On the Mastodon Experiment

Nov 15, 2022

Users are claiming to leave Twitter over their disagreements with Musk and switching to a decentralized (federated) alternative called Mastodon.

I played around with Mastodon in 2017, and my reviews were fairly negative. It was slow, inconsistent, and had a worse user experience than Twitter. However, I think the product makes interesting trade-offs relevant today in the context of other domains, i.e., crypto.

Some observations

  • Users don't want to host their own servers. Even the most technical ones. SaaS is the optimal solution. I'm an avid believer in running open-source software and controlling your own destiny, but for the vast majority of services, I don't want to manage it.
  • Moderation and user experience vary from instance to instance. It's hard to sustain at scale. In practice, decentralized moderation ends up as a series of fiefdoms without accountability (e.g., if you don't like it, find another server).
  • Chronological vs. algorithmic feeds. While many users claim they want chronological feeds, algorithmic feeds are clearly the future (see TikTok). More content means a greater need for curation. Curation works best at scale when data can be analyzed in aggregate. While Twitter's algorithmic feed is subpar at best, a chronological feed seemed worse.

Decentralize vs. centralization is a trade-off and brings a new set of problems in exchange for solving a different set.  It's an interesting experiment to see where users land.