Does the community subscriber count only show statistics from that instance? For example, I have accounts on two servers and I wanted to find popular communities across the fediverse. For the most popular community, two different instances show different numbers for subscribers and users/month. For example, the top listing for these two instances is this:

https://lemmy.world/communities?listingType=All&page=1

[email protected] 10.4k 12.9k

https://lemdro.id/communities?listingType=All&page=1

[email protected] 34 8.48k

  • PriorProject@lemmy.world
    link
    fedilink
    English
    arrow-up
    2
    ·
    1 year ago

    Interesting, I feel like that active-users column is new in lemmy v0.18.x, I don’t remember that being there and I didn’t initially read your post clearly enough to understand that bit.

    Here’s my best-guess at what’s going on. It’s purely speculation, but speculation informed by study of how lemmy does other things in general and also by lots of exposure to system designs:

    1. The subscriber count works as I described in my last post. It’s a db query that fetches the count of local subscribers and is therefore… pretty much totally useless.
    2. The active-user, post, and comment counts are calculated based on the local db state also. Now this generally SHOULD be useful. Posts and comments are supposed to be federated, and active user counts should be locally derivable simply from the posts and comments themselves. So these figures should generally match.

    They obviously do not though, so what’s the skinny?

    1. For post and comment counts, lemmy.ml has post and comment history back to the beginning of time. Whereas lemmy.world and lemdroi.id are both relatively new servers. When the first user on an instance subscribes to a community, their server tells the instance’s server “I want you to send me a copy of all new posts and comments going forward… also send me the most recent dozen or so just to get me started”. There’s no historical backfill. So you don’t expect to see post and comment counts match. I guess that makes them useless.
    2. For any server more than a month old, you SHOULD expect to see active-user counts match. Unless there’s some rolling average that takes a few months to settle (which is possible). But if that’s not what’s up, then I think we’re seeing side effects of broken replication: https://github.com/LemmyNet/lemmy/issues/3101. Perhaps lemdro.id is missing actual posts and comments and therefore cannot calculate the active-user counts accurately.

    In any event, the workarounds are the same…

    1. Browse communities on lemmyverse.net, not in the native community browser. It will give comparable activity numbers for communities on all servers.
    2. Visit the home instance for each community, and only use the numbers there: https://lemmy.ml/c/asklemmy. That gives a third set of numbers that are bigger than any of the other two, and the most reliable. If the community’s instance doesn’t know about a post… it didn’t get federated anywhere so it mostly really doesn’t exist (except maybe on the instance of the user that posted it).

    This is all kind of a wreck, I’m not trying to defend any of it. Just trying to explain why it might be that way. I have basically decided that the lemmy community browser is not useful except to look up a specific community and subscribe to it. Otherwise I always use lemmyverse or the community’s instance to find info about it.