A few weeks ago I deleted my Mastodon account.

I can’t swear that I’ll never venture into the Fediverse again, either via Mastodon or some other ActivityPub-compatible service. And I’m not writing off either ActivityPub or the Fediverse as failures. I’ll probably dabble in the Fediverse some in the future. But for now, I’m pretty sure neither is for me. This post is a post-mortem of sorts, detailing why I found the whole experience somewhat dissapointing.

There’s a scene in HBO’s Silicon Valley where we find out that perpetual sort-of villain Jian Yang has been brainstorming blatant copies of existing successful websites. The whiteboard behind him has lots of suggestions, like “New Stack Overflow” and “New Facebook”. That’s pretty much the impression I get when looking at the various Fediverse platforms. There’s New Twitter, New Reddit, New Instagram, New Goodreads, New Twitch, New YouTube, New Meetup, and many more. Most of these do offer some improvements, or at least differences, from the walled-garden platforms they imitate. But it’s unquestionable that the prevailing motivation for Fediverse platforms is “What if (popular commercial website)… but federated?” I’d like to see more services that take advantages of the unique properties of ActivityPub to offer something that doesn’t have an obvious parallel in the commercial world.

Part of the problem with this imitative approach is the assumption that, because services are fragmented in the commercial world, they need to be fragmented in the Fediverse as well. I don’t know that there’s a technical reason why I need to sign up for different servers with different APIs in order to share short snippits of text, photos, videos, and so forth. And most of these Fediverse platforms will let you follow instances of other platforms; they all speak ActivityPub, after all, so they all work together. Sort of. But ActivityPub is deliberately somewhat vague, and each platform has the leeway to interpret certain federated data in different ways. As a result, my experience of following someone’s Pixelfed instance from another Pixelfed instance will be very different from my experience of following the same from a Mastodon instance. So if I really want to see someone’s content as they intended it to be seen, I will need to sign up to multiple services.

The lack of nomadic identity on most Fediverse platforms is a problem as well. If I have an account on one Mastodon instance, and I want to move to another instance, it is relatively easy to move over. I just have to export a bunch of CSV files on my old instance and import them into my new instance, and my followers, followed accounts, blocked accounts, and so forth will be moved over. My previous posts, however, will not. Plus, this moving-over process takes some effort on my part. It can’t happen automatically.

Why is this a problem? I discovered the downside of manual migration when a Mastodon instance where I once had an account suddenly disappeared. It’s not that the server was temporarily offline; it appeared that the domain name had expired, and no longer pointed at anything at all. There was nothing I could do in this situation. I couldn’t export my data, because I couldn’t log into the server. My followers wouldn’t be switched over to the new account, because for that to happen, my defunct old server would have to inform their servers of the change. About the only thing I could have done to prevent that situation would have been to regularly export CSV backups, and, in the event that my server died, manually create a new account somewhere else, import my most recent backups, and direct-message all my followers telling them to start following me at the new place. Or just start over from scratch.

What I wanted was a way to maintain linked accounts on multiple Mastodon servers: anything I posted on one would automatically be copied to the others, anyone who followed me on one would automatically be following me on the others, and posts would be seamlessly deduplicated so that people don’t see the same post multiple times, one for each server. If a server died without warning, anyone who followed me at that server would still be able to see my posts, because their server would know that the backup servers were associated with my identity and would pull the content from there instead. In short, my identity on the network would be decoupled from the server or servers on which I had accounts. Only after griping about this on Mastodon did I learn that the concept has a name – nomadic identity – and there are perpetual plans to add this to the Fediverse, but little has happened so far, for the simple reason that it’s a difficult problem to solve, especially given that it wasn’t baked into the ActivityPub protocol from the beginning.

But my real problem with Mastodon and the Fediverse had nothing to do with any technical shortcomings. It was this: despite the fact that the various Fediverse platforms have made efforts to correct the mistakes of other social media, the fundamental problem remains that they are social media. And the problems with social media are social, not technical, so no technical solution will ever fix them.

Through most of human history, there have been two basic ways in which communication occurred: Conversations among small groups of people, in which anyone could talk to anyone else; and mass communication, in which one or a few people address a large number of others, who, for the most part, cannot individually respond. The latter encompasses everything from an orator addressing a crowd, to a newspaper, to a television broadcast. A few people in a crowd can yell back to the orator, but not many can speak at once before their voices become an indistinct roar. One newspaper can publish a response to something written in another newspaper, or a newspaper can publish a handful of letters to the editor written by its readers, but there’s no easy way for any random newspaper reader to send a message to every other reader.

Social media changed everything by combining all the worst aspects of these two forms of communication. Now, anyone could reach anyone else. Conversations now happened in public: if I post something, and you respond to it, millions of people can witness our back-and-forth, and jump in at any time. This is unprecedented in human history, and I do not think we are psychologically configured to handle it. I have found that my go-to instinct when seeing a different opinion on social media is to view it as a confrontation. Whereas in a small group setting I can hear different views, ask for clarification, and try to find common ground, social media turns discussion into gladitorial combat, in which any disagreement takes place in front of millions of onlookers. This often encourages people to assume the worst, to avoid nuance, and to attempt to humiliate or “own” their interlocutors. This isn’t just some personality flaw of mine; I see this play out constantly online, and it is a very common, if not universal, reaction to the experience of participating in an always-on global forum.

