I chased the allspam methods for a while, didn't find anything wrong. Apart from the why data saying "Post manually reported..." which isn't the problem here.
> Bad keyword in body, bad keyword in title, bad keyword in username ---------- Title - Position 5-14: lumagenex Username - Position 5-14: lumagenex Body - Position 5-14: lumagenex
Hm... I don't like it that some of the websites only get detected as "bad keywords", but OTOH the list of websites is already quite long, and repeating keywords there wouldn't help.
I'm thinking of concatenating the no-word-boundaries regex, adding the usual TLD bit \.(co|net|org|in\W at the end, and adding this to pattern-matching.
The idea is that seeing "pattern-matching website" in addition to "bad keywords" (i) tells me there's no need to look at the source to add the URL to patterns; (ii) adds confidence to hypothetical auto-flaggers.
Yeah, worth doing. There aren't that many nwb keywords, and the regex of them is pretty cheap, so RPi will likely not melt.
And if we wanted to save CPU on regexes, we would use a cheaper method than findall...
[ SmokeDetector ] Bad keyword in body, bad keyword in title, link at end of body, pattern-matching website in body: What is Weight Loss? by JoshuaRudd on security.stackexchange.com
[ SmokeDetector ] URL in title, bad keyword in body, bad keyword in username, pattern-matching website in body, pattern-matching website in title: http://provasilfacts.com/ by Provasil on gaming.stackexchange.com
Yesterday, a whole slew of new questions began to appear from users with a reputation of 1 and as @w3dk pointed out, were registered 4 days ago (now 5). @dan made the first attempt to clean-up the mess by removing all the questions that were submitted by a single IP address and using any criteria...
Imagine getting a templated mod message ending with "Sincerely, Move more comments link to top, Motor Vehicle Maintenance & Repair Stack Exchange moderator"