Search

Friday 25 November 2022

Mastodon

The last couple of weeks have seen the emergence of a new sub-genre in the press coverage of social media: my struggles with Mastodon. While not uniformly negative, these tales have relished the authors' cluelessness in the face of baffling concepts like "federated" and "instance". This is partly the traditional journalistic pose: you, dear readers, are idiots so I will perform idiocy as a way of explaining this new thing to you. But it also points to the underlying hatred for social media (which is really a hatred for democracy) that runs like a thread through newspaper coverage of the subject, and which is by no means limited to Twitter (the journalists' drug of choice), even if these articles inevitably serve as yet another excuse for linking the bird site in particular to the "toxic slurry" that supposedly disfigures our society. It becomes clear reading this sub-genre that what journalists really want is an environment that reflects their own self-worth. Unsurprisingly, some competitors to Twitter have sought to emphasise this, e.g. Post claims to be a platform for "Real People, Real News, and Civil Conversations". If that doesn't attract the liberal scolds then nothing will.

It's therefore amusing that Mastodon should have become the early leader in the field of alternatives to Twitter, given that it isn't particularly user-friendly and its politics (or at least those of its creator, Eugen Rochko) appear to be to the left of your average journalist. This is partly because so many other challengers have been promoted by the right, e.g. Parler and Truth Social, but partly because the politico-media caste have never really wanted to leave Twitter. Their fundamental criticism was not that it didn't do enough to counter abuse - it should be obvious now, as Elon Musk dismantles much of the moderation regime, that it actually did a good job in keeping a lid on it - but that it still allowed ordinary grunts on the platform. What they wanted was a more exclusive environment in which they could praise each other's "brilliant" articles or issue vapid press releases without getting laughed at by the great unwashed. Of course the problem with such exclusivity is that it can backfire, as the hilarious tale of journa.host reveals (in a nutshell: journalists set up their own instance; lots of other instances immediately defederated because they didn't want their content being scraped for stories).

Mastodon isn't going to supplant Twitter and the reason for that is as much to do with the technology as the cultural values. It is a halfway house between a fully distributed social network and a cloud-based common service. Conceptually, the former would require every user to provide and administer their own server. The point is that there would be no shared infrastructure beyond the Internet itself - i.e. the protocols that allow you to send and receive messages over a network. In practice a fully distributed system is constrained by the need to provide servers. Most users won't want the cost and hassle of providing their own, and they certainly don't want the administrative overhead this would entail. In other words, what Mastodon provides is both a collection of servers and (crucially) a devolved community of administrators. The rapid growth in the user base is causing the former to buckle under the strain, though this can be alleviated by throwing money at the problem and upgrading capacity, but the real choke point is the latter. Who wants to spend their free time reviewing sign-up requests and moderating content?

We've been here before. Mastodon is esentially a modern version of Usenet, the distributed bulletin board system that dates from the early 1980s. The similarities between Usenet groups and Mastodon instances, notably the discrimination over which peers to share feeds with, should be obvious. It should also be obvious that social media developed in part as a solution to the limitations of that older technology, not least the relative lawlessness of its group moderation (by the end, it was mostly being used for illegal file-sharing). Once access to the Internet was cheap and reliable, services could be consolidated "in the cloud", but this entailed a centralised administration as well. To pay for that, most social media services mixed advertising in with the user-generated content, which in turn created the incentives for surveillance and opaque algorithms. There's recently been a move towards more subscription-based services (e.g. Substack), but that has been driven more by the desire to monetise user content than to find an alternative to advertising. 


Mastodon cannot service a population the size of Twitter's near 400 million (it currently has less than one percent of that figure). The announcement by celebrities like Stephen Fry that they have moved over should alert us to the fact that its new users are overwhelmingly unlikely to want to provide their own servers or conduct content moderation, even if they knew how to do so. The "fediverse" depends on a community of nerds, and for those nerds to be a significant percentage of the total user base, willing to dedicate their own equipment and time to the common good. What's happening now is that Mastodon is attracting the jocks. To service this larger community individual instances will need to add both computing and moderation resources, and that ultimately means charging users. Given that the jocks are used to free stuff, this is either going to kill the growth or lead to Mastodon abandoning its principled objections to advertising and therefore surveillance. That is a bridge too far so pleas for donations are likely to dominate for the next few months. Ironically, the much foretold death of Twitter may be pushed out of the news shortly by the sudden death of multiple Mastodon instances.

