Banning spam accounts on kbin.social is a cumbersome affair.
E.g., today @bayaz tried to ban several spam accounts. But that just did not quite work:
Instead of straight forward banning the accounts responsible for spam, those accs got unbanned.
How come?
If magazine owners ban a spam acc which prior went unreported, the ban button triggers an unban command.
To effectively ban accounts, they must be reported first. Approving the report will trigger a ban. I.e. magazine owners must report the account identified as spam to themselves to enforce a ban.
Therefore, pre-emptive banning of spam accounts does not work on kbin.social.
This is a serious problem which needs to be addressed asap.
I did. I would still be commenting about it as I don’t think extra stuff should be necessary to fix a problem like this. Filters should exist especially for new accounts (even the most cautious implementation could make a big difference), comparing names to banned accounts before account creation too (or shadowbanning so they don’t just choose different names).
Oh yeah, funnily enough the one thread in my image that isn’t spam was from a community I blocked. (at least I think it was, hard to tell with different instances)
Also to add to my list above, I just noticed a lot of spam posted in the food community. Also checking from the top of the magazine list with default sorting: tech, TodayILearned, space, showerthoughts, programming (though some of the spam is related SEO-type garbage). Books has 1 piece of spam and 1 user (probably bot given the post with 503 - Service unavailable in a title) who just aggregates Amazon links+descriptions.
Of course, I’m not trying to suggest that a third-party prophylactic tool is a definitive solution to what is ultimately a separate problem, just trying to be pragmatic here and restore basic readability for end-users, whether the filtering is done at the source or after the fact.
Let’s be real here, we are talking about unmoderated magazines on an instance where the developer is AWOL and using a framework that is lacking many basic features. Even with moderators, manual moderation can be a big ask and is time-consuming for free volunteers, depending on the volume of posts or how rudimentary the moderation tools are.
I actually don’t read kbin magazines much, so I wasn’t aware of the extent of the problem until I started opening those magazines more closely, and felt that something is better than nothing.
On the magazines you mentioned, I do see a few anomalous patterns that I’ll start filtering. For the most part, with filtering enabled, they were almost entirely free of garbage, save for a few patterns I may have missed on the first few passes. /programming and /food I need to take a deeper look at. The /food thing is good intel, because the use of Amazon referral links in the threads is something that can be generalized to other situations beyond books. Posting referral links is definitively block-worthy.
I also noticed some stuff that by any other name would be considered a thinly-veiled ad, such as specific users only posting articles to web sites they own and operate. I’m not talking about bots as such, but actively promoting one’s own content–even when such content is on-topic for the magazine. I declined to filter this stuff yet, because it received a lot of upvotes and seemed to be received favorably, maybe because the readers felt it was at least germane to the topic at hand? I think this is probably true for /food as well, because the line between “content” and “promotion” is unclear here, since what is a food blog if not a product generating click revenue? It seems like the tolerance threshold for that sort of thing is higher in a magazine like /food versus some other magazine. Anyway, I digress. I’m not treating such stuff as in scope, just filtering what is blatantly noise.
On food I was actually talking about the non-related spam, such as the x8 newest. Mostly gummies and pills, though maybe you’re already blocking those like you said (and this might apply to the below text too).
Though looking at it there are also some semi-related local-specific self advertisements by likely bots (Best X in location Y / near me). So I wasn’t talking about food blogs, but I do see some of those that are downvoted (and they are clickbait-y). One of the blogspot ones (the one who marked it as 18+) has one thread about Quora SEO and another thread called “hot girl” (random woman outside)
Aside from blocking I could see soft hiding (especially with ratings and/or grouping), though I’m not so sure how well that would work with assessing a thread or a user. Though I notice a lot of accounts like this don’t seem to comment much if at all.
@shazbot @insomniac_lemon To be clear, you’re talking about /m/food on kbin.social? And particularly the threads side?
I’m the mod for that community, and I’m not seeing any amazon links, gummies, etc. I’m usually pretty good about deleting those within at most 24 hours of them being posted. But, if you’re still seeing them, either there’s a glitch or I’m doing something wrong.
Right now, I’m seeing 51 threads total, and the newest one is a month old (tagliatelle link). Are you seeing something different?
I agree with you about the questionable food blogs and probably-ai-generated content. I’ve been on the fence about whether to delete those, but I decided to let it slide and hope that upvotes/downvotes would take care of it. Also, I didn’t get any user reports about them, so that was another metric to consider. For now, I’m just doing the absolute minimum of deleting obvious drug spam and amazon links (or, at least, I thought I was). If you notice anything especially egregious (where on earth do you see this 18+ spam nonsense?!) and could take the time to report it, I would really appreciate it.
Amazon links no, that was Books and the other user misspoke in one of their comments.
The other stuff, you deleted some of it after I commented, some is still there. Screenshot attached, the last one is about food but is that user’s only post and looks awfully like a thinly-veiled advertisement.
EDIT: I should have started with “thanks”, by the way. I appreciate the response.
This is so weird. I don’t see any of those first four posts. I see the fifth (I’m actually the downvote), and I agree it’s sketchy, but I’m trying to get just the absolute worst out for now.
Also, I have two posts you don’t have. I’m viewing this directly in Firefox – are you in an app of some kind?
That is weird, and nope also Firefox. I see what you do if I log out though.
I checked, I don’t seem to have that user blocked, and the spam still there then gone on logout/private-window is something else entirely.EDIT: I don’t see those 2 posts because they are categorized as Macedonian. I have it set to English only. Though the posts look like they are in English.
Ouch, what a bug. I knew some of the moderation wasn’t being federated, but I can’t imagine how a kbin user isn’t seeing the latest version.
To be clear, you see the spam when logged in, then don’t see it logged out, then see it again when logged in again?
I don’t see it regardless of whether I’m logged in or not. Also, I don’t think I’ve ever been able to see it because I don’t see the posts in a quick look through the moderation log.
Would you mind posting an issue about this? https://codeberg.org/Kbin/kbin-core/issues Or, I can do it if you don’t feel like it and don’t mind me using your screenshots. If you do post it, just please emphasize that this makes it impossible for moderation to happen because the moderator literally cannot see the posts.
Thanks again for trying things out and sharing your info!
@bayaz
Correct.
I haven’t used codeberg yet, so I’ll just say sure you can use my screenshots/words. It also directly affects you as a mod more as you’ve said, anyway.
Using those usernames/profiles to look at the posts directly, I don’t suppose there is anything that might detail what is going on? There are ~500 open issues, maybe this is some existing database/caching issue possibly related to post/community IDs? Though I am still not sure why viewing would differ by user.