3 ms·

Everyone is approaching the social media issue as if it was a technical problem - it's not. No amount of decentralisation or protocols address (or even attempt)

Everyone is approaching the social media issue as if it was a technical problem - it's not. No amount of decentralisation or protocols address (or even attempt) the root cause. The closest would be the crypto-based social networks, which while they have their own problems at least attempt to address the issue of funding the platform. You know it's bad when the closest thing to a solution comes from crypto grifters. Same issue with the rest of the "alternative" world, whether it's OSes or software. Lots of time spent on technicalities or ideologies, zero time spent on addressing the actual problem - that's why the "year of the Linux desktop" is still a recurring joke. The problem with social media right now is the lack of a non-adversarial, sustainable business model. All these changes stem from the fact that advertising-based business models are on their last legs and are fundamentally flawed because they are adversarial to the users - the Twitter API shutdown is at least partly because they want to drive everyone to use the official client where it's easier to impose user-hostile functionality. Decentralisation merely side-steps this problem which works on a very small scale but not only will break down at a larger scale (operating a social media platform costs money) but also brings a lot of its own issues. Part of the appeal of a social media platform is its popularity, network effects and a sense of community where most people are happy with or at least tolerate the rules and moderation policy. A Mastodon-powered future will have 2 outcomes: 1) every instance federates with everyone and the entire thing becomes flooded with spam and other unsavoury (or outright illegal, at least in some jurisdictions) content because there is no common moderation policy. Users eventually get fed up and leave to a centralised competitor. 2) instances federate on a case-by-case basis which fundamentally breaks network effects and makes global conversation and community building impossible. Good luck explaining to non-technical users why they can't see/interact with the same posts as their friends because they happen to be on different instances that don't federate with each other, or because the content they both want to see is on a separate instance that doesn't federate with theirs. Users get fed up & leave or can't get started to begin with and sign up on a centralised competitor instead. In both cases I haven't even addressed the issue of funding the network itself - there is still no business model (and any business model where users pay would require the service to have enough value for them to begin doing so - chicken & egg problem when the value of a social network is in its network effects), and even if there was, it will be more expensive because decentralisation requires a lot more system resources.

tptacek 4y ago
No. You're still thinking about Mastodon as a federation of medium- and large- sized small social networks; star systems in a galaxy. That's how it works now, but I don't think it's going to stay that way for long. I don't think large shared Mastodon instances, at least larger than you'd see for a group blog, make much sense. It doesn't make sense for all the reasons you're spelling out: the amount of effort to keep those big instances going is expensive, and there's no real funding model for it, and beyond that there's not enough win to justify the work. But a "network" of hundreds of thousands or millions of tiny instances, maybe paradoxically, doesn't have this same spam/trolling problem, any more than RSS did when we were all using Google Reader. You see posts from people you follow. Because most people aren't on big shared instances, you're not pulling lots of random traffic in to begin with. Again: think of this like Blogger and Google Reader mashed together, with Twitter's surface-level user experience. I think it'll work fine, and I think it does most of what Twitter did (and some additional things). People forget that the "blogosphere" worked pretty great before Twitter and Facebook wrecked it. I think there's a good chance we're heading back towards something that works like blogs did.
- Nextgrid 4y ago
  > But a "network" of hundreds of thousands or millions of tiny instances, maybe paradoxically, doesn't have this same spam/trolling problem, any more than RSS did when we were all using Google Reader. RSS/Google Reader was mostly a "pull" medium. Creators create content while consumers passively consume it but can't write back. This works but is fundamentally different from Twitter and the modern concept of social media. Furthermore, RSS offloads all the work to the clients, so RSS servers don't have to store nor process any state regardless of how many consumers there are (you can serve the same cached response to all of them). If you want it to be like social media, you need people to also be able to "push" things (that others who don't directly follow you see) so people can reply and comment and do social-media-y things. This then not only opens the floodgates for spam, but also makes everything harder if every instance has to federate with tens of thousands of small ones instead of a few hundred like we currently have (the current system works fine with a "hosts"-like file manually curated by instance admins, but if instances start proliferating you'd need a DHT of sorts so they can all discover each other, and again need some way of vetting the quality of them so you don't get spam into it). This turns instances into hosts not just for the original content of the instance admin but also for the replies of people, again raising both server resource issues as well as issues relating to spam, illegal content and moderation in general. > think of this like Blogger and Google Reader mashed together, with Twitter's surface-level user experience Blogger could potentially pull it off because it's a closed system that has one consistent moderation policy. It doesn't have to broker or host unvetted third-party content. But blogging/publishing is fundamentally still a different medium to something like Twitter. > People forget that the "blogosphere" worked pretty great before Twitter and Facebook wrecked it. I think that the blogosphere and social media are fundamentally different mediums. Blogging is write-once, read-many, with minimal, mostly consumer-to-publisher interactions (comments on the blog itself). For a lot of people, social media also involves direct consumer-to-consumer communication, merely adjacent to the publisher's content but completely outside of their control. Your proposal does not address this use-case.