As others have noted, the volume of users and posts that a celebrity brings in their train can amount to an unwitting denial of service attack for a Mastodon instance near the limit of its server capacity. That many instances now appear to be blocking or at least slowing new sign-ups is hardly a surprise. This highlights that the "cloud" has long served to obscure the fact that a person's online social network is not simply a matter of follower numbers but of linked activity, and that activity has a material cost in terms of the infrastructure required to support it. The blitheness with which media personalities announce that they're moving from Twitter to Mastodon (or to other platforms) is as good an example of dumb privilege as you can find online these days. When celebrities moan about their fans' presumption ("They think they own a bit of me") they forget that this proprietorial attitude cuts both ways. Imagining that your followers will loyally troop over to another platform ignores that a social network is not the product of your person but an emergent property of the medium. 

The growth of Mastodon, and the challenges faced by users in trying to join and navigate it, may create an opportunity for someone with a more commercial mindset than Rochko. Just as tools like TweetDeck sought to make Twitter more user-friendly (before being absorbed into the product), so we are likely to see a flurry of interfaces that seek to "simplify" Mastodon. In seeking to mask the limitations of the platform from "ordinary" users, it will make technical sense to consolidate this interface layer, creating a cloud-based superservice that will function pretty much like Twitter. But because of that consolidation and user affordance, it will face the same issues of moderation, regulatory observance and the need for a revenue stream to pay for its infrastructure. What this should remind us is that what people like about Twitter is the fact that it isn't federated: that it provides the serendipity of "weak ties" rather than the claustrophobia of a parochial interest group. That so many users find the "fediverse" baffling should be a clear indicator of that.

Twitter has barely changed since Musk bought it. The company may have seen upheaval aplenty in its personnel and working culture, and you can't rule out the possibility that bits of the system are going to start failing, but the idea that users would be driven away is clearly nonsense. The liberal media's critique is essentially that they'd prefer another owner - less rightwing and less vulgar - but they actually share Musk's vision of a social space that divides the fee-paying sheep from the goats: that is an environment they are familiar with and value. More democratic platforms like Mastodon are not what they want, and you can expect journalists who persist with it to soon take umbrage when administrators moderate them rather than the uncredentialled. The Mastodon sub-genre of newspaper comment is likely to prove short-lived but its demise will probably herald the arrival of another navel-gazing sub-genre: why I've come back to Twitter.

7 comments:

  1. Do you have a Mastodon account of your own yet, and if so then what is your handle?

    ReplyDelete
    Replies
    1. I don't, and I have no current intention of getting one.

      Delete
  2. While Mastodon and Usenet are similar in their decentralized nature, aren't they very different in that Mastodon (like Twitter) is a general-purpose microblogging platform, while Usenet is divided into newsgroups based on subject matter?

    The centralized corporate platform corresponding to Usenet wouldn't be Twitter: it would be Reddit (with its subreddits corresponding to Usenet groups).

    ReplyDelete
    Replies
    1. I wasn't comparing Usenet and Twitter but specifically comparing the decentralised structure of Usenet and Mastodon. My point is less about the techncial architecture and more about the moderation problems that this gives rise to.

      Delete
    2. A more detailed discussion on the logic behind Mastodon et al can be found at Mike Masnick's 2019 article Protocols, not Platforms.

      Delete
  3. It looks like one important design decision made by Mastodon (which AFAIK isn't dictated by its federated architecture) is that the plain text of its toots (equivalent to tweets on Twitter) isn't searchable: searches can only be for usernames and/or hashtags.

    I wonder if this was down to a concern that (even without centralized social media corporations) allowing general text searching is too facilitating of a surveillance society (with all the pathologies -- eg "cancel culture" -- that go with that): it was a point that Charlie Stross made with respect to DejaNews (a now-defunct search engine for Usenet archives) back it the '90s.

    ReplyDelete
    Replies
    1. The simplest explanation is that it's a technical constraint. Usernames and hashtags, by their nature, will be indexed. Searching large sets of unstructured data is expensive. Trying to do so across a fediverse would be prohibitively expensive.

      Delete