Participation in the Fediverse

Parts of the fediverse have been in something of an uproar recently over an experimental search service that was under development called (appropriately enough) Searchtodon. The project aimed to enable people to search their own home timeline and worked by being authorized by a user to access that user’s timeline via the ActivityPub API and downloading a copy of each post that is visible to that user. The service then indexed the posts and gave the user tools to search through the history of their own timeline.

This project generated a large amount of feedback, some positive, and some very negative. The author of the project has since taken down its website and replaced it with a retrospective that is worth reading. There’s a lot that Searchtodon did right, and the author was explicit about wanting to make sure that the capability “gels with the community” and, when the feedback received made it clear that it did not, they pulled the plug on it.

Some have looked at the response to Searchtodon and saw in it a disproportionate attack and even a pile-on (though the Searchtodon author themselves does not say this anywhere). It seems like there’s a good chance that Searchtodon got caught in the crossfire of a larger sentiment within the fediverse community against making the fediverse broadly searchable, even though that wasn’t quite what Searchtodon was doing. That sentiment is based on some of the cultural realities that drove early adoption and growth of the fediverse; there are deep seated suspicions of features that might seem innocuous in other contexts. Search is one example. “Quote tweets” is another. In particular, there is a widely-held feeling that network-wide search (that would let you pull up every post mentioning an arbitrary term across every server in the federation), while certainly useful in some circumstances, enables and promotes bands of roving trolls jumping into any conversation about a given topic to harass people. The victims of this harassment, unsurprisingly, tend to be people of color, LGBTQ+, and women.

What others saw as a pile on, however, seems like it might be something much more interesting: participation. For what may be the first time, the users of the social network have tangible power over how the systems in which they operate are run. In other circumstances, this sort of outcry would have been screaming into the void, but here it was a very real feedback force, and one made with the expectation that it would be heeded.

A few things stand out about this episode: First, this direct engagement by the actual people affected with how the software that operates their community will run is a good thing. On incumbent centralized services, you are just a user. You can complain and protest, but you aren’t invited to the table where impactful decisions are made. These services ask you to accept the changes or leave. In the fediverse, though, people have more options, including changing your home instances and blocking others. Democratic processes are not always pretty, as anyone who has sat through a town alderman or city council meeting can tell you, but it is the best form of community governance we know of.

Second, for a long time the tech industry has had a tendency to perceive all problems as having technical solutions. (This service doesn’t have search? No problem, we can add search!) In reality, some problems only have social or policy solutions. Put another way, just because something is technically possible doesn’t mean that it’s socially desirable. It is good to see that fact reiterated.

Finally, it must be said that the Searchtodon post-mortem is a great example of analyzing with an open mind what went right, what went wrong, and how the community might find a path to having the home timeline search that some want without causing the downsides that have led to the general opposition to full text search across the entire network. This kind of participatory policymaking is about experimentation as well, and iteration when something doesn’t work.