I do not enjoy this experience. I do not think it is healthy for me, and I do not think it is healthy for others. There is still a lot of good and interesting content posted on places like Mastodon, but I have found that reading replies to posts, or – even worse – posting replies myself, almost always ends badly and increases my anxiety and distrust. Fortunately, Mastodon by default provides RSS feeds for users’ public posts, so I built a tool that can take someone’s exported follow list and create an OPML file that can then be imported into any RSS reader. This allows you to see Mastodon posts by people you follow without ever seeing any replies, or giving you the ability to reply yourself.

RSS (and its successor/competitor Atom) brings me to what I consider to be the golden age of the Web: the early-2000s heyday of blogs. I’m not alone in this opinion; you can’t swing a cat meme on the Internet without hitting someone lamenting the demise of blogging that everyone seems to think happened the instant Google killed Google Reader.1 What is it that made this era so much better than the present for online discourse?

Part of this is no doubt the selective memory of nostalgia. Lots of terrible and even dangerous content spread through the blogosphere, just as lots of wonderful content spreads via modern social media. But there were definitely differences between media then and now, differences which I think made it a better environment overall. I would argue that the heydey of blogging, prior to the dominance of social media, was the pinnacle of mass media, before mass media and conversation became so unfortunately combined.

In those days, it was fairly easy and cheap to start a blog. There were several platforms that would host yours for free, and Web-based interfaces that made it easy for even non-technical users to write. Once you did so, your writing could be viewed by anyone who had Internet access and wasn’t behind a dictator-imposed firewall. However, it wasn’t that common for people to talk back. Some blogs didn’t allow comments at all; if they provided a way for people to give feedback, it was in the form of an email address that you could use to contact the blogger. But that feedback was private and didn’t take place in full view of everyone who read the blog. When a blog did allow comments, it could introduce some of the public-brawl mentality that would eventually make social media so toxic, but even so, it wasn’t quite as bad. Pretty much every blogging platform gave the blogger complete control over comments: you could delete any comment, or block certain users from posting. You could run anti-spam software to weed out the bot comments that inevitably started cropping up. You could turn comments off entirely if things got out of hand, or if you just didn’t feel like dealing with it anymore. You could set comments to not appear until after you had approved them. Plus, most blogs just didn’t have really happening comment sections anyway. Even though any post could potentially be read by billions, in practice most blogs had a small readership, and even fewer non-spambots cared to comment. You didn’t get the pile-ons that often happen on social media, where someone searches for terms that they think are likely to turn up posts they disagree with, and then picks a fight with the poster for the benefit of their own followers. It was much more like a newspaper’s letters to the editor section than like social media.

Speaking of pile-ons, one of Mastodon’s most conspicuous and controversial differences from Twitter is the lack of a “retweet” (or “re-toot” or “repost” or whatever) function. This is a feature Twitter added that allowed a tweet to reference another tweet, with the optional addition of some commentary by the retweeter. It did not show up as a reply to the original tweet, and so it was often used when someone wanted to criticize another tweet without having that criticism immediately be visible to the original tweet’s audience. Someone would search for something they knew they’d disagree with, retweet it with “Get a load of this idiot!”, and all their followers would see it. Some of these followers would inevitably reply to the original tweet with the hopes of starting a fight, or else just drowning out supportive replies. The ultimate goal was a “ratio,” in which there are more replies than likes to the original tweet. (The idea, I suppose, is that people who write supportive comments almost always like as well, while detractors reply but do not like; therefore, a tweet with twice as many replies as likes must be expressing an unpopular opinion. In reality, ratios mean nothing except “This tweet, through some combination of malicious intent and opaque algorithmic machinations, happened to be exposed to more people willing to write critical comments.”)

Mastodon’s author Gargron long resisted a re-toot feature, fearing it would only add to Twitter-style pile-ons, though many users countered that retweets had been a vital way for people in marginalized communities to share information within those communities while not necessarily participating in the original discussion – such as a trans user retweeting someone spreading anti-trans misinformation, for example. But it occurs to me that the golden age of blogging, in which discourse was not always civil but seems to have been far better than any social media today, made almost the exact opposite decision. If I read something on another blog that I want to discuss at length, I simply write a post here and link to the original content. My blog has no comments (and if it did, they would be heavily moderated, by me, and me alone) so the only way anyone else can respond or add to the discussion is to write about it on their own blogs and link to my post. In other words, Mastodon allows replies but not reposts; blogs allow reposts but not – or not many – replies.

I don’t know whether the golden age of blogging will ever return even with the widespread disilluionment with Twitter and Facebook. I doubt there will ever be the “year of Linux on the desktop” either, but that doesn’t stop people who prefer that sort of thing from running Linux on their desktops. So I will continue to blog, and to read other blogs, and to not allow comments, and to not read comments on other blogs. And that gets me everything I really want from social media without most of the things I hate. No ActivityPub necessary.

  1. This narrative, despite its popularity, is laughably false. RSS both predated and survived Google Reader, and remains incredibly popular; it’s the technology behind podcasts! Lots of people still blog, and it’s not that hard to find RSS feeds worth following. There are a lot of complex reasons why so much online writing moved into the walled gardens of social media, or onto mailing lists, but I doubt the demise of a single feed aggregator, no matter how popular it was, is even in the top ten reasons why blogging doesn’t seem to be as big as it was in 2010. If anything, Google Reader being killed was a symptom of blogging’s decline more than a cause. ↩︎

Last modified on 2023-08-31