On Twitter and Mastodon: First Thoughts

I have, unfortunately, felt it necessary to establish a Mastodon account. You can find me at @ravenonthill@mastodon.social.

So Elon Musk now owns Twitter. He has started his tenure by firing the CEO, the CFO, the head of legal policy, trust, and safety, and the general counsel. He is reportedly planning on firing 75% of Twitter’s staff. He has loaded Twitter with $1 billion in debt. This is a recipe for wrecking a business. Dr. Caroline Orr Bueno comments (on Twitter!):

So basically Elon Musk is planning to make Twitter profitable by firing nearly *75 percent* of Twitter’s employees, including most of the people who know what to do to prevent security breaches and how to mitigate them when they happen. That’s a serious national security threat. This is probably about to be an incredibly costly, dangerous, and totally predictable lesson in “no, AI can’t just replace humans.”

So while I wait to download my Twitter archive (if that is still allowed) here’s some thoughts about Mastodon as an alternative. My main thought, unfortunately, is that it isn’t one, or isn’t a very good one. Mastodon, for people who aren’t familiar with it, is part of a distributed network of Twitter-like services. The source code is entirely open and owned by Mastodon gGmbH, a German non-profit whose CEO is Eugen Rochko, the original developer of Mastodon. As with the Free Software Foundation, it is not clear what will happen to Mastodon should Mastodon gGmbH or Eugen Rochko become corrupt. Mastodon is written in the interpretive object-oriented language Ruby. My quick research on this suggests that any Ruby implementation is too slow to accommodate as many users as Twitter’s system; Twitter itself found it was necessary to extensively rewrite parts of its Ruby server code in Java and Scala, at what must have been huge levels of effort and frustration.

The theory behind Mastodon and the various “federated” services – the “Fediverse” – is that of a decentralized network made up of a large number of semi-independent nodes. The original model of these was probably one of the original social networks, Usenet, which was created by people to get around lack of access to the then-nascent internet. It used its multiple-node network to overcome limitations of computing power and bandwidth. Usenet is still around, but is quite minor. Its user interface was not updated to modern standards, and its distributed model turned out to be vulnerable to trolls and spammers; the term “spam” was popularized on Usenet.

And this is the model that the Fediverse perpetuates. In principle, modern authentication techniques could control the spam and the trolls. In practice, it is left to the management of each Fediverse site to do so, and the job is often done badly. So the Fediverse is utterly vulnerable to trolls and spammers, well-heeled corporations and state actors. It is rotten with Chinese and Russian bots and propagandists and policies of blocking lead to loud cries of racism. The Fediverse is also a refuge for dissidents, but they must be very, very careful; because of the limitations of authentication on Mastodon there is no reliable way to know if any contact is in fact a government agent.

I will go on to discuss the ideological and technical problems of the Fediverse network in subsequent posts.