So we’ve seen the complaints and the reports and boy oh boy are there complaints and reports.
I’ve discussed the account with the other mods and admins multiple times, and while we agree the volume is a lot, it doesn’t point to a botfarm or multiple people using the account.
Obsessive? Absolutely, but not technically rule breaking… Until today.
Today they indescriminately posted the same story three times from three different sources apparently solely to flood the channel showing a decided lack of judgement.
It’s a valid story from a valid source, the original has been kept here:
https://lemmy.world/post/21098916
The others have been removed as duplicates.
I’m also applying a 15 day temp ban on the account.
“15 days? That’s oddly specific! What’s in 15… OH!”
Talked about this with the admins and the suggestion is to use some kind of automod, but implementing something like that is a little beyond my ability.
Ah - I can see why they’d prefer an automod (they’d not have to worry about configuring or changing the server software in that case).
Unfortunately, as far I can tell, none of the existing automods out there support deleting posts by a rate limit. It’s not impossible to add this functionality, but it’d take a bit of work and time from a dev.
For example, I think for lemmymodbot one can modify the User Processor at https://github.com/noenfugler/LemmyModBot/blob/master/lemmymodbot/processors/user_processor.py to accomplish this.
Under line 8, add this line to create a user/seen hash
user_post_seen_hash = {}
and replace the entire execute function with something like this,
if content.actor_id in self.user_post_seen_hash:
if int(datetime.now().timestamp * 1000) - int(user_post_seen_hash[content.actor_id].timestamp * 1000) <= 300 * 1000:
handle.remove_thing(“Posting too frequently, take a break”)
self.user_post_seen_hash[content.actor_id] = datetime.now()
(Oh, and at the top of the file, also add above the first line,
from datetime import datetime, timedelta
)
But for lemmy.world (even if just looking at /m/politics) it would likely OOM from the in-memory hash due to the volume of users, so it’d need to be extended leverage the database for the lookup.
Something similar could be implemented on top of threativore, I think around this line might be the easiest place to implement the check of the username/timestamp, https://github.com/db0/threativore/blob/main/threativore/threativore.py#L285
Edit: Forgot to add, all the above code changes are completely untested by me, use at your own risk, etc.