[ SmokeDetector | MS ] Blacklisted website in answer, link at end of answer, potentially bad keyword in answer, potentially bad keyword in username, username similar to website in answer (221): Scan through wifi with mx870 printer by IJ Start canon on askubuntu.com (@Ollie)
@Zanna Thanks for the response. That's good to know. Given that the answer was posted at most a few hours after the blog entry/driver was posted on the linked site, I'd be quite surprised if the user wasn't affiliated. The vast majority of links to such sites from unaffiliated people are to posts that are substantially older (i.e. to posts that have had some time to be indexed by search engines and visited by many people).
@Zanna I don't know how it's today, but 20 years ago you could get drivers anywhere except from the manufacturer's website. Sites like driverguide.com were very useful.
@Zanna Yeah, that's sketchy, at best. Sure, the copy could be unmodified, but why would someone want to chance it when there's working drivers available from the original manufacturer?
yeah. a lot of the time it's subjective, usually they're low quality ("help i want to write code pls halp what should i use"), etc. also i agree; software recommendation isn't really something that fits SE's model of being part of a permanently repository of information; they get outdated really fast
also spammers will take it as a perfect opportunity - the question is literally asking for product promotions
@Mast you'd think a filter for a username of Mel and either of the word photographer or illustrator would remove them all with no human intervention...
@Mast well, I wasn't being entirely serious... but even on those sites you could add "I won't give you a prior notice again" and that would probably do the job.
@Nick I don't have my MS bookmark handy, but I'll check when I'm back home whether that's sufficient. I think they mix up their MO a bit every so often.
I noticed that the posts of the recent spam surge on Ask Ubuntu were all edited by community members into something like
EDITED - REMOVED SPAM ANSWER
or similar.
While I understand why the users have done this and I too don't want to leave the spam content visible for longer than absolutel...
Earlier today, I came across a question asking for recommendations for software for an automotive repair shop. This question is clearly off-topic, and I voted to close it as "Questions seeking product, service, or learning material recommendations are off-topic because they become outdated quickl...
> Url in title, pattern-matching website in body, pattern-matching website in title ---------- Pattern-matching website in body - Position 20-31: replica.com Pattern-matching website in title - Position 20-31: replica.com Url in title - Position 0-31: https://www.perfect-replica.com
@Ollie 'perfect-replica\\.com' is not caught by a blacklist or watchlist item.
@Ollie 'perfect-replica.com' is not caught by a blacklist or watchlist item.
@tripleee @Mast @Nick We already identify them with 100% accuracy, and have been for months. There's no need for machine learning. We could remove them completely automatically, but don't have permission to use more than 4 autoflags. To get that permission, we would need to either explicitly get it from the CMs, get it individually from a consensus of moderators on each particular site, or go through Meta Stack Exchange.
While we might choose to put more than 4 autoflags on a set of posts if SE was being hit by a concentrated wave of spam (e.g. at a rate reasonably measured in #/minute, maybe #/hour), we should not be exceeding what we have permission for just because we are annoyed by a spammer posting a few spam posts every few days.
@Xnero Because you complained, which made me check, which made me see that your flagging conditions were below the minimum required, which meant I needed to run the automatic check for all of your flagging conditions, as we require that they not be below the minimum.
@Xnero Then set your overall combination of your flagging conditions to somewhat above the absolute minimum, rather than below it. The accuracy for your previously enabled set of flagging conditions was 99.72%, which is below the 99.75% that is required. I also note that one of your flagging conditions is actually below 99.75% all by itself.
I'm aware that without the Core role you were unable to access the above database query, but that just means that you needed to adjust your flagging conditions to be more conservative, or risk having them disabled a lot. In general, that's something which several of us have gone through and just decided to not have our conditions be on the bleeding edge of the limit.
If this happens a lot, we may want to open a channel of sorts to get permission for 5-flag situations. But I imagine it doesn't happen enough to warrant this kind of force from MS, all the rewrites it would require and attention from the team.