Next post Previous post

Mastodon instance comparisons

May 26, 2025 at 10:43 AM by Dr. Drang

I forgot to mention in yesterday’s post another way (apart from looking at the FediDB directory) to check how much an instance is being used. That’s the activity API call. Running it on Fosstodon this morning via

curl -s https://fosstodon.org/api/v1/instance/activity | jq

I got

[
    {
        "week": "1748188656",
        "statuses": "1383",
        "logins": "2663",
        "registrations": "1"
    },
    {
        "week": "1747583856",
        "statuses": "16416",
        "logins": "6996",
        "registrations": "5"
    },
    .
    .
    .
    {
        "week": "1741535856",
        "statuses": "21653",
        "logins": "7844",
        "registrations": "24"
    }
]

The result was an array of 12 JSON objects, each with a week’s worth of activity.1 The week entry for each item is the Unix timestamp for the beginning of the week. Converting the first week value to a human-readable form through

date -jr 1748188656

I got

Sun May 25 10:57:36 CDT 2025

so the first set of activities represents only about a day, not a full week.

Running

curl -s https://fosstodon.org/api/v1/instance/activity |\
jq '.[1:].[].logins | tonumber'

which omits the number of logins for the partial week, returns

6996
7140
7212
7622
7498
7440
7490
7516
7666
7882
7844

So the activity on Fosstodon has gone down, and some of that may be due to the recent controversy. Let’s check this against Hachyderm, an instance that I’ve always considered comparable to Fosstodon. Here,

curl -s https://hachyderm.io/api/v1/instance/activity |\
jq '.[1:].[].logins | tonumber'

returns

6531
6539
6651
6653
6579
6631
6735
6731
6786
7187
7223

so its logins have also gone down a bit over the past several weeks. Plotting both timelines, we get

Login timeline comparison

So both instances have had an overall decline in logins over this period. The short-lived jump in Fosstodon logins four weeks ago might well have been due to the controversy.

Another way to compare the two sets of data is to plot the ratio of logins:

Ratio timeline

There’s about a 7% variation in the ratio over this period, but it’s hard to argue for a trend. Certainly not an exodus from Fosstodon.

I should mention that “Weeks ago” is an imperfect scale because Fosstodon and Hachyderm start their weeks about a day apart. As we saw above, Fosstodon starts its week on Sunday morning (Chicago time); Hachyderm starts its week on Monday morning.

I suppose I could have compared the statuses data instead of logins, but I care more about the number of people checking their timelines than how much they’re posting. What’s nice is that the API gives you the information to compare either one.

  1. The docs say it returns three months worth of activity, so presumably sometimes it will return 13 objects instead of 12. ↩

Next post Previous post