3
$\begingroup$

This is in regards to https://blender.stackexchange.com/questions/292724 , which is already closed, so maybe some folks can't see it.

This question quickly gathered a lot of downvotes. Community bot closed, giving a reason of "abusive or spam." However, this question is neither. It's not a good question, but the reasons it's not a good question have nothing to do with either of the reasons offered by Community Bot.

The user is probably being told their question was closed for being abusive or spam, and scratching their head-- or, perhaps, just writing off Blender SE as full of it. They may not realize that it was a bot that closed the question. Apparently, there were no comments from any users offering suggestions to the questioner as to how to find out what they wanted to know, just downvotes. No comments can be left on this question currently (I don't think there are any privileges left for me to earn related to this....) I suspect that if this post were closed for other reasons, I could comment to offer advice to the questioner-- IIRC, I've done so in the past.

Is it possible to get community bot to be a little less certain about the reasons that a particular post garnered so many downvotes in such a short period of time? It is not necessarily because a post is abusive or spam.

Note that there is an additional weirdness here-- atm, the post is marked as viewed 4 times, with a score of -6, implying that some downvotes were cast by users who never viewed the question. I'm not sure exactly where that comes from, maybe the view count resets when the question is closed, but it's definitely weird.

$\endgroup$

2 Answers 2

3
$\begingroup$

Thanks for bringing this to our attention. It all happened while I was absent, I'm only aware of the events because of this post.

As far as I know community bot doesn't erase posts of its own accord, it generally acts on flagged post by other users only.

Looking at the question flag history it was first flagged by Smoke Detector bot as possible spam, then put up for review and confirmed as spam by four other users doing reviews. This is what prompted community bot to erase it.

The fault here (if any) would be from those users who confirmed the post as spam despite not looking like it at first sight. The post however originated from a new account which matched previously known spammer names and also matched some of the same keywords keywords and patterns of previously confirmed unwanted posts.

As for the six negative votes, when a post is flagged as spam the community bot user downvotes the question on the user's behalf (to spare the user the associated reputation penalty for downvoting answers), this is in addition to any optional user cast downvotes, hence why the post appears to have more downvotes than views.

$\endgroup$
4
  • $\begingroup$ Are users actually capable of confirming SD's suspicions? It sounds like this is on the basis (primarily) of IP address; are users offered the similar posts to judge and say, "This is the same user"? Maybe in some cases, we could say, "Yeah, this is just guerilla marketing," but that doesn't seem to be what's happening. I think that we should expect that when a user has a question deleted, that they try to improve and repost the question, and that such shouldn't be considered as spam, but as how things should be, even though some tries won't reach the target. Spam is 2+ open questions. $\endgroup$
    – Nathan
    Commented May 12, 2023 at 23:13
  • 1
    $\begingroup$ For most spam posts the offense is quite obvious, users can easily identify by simple visual inspection, regular users obviously don't have access to advanced info about the account like IPs or nothing of the sort. The users in question are all low or no rep, "trusted in other communities", with little involvement in this community. Not sure why they confirmed the flags, either they are benign bot accounts themselves, they are aware of smoke detector insight, or they just arbitrarily confirmed spam flags by "me too" blind flock behavior. $\endgroup$ Commented May 12, 2023 at 23:39
  • $\begingroup$ "the community bot user downvotes the question on the user's behalf (to spare the user the associated reputation penalty)" - only downvoting answers cost rep, as far as I know. Will downvoting a question cost reputation?, Why does downvoting an answer cost reputation while questions not? $\endgroup$ Commented May 13, 2023 at 21:17
  • $\begingroup$ @MarkusvonBroady True, updated for clarity $\endgroup$ Commented May 13, 2023 at 21:27
2
$\begingroup$

The question itself may seem legitimate here (even if its content is questionable).

The reason why the bot deleted the question is because the user behind it had created some accounts with obviously dishonest intentions (some of them have been deleted already) and some (or all) of his posts were actually spam.

So the bot worked splendidly and showed in a very precise way that there has been abuse here before, and in this case that there will be abuse here with the highest possible probability.


To describe the situation a little more precisely:

There are some mechanisms running in the background here, which check the processes according to certain patterns for every (!) action on SE page, in order to be able to detect any misuse at an early stage.

The automated processes remain hidden from the user for the most part, but the information is basically publicly available.

One of these mechanisms is metasmoke.

"metasmoke" is a SmokeDetector, which is run by Charcoal, a group of wonderful and hardworking people whose effort is to keep spam away from SE-networks.

This mechanism operates on SO/SE with two accounts:

These two bots converse on Charcoal HQ. SmokeDetector detect and verify posts there and rate them.

In the specific case you mentioned, here's what happened:

  • Due to the fact that it is currently (still) was possible to post questions/answers here on BSE without having to provide a valid email address and register an account, this site once again fell victim to a certain spam wave. So it was a certain user who continuously created unregistered accounts and thus created a lot of unpleasant content on several SE-networks.
  • This "user" was recognized by metasmoke due to a certain string in the username, which was meanwhile considered so critical that it was even a blocklisted username. In addition, this user used certain keywords in several posts previously recognized as spam, which also appeared in the title of this post you mentioned.
  • Based on this precise (and reliable) assessment, the post was further processed as spam by metasmoke.
  • Since Charcoal was authorized by the CMs to remove posts in case of spam detection, this bot then acts on its own and removed the post. However, since non-moderators do not have the rights to directly delete a post on SE platforms, the bot acted on behalf of several Charcoal users by assigning a total of six downvotes, which caused this post to be automatically deleted by the BSE community bot.
  • However, since there are no real page views, the post on BSE did not receive any further views, because the processing (and also a possible manual review) was done fully automatically on a copy of the post in metasmoke's network.
  • The detection and processing subsequently deleted the post, but not the user. This had to be processed manually by the moderators.

So it may not always seem very clear why a post was deleted, but in almost all cases, metasmoke's assessment is very reliable and helps us keep the site spam-free.

BTW: In the last three days, due to this spam wave, more than 50 posts and users were reliably deleted, which would have otherwise quite harassed this wonderful platform. I therefore stand by the conclusion: The mechanism works very well!

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .