@HenryEcker Until something's been on the lists for a while, you never know what the actual TP/FP rate will be. The data on MS is heavily biased towards TP posts vs. everything that's actually on SE. About all you can really tell is that the real %TP rate will be lower than what you find when searching on MS for a new term.
Yeah, I think that rule is going to need some tightening, but it might be worth leaving as-is for a little bit to collect data on what we need to be ignoring.
@SmokeDetector Entire answer copy-pasted from linked website, answerer failed to disclose affiliation. And, doesn't answer the question further indicating spam. Flagged as spam.
@RyanM 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.
@SmokeDetector fp "Dumping an API token really isn't a great idea... @CodyGray can you redact that from the revision history? And there isn't a question there"
@cocomac Yes, I'm aware; I noted as such in a comment. I don't generally redact this kind of stupidity unless specifically requested by the user, much as I don't bother trying to put cats back in bags. I did, however, delete the question so it becomes much less visible.
@CodyGray Fair enough, what about on MS...? I highly doubt that this user knows about that, but it doesn't seem great to share the key there, and it probably isn't great to tell someone that is brand-new to SO about MS, so can it be redacted there, or is the policy that the people here are more trusted than 'random' SO/SE users so it's fine?
@CodyGray I'm fine with not doing it as you said, I just thought that maybe it was easy like on SE where there was built-in post redaction. Seeing as there isn't, ignoring the flag I raised on MS seems fine to me. But I'm content to do as you said and ignore it
@cocomac I would not exactly say it's "easy" on SE. There is, in fact, a built-in feature for post redaction, but it requires a moderator to submit a redaction request, which is fiddly enough, especially in the common case where you don't know exactly what information is sensitive and needs to be redacted, and then another moderator needs to approve that request.
It's a fair number of steps, since it's dangerous (permanent removal of information), so I would caution against developing the impression that it's simple or routine.
@Nick Why did you explicitly add bookending to this? That should be added implicitly.
!!/bisect situs judi PUSATMENANG dengan kemenangan sensasional
@Nick I suspect Makyen would request that this be changed to a non-capturing group. What I am unsure about is whether it is advantageous to include an optional trailing numeric match like that at all, since it will match just as effectively without it.
E.g., pusatmenang vs. pusatmenang(?:\d+)? or even pusatmenang\d*.
I suppose the difference is what is consumed, but judging which form is better is a bit above my head at this point.
@CodyGray This comes down to the same reason I added \b to the other regex, I wasn't sure if word boundaries were automatically added. In which case the \d part might be necessary to match. But \d* would have been better...
I am, actually, not sure. Entries on the watchlist have \b implicitly prepended and appended to them, as do entries on the keyword blacklist, but that is not true for entries on the website blacklist. Since website entries added to the watchlist may be promoted later to the website blacklist, I am not actually sure whether it is desirable to add the \b bookending as a matter of course when adding to the watchlist.
I do not typically see people doing this, but I am unsure whether that is out of principle, ignorance, or laziness. Thoughts, @RyanM, @Makyen? (In reference to this proposed watch of an IP address.)
@RyanM Which userscript requires disabling to perform a redaction? Flags ToC shouldn't require disabling. It won't let you approve/reject it from the flag ToC entry, but can still do so from the flag dashboard, which is generally how you come across something requiring redaction approval, anyway.
I don't think we're relying on it for anything that I've noticed, and if we are doing so unintentionally, then we'll fix it on a case-by-case basis, rather than accidentally applying a very heavy-weight rule to the wrong website.
In my opinion, the risk of a 90+ weight rule hitting the wrong site and triggering autoflagging is more concerning than missing a particular piece of spam until we tweak the rules shortly thereafter.
But the realness of the Scunthorpe problem is exactly why I prefer that it be explicit, then if you don't see it or don't want it, it's immediately obvious what is wrong and/or what to do. But, yeah, I get that it would be a bit annoying to have to do it explicitly when it's what you typically want.
Kind of like having to say "hello" and "goodbye" to people. :-)
@CodyGray 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.
I'm thinking this is a very specific type of spam that I've seen before, so a targeted watch should be sufficient, but I am checking into alternatives, too.
