. If you ask people old-enough to have known and used WhatsApp in its early days, they will depict a picture about as rosy as the one you paint today for Signal. All that to say, once again, that nothing is eternal. Especially in today's extremely consolidated internet (like, who would get in the way of Meta, Alphabet or Microsoft buying off Signal if they ever want to?).