@Makyen Would it be a terrible idea for the bulk feedback feature to just ignore any non-report message when counting? That is, if one of the messages within the range is not a report, it would simply apply the feedback to the next message, and continue from there. It seems like that's far easier for a human to understand than the current system, which IIRC also involves knowing which instance sent certain messages.
(and even if it doesn't, it's still tripped up many experienced users)
@RyanM That's been brought up more than once. It's been declined each time. Currently, it's a simple count of every SD message starting from the current one and going back in time. If we have SD not count some messages, then the human giving feedback has to keep track of which messages are not going to be counted, which is harder than just counting every message. Overall, it gets more confusing to not count every message.
Over the while, I've noticed that the username of naïve spam-posters matches the company they link to:
A short bit of code to identify name being congruent with a part of the link-string could be used to identify possible candidates for the spam filter. Perhaps this has already been done, we may...
@SmokeDetector I'm not sure this is spam per se, because it's not, well, promoting anything. However, it's complete nonsense in relation to both the question and site, so TP and red-flagging seems appropriate.
@Makyen That pattern looks like it's already caught by Potentially bad keyword in body and Potentially bad keyword in answer Append -force to the command if you really want to add the pattern you entered.
@cigien With this done, the watch of koyowater(?!\.com) needs to be adjusted such that it doesn't double-detect the username. When adjusting it, it would be reasonable to move the watch to the keyword blacklist.
@Makyen That pattern looks like it's already caught by Blacklisted username Append -force to the command if you really want to add the pattern you entered.
@Makyen Hypothetically, is there a reason in this case not to simply combine both blacklists into koyowater(?!\.com)? Since the keyword blacklist is checked against usernames.
(the weights probably differ slightly, but that's probably negligible)
we are not currently doing any higher-level analysis of the blacklisted domains but if we wanted to, it would be nice if they were all consistently listed in one place
@RyanM Basically, the primary issue is we shouldn't have two detections detecting exactly the same thing, unless we are specifically choosing to do so (which should be quite rare).
[ SmokeDetector | MS ] Bad keyword with email in body, potentially bad ip for hostname in body, potentially bad ns for domain in body, potentially bad keyword in body, blacklisted user (129): Masonic Supplies and about them by London Regalia on genealogy.SE
@tripleee That pattern looks like it's already caught by Blacklisted website in body and Blacklisted website in answer Append -force to the command if you really want to add the pattern you entered.
(Rather incompetently) plagiarized from superuser.com/questions/27446/… with known spam phone number added to the title. Probably the stupidest spammer of the month (so far ...) — tripleee8 secs ago
I was hesitant too, but the misleading link formatting about a frequent spam topic from a 1-rep user was enough to persuade me ... but not entirely sure, this is all circumstantial evidence
Bad keyword in link text in body - Bad keyword *in Delhi* in link text Potentially bad keyword in body - Positions 836-855, 2070-2089: ameuromigration.com, Position 2118-2122: Best
I wonder if there's a race condition there, actually...SD restarted reaaaaally close to when it posted that report in the room - to the point that the restart message was posted first. Perhaps it didn't let the report finish posting to MS?
Python concurrency is not my area of expertise, though.
@SmokeDetector k Likely spam: this has nothing to do with the problem in the question, which is about controlling the torch programmatically (custom flagged)
@tripleee That pattern looks like it's already caught by Potentially bad keyword in body and Potentially bad keyword in answer Append -force to the command if you really want to add the pattern you entered.
[ SmokeDetector | MS ] Bad keyword in answer, blacklisted website in answer, potentially bad ip for hostname in answer, potentially bad ns for domain in answer, potentially bad keyword in answer (184): Which one is correct is phrase? by Sheira Smith on english.SE