cross-posted from: https://sopuli.xyz/post/12944261

The psychology of this problem is that users are too lazy to maintain multiple accounts when all they have is Lemmy’s stock web client. So they choose one of the big nodes: lemmy.world, sh.itjust.works, lemm.ee, lemmy.ca, etc.

These Cloudflare-centralized nodes are able to greedily exploit the #networkEffect because due to lack of multi-account software. If there were some well-made 3rd party client apps for Lemmy that would be designed for multiple accounts, then more users would be willing to create accounts in more decentralized parts of the fedi.

Mastodon somewhat proves this because the client-side tooling is in place to make it convenient to have 6 or Mastodon accounts. And Mastodon nodes are better balanced.

  • freedomPusherOP
    link
    fedilink
    arrow-up
    1
    ·
    edit-2
    6 months ago

    I don’t think those figures are trustworthy. I recall a page that tracks user counts which named some server I never heard of with a count an order of magnitude higher than lemmy.world. Might have been lemmyverse.net, not sure.

    Counting active accounts is a bit tricky I can imagine. So I judge by looking at the activity. Lemmy has ghost towns and 1 person communities which appear from the timeline like an announcement community but in fact they are open discussions where hardly anyone participates except the moderator. These are not niche topics either. It’s because users only want to manage one account. The stock web client dominates, which is inherently a one-account client. So the single most popular app fails to resist the gravity of the giant nodes. There is a paltry selection of 3rd party apps. Nothing in the Debian official repos.

    I went to phtn.app and just got a 500 error. So whatever that is, it’s probably not a significant factor here either way.

    In Mastodon threads I see more diversity of nodes people are coming from. Whereas when reading a quite active Lemmy thread you see something like 90% of comments coming from the top 5.