tptacek 4y ago

No. You're still thinking about Mastodon as a federation of medium- and large- sized small social networks; star systems in a galaxy. That's how it works now, but I don't think it's going to stay that way for long. I don't think large shared Mastodon instances, at least larger than you'd see for a group blog, make much sense. It doesn't make sense for all the reasons you're spelling out: the amount of effort to keep those big instances going is expensive, and there's no real funding model for it, and beyond that there's not enough win to justify the work. But a "network" of hundreds of thousands or millions of tiny instances, maybe paradoxically, doesn't have this same spam/trolling problem, any more than RSS did when we were all using Google Reader. You see posts from people you follow. Because most people aren't on big shared instances, you're not pulling lots of random traffic in to begin with. Again: think of this like Blogger and Google Reader mashed together, with Twitter's surface-level user experience. I think it'll work fine, and I think it does most of what Twitter did (and some additional things). People forget that the "blogosphere" worked pretty great before Twitter and Facebook wrecked it. I think there's a good chance we're heading back towards something that works like blogs did.

Nextgrid 4y ago
> But a "network" of hundreds of thousands or millions of tiny instances, maybe paradoxically, doesn't have this same spam/trolling problem, any more than RSS did when we were all using Google Reader. RSS/Google Reader was mostly a "pull" medium. Creators create content while consumers passively consume it but can't write back. This works but is fundamentally different from Twitter and the modern concept of social media. Furthermore, RSS offloads all the work to the clients, so RSS servers don't have to store nor process any state regardless of how many consumers there are (you can serve the same cached response to all of them). If you want it to be like social media, you need people to also be able to "push" things (that others who don't directly follow you see) so people can reply and comment and do social-media-y things. This then not only opens the floodgates for spam, but also makes everything harder if every instance has to federate with tens of thousands of small ones instead of a few hundred like we currently have (the current system works fine with a "hosts"-like file manually curated by instance admins, but if instances start proliferating you'd need a DHT of sorts so they can all discover each other, and again need some way of vetting the quality of them so you don't get spam into it). This turns instances into hosts not just for the original content of the instance admin but also for the replies of people, again raising both server resource issues as well as issues relating to spam, illegal content and moderation in general. > think of this like Blogger and Google Reader mashed together, with Twitter's surface-level user experience Blogger could potentially pull it off because it's a closed system that has one consistent moderation policy. It doesn't have to broker or host unvetted third-party content. But blogging/publishing is fundamentally still a different medium to something like Twitter. > People forget that the "blogosphere" worked pretty great before Twitter and Facebook wrecked it. I think that the blogosphere and social media are fundamentally different mediums. Blogging is write-once, read-many, with minimal, mostly consumer-to-publisher interactions (comments on the blog itself). For a lot of people, social media also involves direct consumer-to-consumer communication, merely adjacent to the publisher's content but completely outside of their control. Your proposal does not address this use-case.