AI Stack Exchange is definitely letting more through in the last week or so. I'm flagging a half dozen per day at the moment, whilst before about mid-July this was very rare.
I don't know if we've hit some spammer groups' radar and are under attack (against the general trend), or if something has happened to spam approaches and/or auto-detection. It makes the site less pleasant to use, so I hope someone with access to the anti-spam toolkit is working on this?
I'm not quite sure what that SEDE query is showing (not familiar with the data model), but the numbers seem off? I've flagged over 10 posts in August already, and seen that they have been removed - just on AI Stack Exchange - but the graph only shows 5 "spam" posts across all SE sites. That implies the query is making a strict definition of spam that doesn't include all the ones I have seen personally
There have been several posts complaining about increased visibility of spam posts.1 At least part of the reason was the fact that the Charcoal team participated in the strike and SmokeDetector wasn't running during the strike.
Now the strike has led to some results. (Thanks once again to the peo...
On closer inspection, I think the query is just for AI Stack Exchange (at least there's some descriptive text that implies so), but the numbers still seem off. I have seen more spam personally on the site in last few days than the query is showing
I had no idea there was strike action, and that we relied on a volunteer service for spam removal . . .
I would have thought spam management would be a core SE feature provided by the company, as it heavily affects the quality of the site in a way that has nothing to do with the expert userbase.
Enough spam flags raised on a post will automatically delete it, and there are some measures to prevent spam from being posted, but SmokeDetector is much more effective.
Well I hope it can start running again. I'm happy to help out with flags for a few days, but I don't want my volunteer time here to be used up being a human spam filter.
Ah, so most of the posts I am complaining about have been since then, presumably overlapping when Smoke Detector went down and we started seeing what is otherwise the normal level of spam attempts.
The fact the SD and many active flaggers were on strike is certainly one of the contributing factors why spam was more visible on several sites, recently. AFAICT Server Fault and Super User were getting quite a lot of spam.
This probably means that this wasn't dealt with by the Community Bot. (I.e., removed revision history and all that stuff.)
To quote from Laurel's answer: "It looks like these are just Community's votes to delete posts as spam or abusive, which happens after 1 moderator red flag or N red flags from other users (currently 4 but usually 6)."
"You say that you had a day with twelve." I think 6 was highest count this week - a half dozen. Although I have not neen actively counting. I may be able to spot this in my history - will check
Sorry, I read dozen (and missed that it was half dozen).
What Mithical writes is in agreement with Laurel's answer. (But I wasn't sure about the correct terminology.). If it was locked as spam, the database should contain VoteTypeId=12 too (from the Community User).
In any case, this is probably not that important - even if not every spam post is caught, the queries at least give some reasonable estimate whether the site is getting more spam now than in that past.
I was playing with the queries mainly because I was curious whether the stats will show some significant difference comparing the times "without SD" and "with SD".