[ SmokeDetector | MS ] Blacklisted website in body, body starts with title and ends in url, link at end of body, pattern-matching website in body, potentially bad ns for domain in body, +3 more: Tonaki Tinnitus Protocol by zuswpnd on superuser.com
[ SmokeDetector | MS ] Bad keyword in body, blacklisted website in body, link at end of body, pattern-matching website in body, potentially bad ns for domain in body: natural process that happen to by Ingre1970 on drupal.SE
@EriktheOutgolfer Yeah, I just saw that. I suspect that there was more we didn't find (e.g. perhaps plagiarized content).
@EriktheOutgolfer It usually gets the gist across.
@EriktheOutgolfer Duplicate answers by the same user? If you're clear about that, a flag shouldn't be declined. In fact, actual duplicate answers by the same user generate an autoflag.
@Makyen not particularly from the same user, but I have flagged answers which are dupes posted after the original ones and they have a subset of the original one's contents, and my flags have been declined
@EriktheOutgolfer That's a more nebulous situation, which may, or may not, result in the flag being declined. It would be upheld for plagiarism, which is often not clear, but for just similar/duplicate answers, that's going to depend on multiple factors. But, yes, it's often just something you could down-vote.
for example, if one answer says "do this" and the other says "you should perform these actions", and both essentially describe the same thing, while the latter doesn't attempt to clarify the former, then I think the latter can be considered a dupe
It's a lot easier to just write a good/decent comment once (or get it from a repository/someone else), rather than having to type something out each time. You can easily add your own comments.
@EriktheOutgolfer I didn't like it as an extension, preferring to have it as a userscript. IIRC, the impetus at the time I switched was that it lost all of my comments, but the real reason is that I prefer to be able to modify it. I particularly didn't like the way that it downloads the script from off-site for every single page.
It uses a poor method of checking that the script is the most recent, which involves inserting the script and running it from the repository into every page, in addition to the one that's inserted from the userscript. I suspect it's the same for the extension, but I never double-checked.
It was relatively easy to modify it such that there was no need to actually run the script from the repository to check the version.
@EriktheOutgolfer Well, there are a variety of alternatives. Which browser to use is a personal choice. Try out various ones and see what works for you.
if I were a mod myself, I'd at least discuss it with the rest of the team to get a strong vote (like, +1 is strongly more apparent than -1, not just one more +1)
On the other hand, I still don't like this per-site-autonomy thing around network-wide issues like that.
I generally think that in a production release, we'd make the decision at a network level and let moderators convince their community that 100.00% isn't good enough.
Decisions about the United States Budget are made on models that are maaaaybe 60% accurate, historically... and people get nervous about spam deletions at 100.00%
At some level, it's some kind of human resistance to handing over the reigns to a machine.
Someone should run numbers on whether the conditions proposed in that big meta post would have hit an FP in the months since
on a serious note, the reason 2 spam reviews is kinda shocking to me is because if one suffers from human error, there will be only 1 review left, so there won't be any vote
also, metasmoke is a way safer environment to look at stuff. Only shown to people who want to be exposed to that kind of crap, and doesn't render images.
It's N=1 evidence, but I remember a SO post (borderline R/A) having erroneously receiving six autoflags and two members immediately taking action on it.
thorough | ˈTHərō |
adjective
complete with regard to every detail; not superficial or partial: planners need a thorough understanding of the subject.
• performed or written with great care and completeness: officers have made a thorough examination of the wreckage.
• taking pains to do something carefully and completely: the Canadian authorities are very thorough.
• [attributive] British absolute (used to emphasize the degree of something, typically something unwelcome or unpleasant): the child is being a thorough nuisance.
I agree that something similar is likely to be caught. Under other circumstances, it might take a bit longer to be recognized, but it probably would be (we generally have several pairs of eyeballs on each report). OTOH, without specifically knowing that changes had been made to that part of the code, I'd have been more tempted to wait for a 2nd occurrence prior to actually shutting off autoflagging (as long as it was flagging actual spam).
Even if something that is FP does get 6-flagged nuked, that isn't the end of the world. There have been a few posts that have come through here where I've disagreed with it being (manually) nuked. It's entirely possible (I've done so) to raise a custom mod-flag to explain the issue and have it corrected.
The possible, but unlikely, case is where a problem wouldn't be detected is if MS does autoflag, but keeps no record of doing so. In that situation, we would not be able to detect the issue, other than by the fact that posts were mysteriously getting nuked faster and people were seeing that they cast flags on things where the flag was not reported. This would eventually be detected because our tools would report to users that they have already flagged, when they were not aware of it.
@EriktheOutgolfer The quota starts whenever that IP begins making requests after >24 hours of not making requests. There is no set time, until a request is made.
Yeah, I'd consider it highly unlikely, but even then, it's something we'd eventually detect, just by people failing to be able to re-flag things they expected they had not already flagged.
I guess the pathological case is if MS started to flag posts which were not reported by SD (e.g. used the post ID form one report on a different site). If it was also 6-flag nuking things, and wasn't logging the issue, then we wouldn't have anything in place to see that it was happening, but that's some pretty far-fetched contrivance.
I believe that it's the decision to do autoflagging that's made prior to any feedback. I'm unsure if it's impossible for a feedback to be made prior to all of the autoflaggs being placed.
Flags also don't look at feedback, since it's assumed to not be present. Might be interesting to play with a few seconds' delay while waiting for FP feedback
scrutinizing it further i guess it barely counts as an answer, but the "here's a teaser of a barely qualifying summary, now follow this link for more" is an alarm bell for me
Does the first WebSocket notice of the go out prior to the autoflags being complete? At least IME, the notice of subsequent autoflags can be delayed by a human noticeable time.
That might be it, but I wouldn't think so. From what I've observed it's that the notification only happens if the first response is FP/NAA. I've seen a couple of times where no notification is made when the first response was tpu-. I don't recall seeing the current code giving a notification when a subsequent response is FP/NAA after a tpu-. Although, I thought I'd seen it happen in the past.
Update: Kevin Montrose makes it happen
The hell-ban is no more! Long live the verbose, obnoxiously evident ban!
Kevin has implemented (more or less) the system described below. Flaggers with a recent (past 7 days) flagging history consisting of at least 10 handled flags where >= 10% of flags we...
@Makyen That pattern looks like it's already caught by Potentially bad keyword in body and Potentially bad keyword in answer; append -force if you really want to do that.
