cross-posted from: https://lemmy.dbzer0.com/post/55628224
I’ve been thinking about discovering underappreciated Lemmy instances. GitHub’s awesome-lemmy-instances used to serve a similar purpose, but it hasn’t been updated in a long time, and I haven’t found anything else like it.
I got the idea from this post about finding decentralized communities in the Fediverse. I’m thinking of a Lemmy bot that tracks Lemmy instances, calculates the average number of active users and standard deviation, and identifies instances with activity below the average plus two standard deviations. It would then rank these underutilized instances by performance metrics like uptime and response time, and periodically update a curated list on Lemmy to guide users toward instances that could use more participation.
I’d love feedback on how you would go about doing something like this. And specifically how to rank by performance.
One thing that comes to mind is that you might find a lot of instances with only a few or even one user because they are intended as personal mirrors.
I guess one way to counter this is to check whether registration is enabled.
A federation map might be intesting. Some groups of instances don’t federate with other specific groups. A visualization of that could show your bubble and islands of stuff that you would never see (for a good reason probably, but who knows without checking)
