Mastodon’s search is restricted to hashtags and users only as a design choice. Full text search isn’t implemented on all instances and where it is only searches your own content, not that of others. The first impression I’ve seen from people not already familiar with the choice is that Mastodon’s search is simply broken, such is the expectation for true full text search across information systems generally.

The restriction means you’re absolutely dependent on conversations centring around a hashtag that is both consistently used by participants and that you know to look for. In practice a lot of conversations don’t work that way.

Content goes untagged, crucially because an originator has no reason to expect a post to turn into a thread when they start it. It not always possible to deduce what the hashtag might be on a topic you’re interested in, particularly when we’re talking about events occurring in real time. It gets even worse when we add regional dialects and different languages to the mix.

Some of this is addressable by having specific, disciplinary based scientific instances and hoping people use them. A planned structure of some sort so you could know where to look. There’s elements of that now.

Contrast this with Bluesky’s very powerful (if currently extremely painful to set up) published feeds system and I think Mastodon’s going to struggle a bit in this space. In my orbit (primarily social policy and support) Bluesky already seems to be winning out even behind the invite wall primarily because conversations and their participants are more discoverable and the control offered over their presentation at both an individual and community level.