User keeps getting autosilenced by system. Not sure what's happening

Hi there. First time posting on Discourse Meta, although I read your content a lot, been helpful in resolving a lot of situations. I couldn’t find anything about this particular glitch though.

Essentially we had a power user who started to misbehave in private messages, so we dropped his trust level from 3 to 0 as part of the punishment (so he could no longer initiate private messages). This caused a whole bunch of weird glitches that are obvious in retrospect. A lot of his posts suddenly got auto-flagged by the system for excessive @ mentions and links, we disagreed with almost all of them.

Then the system auto-silenced him. So I unsilenced him. Then today, I got a note from this user that he was silenced again. Lo and behold, the system had silenced him again after I had unsilenced him. And I can’t figure out why it would do that, especially without notifying us via the moderation/admin panel. What are the parameters within which the system will automatically silence a user? I know there are restrictions on things such as links or @ mentions, but I want to make sure this kind of glitch doesn’t happen again.

Does this user regularly copy n paste post content in? That’s a huge trigger for TL0 users.

Generally no. Most of his content is poorly written (he has a degenerative condition) and he tends to post links to various products, but none of the content is copied and pasted.

Generally it means this TL0 user is hitting spam or flag thresholds to get auto-silenced. You might have better luck forcing them to TL1, but if PMs are a problem, I’m not sure what to do.

1 Like

Ok, I appreciate the advice. We dropped him to 0 because his initiated private messages were… problematic. I’ll see if there are any other steps I can take to prevent that besides outright suspending or silencing him.

Not sure. Any thoughts here @jomaxro? Or anyone else on community team, really.

Not really, no. One guess (and it’s just that, a guess) is that one or more of the TL0 checks runs via a background job, so each time it runs and detects something “bad” the user is silenced. You don’t see a new notification as you already got one about the first silence, and the system is treating these all the same.