I've had a little process running for the past year or so, totting up Mastodon "active monthly users". Occasional bumps up (one outstanding one last summer when Elon did something particularly idiotic) followed by long slow declines.

Was somewhat disappointed to see recent growth of about 250k turning out to be a phantom instance (telling lies via the API).

#mastostats #mastodon #api #fediverse

It started of as an exercise to attempt to sample from the population of Mastodon instances, but developed into an attempt to list and then scan all active instances. There is similar info available from https://api.joinmastodon.org/statistics but this uses an independent method (in particular, using a snowball method to identify (nearly) every working instance).

The number of Mastodon hosts or instances I find doesn't vary enormously over time, 10-11k total. However, most of these have very few active users. If we restrict according to the maximum number of active users observed, we are down to about 2700 hosts who have at least 10. Only about 760 hosts have had more than 100 active users.

(Note log scale, which tends to de-emphasise small temporal changes.)

Here's a histogram of host size, today. What it mostly shows is how crap histograms are for data like this.
Cumulative size gives a nicer view than the histogram (a Lorenz curve?), and even more readable with a log scale.