1. 16

  2. 22

    Interesting topic, but thoroughly meh article. It’s got the same ‘splain of the basic concept that anyone can find on any federated site’s marketing copy, then Drew don’t like ActivityPub (OK, fine, but why?), then rah-rah grab-bag of federated protocols, then… nothing. Unrealized potential for a substantive critique.

    1. 7

      I would be fine with not posting every article from Drew here. A lot of them are great, some of them are.. meh. But he’s got celeb status so I guess everything he posts is ‘on-topic’ here? (I flagged this as ‘off-topic’ because of how shallow it is..)

      1. 2

        That seems like a misuse of the “off-topic” flag to me. A little teasing to elicit a substantive critique (or at least some discussion of the topic, as distinct from the article) from the community here still appears to be a more effective tactic. Maybe Lobsters hasn’t entirely gone to the dogs just yet.

        1. 3

          Eh. I would argue that it’s an off-topic piece posted here merely because the author’s fame. This article doesn’t even try to do any of the following:

          Some rules of thumb for great stories to submit: Will this improve the reader’s next program? Will it deepen their understanding of their last program? Will it be more interesting in five or ten years?


          1. 1

            Fair enough. I prefer to judge content by its own merits and not by the author’s reputation, but it takes all types to make a forum. I do think that if you compare the boss’s guidelines with what the community here actually engages with, you will find some discrepancies, though.

    2. 7

      Federation is not enough. You need an answer to network effects and the convenience offered by major players. Look at what Google and FB did with XMPP, and what slack did with its bridges - you need to make it impossible for a big player to acquire users via federation and then wall off the garden.

      Some reward for self-hosting or for being a participant on a niche node could be cool, but how do you do that without opening up to Sybil attacks? Bitcoin solved this with scaling difficulty and HashCash, but maybe there are better techniques these days that can be backported from the cryptocurrency space?

      1. 3

        you need to make it impossible for a big player to acquire users via federation and then wall off the garden

        Agreed. This article is surprisingly naive in my opinion, and ignores the reality of basically all federated protocols.

        1. 2

          This is also one of my main gripes with Matrix. If I tell someone I can be found on Matrix, chances are 10:1 that they’ll just create an account on the matrix.org main server. And AFAIK they’re not even trying to centralize everything on their server!

          The most popular client, Element, makes this the default and I think there’s no registration workflow for other home servers anyway. User creation on Synapse is hopelessly complicated (I always forget how because I so rarely create accounts for people) and then using a different home server requires some settings in Element which I also always forget.

          The overall experience is also much shittier if you’re using the matrix.org home server, because it has so many users. Everything is rather slow with it, compared to my own home server where everything is snappy because I have only a handful of people on there.

        2. 5

          Part of my answer is that focusing on technology underpinnings is the wrong question, but that’s a half-story. The rest the reason I think OStatus is on the wrong track:

          I need to be able to transparently migrate service provider. I should be able to switch from one service provider to another, and, after doing so, should not have to worry about the old provider any more. Such a feature makes it possible to take risky bets on service providers, and if it doesn’t work out, it’s not a permanent loss. People need to be able to make such risky bets, otherwise they’ll only register with service providers that they know will be reliable, and that means big services get bigger and small services can’t get off the ground.

          That means domain names can’t form part of my permanent identity. If it looks like notriddle@example.com, then it means I’m now chained to example.com, and even if I decide to switch to another providers, I either get to send a message to all my contacts asking them to change, or I get to worry about example.com being able to host a permanent forwarding address. I can also own my own domain, but what happens if I screw up and it expires?

          1. 2

            domain names can’t form part of my permanent identity

            Well.. the alternative is cryptographic keypairs as in SSB. But this makes multi-device hard, and making keys manageable by normal people is also hard (I guess deriving keys from passphrases is convenient but then the hard problem is recovery from loss/theft).

            I can also own my own domain, but what happens if I screw up and it expires?

            Don’t screw up? People have been owning personal domains for a couple decades now, with good results mostly.

            Seems like the bigger barrier to mass adoption of personal domains is having to pay money at all in a world where most online communication has been free :/

            1. 1

              In your version you don’t actually really belong to a federation instance anymore. It acts more or less like a cache. I think this is a cool idea but doesn’t really match what most people mean when they discuss federation.

              It’s probably difficult to form a sense of community in this case or to set up an instance for your friends since people don’t belong to an instance in any meaningful sense.

              Conceptually, it seems more like p2p to me.

            2. 5

              Unlike P2P systems, the federated model allows volunteer sysadmins to use their skills to expand access to the service to non-technical users, without placing the burden on those non-technical users to set up, understand, maintain, or secure servers or esoteric software.

              We would be better served educating non-technical users than enabling them.

              There is no end of trouble for choosing to cultivate Eloi; we’ve learned this in the past three decades.

              1. 4

                I’d love to see research into solving the hard problems of federated services – like social networks.

                This 2017 article summarizes my skepticism – https://www.wired.com/story/decentralized-social-networks-sound-great-too-bad-theyll-never-work/

                I’d love to be proven wrong

                1. 4

                  It’s interesting that this article fails to define success until the second to last paragraph of the article - tackle the problem of platform monopolies.

                  This is like saying “Linux will never replace Windows on the desktop” or “Instagram will never replace Flickr for serious photographers” or “Racket will never replace C++” You’re casting the platform as a challenger for dominance in a well-established space, and of course it will never win against the incumbent. But that doesn’t mean it can’t be successful in other ways.

                2. 3

                  This feels naive. Federation and decentralization answer a question that nobody is actually asking, and the compromises they require, especially around addressability and discoverability, are both fundamental and kind of intractable.

                  1. 3

                    a question that nobody is actually asking

                    There are problem sets that decentralized apps solve very well, but the problem is that they’ve been applied to areas where they aren’t needed just because they’re sort of “cutting edge”. This is probably true of lots of technologies, though. I imagine the same type of improper application of a new tech happens just as frequently in, say, materials science.

                    1. 3

                      Nope people are absolutely asking questions about centralized providers of software services and identities, where decentralization and federation are potential solutions to the problems people have with those services. That’s not to say that decentralization and federation don’t introduce their own problems, but it’s certainly not the case that the compromises necessary for centralization are obviously superior to the ones necessary for decentralization (and I think they are manifestly worse).

                      1. 2

                        I mean, this also feels naive. I’m sure the set of people asking these questions isn’t empty but, like, is it statistically significant? I don’t think anyone who isn’t actively involved in building the decentralized web is really thinking in these terms…

                    2. 3

                      I don’t understand how spam is a reason for federation. Spam is exactly the kind of problem that you get because of federation since there is no central way to reprimand users. If you block an entire server you are pushing around the responsibility and punishing people who don’t deserve it.

                      1. 1

                        If you block an entire server you are pushing around the responsibility and punishing people who don’t deserve it.

                        Not true. You are pushing the moderation decision on the server operator. They have the option to punish the individuals on their server that have triggered the action in the first place, and then be reinstated, or not. The good faith users of the server should have the option to lobby for punishing of the bad faith ones.

                        I feel like this system is better for everyone, as it allows small groups to police themselves, instead of relying of an uncertain central authority.