this post was submitted on 10 Sep 2026
25 points (93.1% liked)

Fediverse

43850 readers
534 users here now

A community to talk about the Fediverse and all it's related services using ActivityPub (Mastodon, Lemmy, Mbin, etc).

If you wanted to get help with moderating your own community then head over to !moderators@lemmy.world!

Rules

Learn more at these websites: Join The Fediverse Wiki, Fediverse.info, Wikipedia Page, The Federation Info (Stats), FediDB (Stats), Sub Rehab (Reddit Migration)

founded 3 years ago
MODERATORS
top 9 comments
sorted by: hot top controversial new old
[–] Zen_Shinobi@lemmy.world 9 points 17 hours ago (1 children)

I think Lemmy and Piefed are handling the spammers and ai reposts well enough. I'm also on Discuit and it's ramped with reposters, ai bots, meatspace proxies, and I assume bot upvotes, which is sad.

[–] Hal@piefed.nz 7 points 12 hours ago (1 children)

Lemmy definitely has LLM bot accounts, but the Lemmy is operating at a scale of tens of thousands of users where Mastodon has a couple of million. There's a lot of manual action by Lemmy admins that probably won't scale well when there are millions of users.

[–] Cris_Citrus@piefed.zip 1 points 9 hours ago (1 children)

Only if that scale happens across the current number of servers. If the server base grows along with users the manual management may scale much better

[–] Hal@piefed.nz 2 points 8 hours ago* (last edited 8 hours ago) (2 children)

Unfortunately Lemmy has one major problem when it comes to banning spam bots. Only the instance that a user belongs to can ban the user in a way that federates to other instances. This means if that particular instance doesn't quickly ban the user, then hundreds of other instances need to do it manually. Currently, a pretty significant percentage of spammers get banned individually by many instances.

If we had 100x as many users and 100x as many instances, we would have 100x as many bot accounts for every instance to ban.

The solution is probably something like shared ban lists or ban mirroring (setting an instance up to ban every user that some trusted instance bans). But right now it's just people clicking, and it's not scalable.

[–] rglullis 2 points 7 hours ago (1 children)

This could be mitigated quite easily by having a clear separation between instances run for users and instances for communities.

[–] Hal@piefed.nz 1 points 6 hours ago (1 children)

Can you explain some more? I don't get how this separation helps unless you keep a small number of community instances.

[–] rglullis 1 points 4 hours ago

Yes, I think you get it. My assumption is that there would be a lot less of "community instances" compared to "user instances".

[–] Cris_Citrus@piefed.zip 2 points 8 hours ago

Hmm, yeah that seems like it does need a technical solution

Thank you for taking the time to explain! ☺️

[–] corsicanguppy@lemmy.ca 0 points 8 hours ago

a [...] software

That word doesn't work like that. It's not "a software" any more than it's "a happy". It's just "software" without the indefinite article.