@SmokeDetector I agree with the flag being FP (and not NAA, nor TP) in both it's original and edited form. The Autoflag was correct both times (the assumption being that after editing the swear words would be removed - actually viewing the post is impossible (for me) since it has been deleted).
@SmokeDetector See prior response. I agree with the Autoflag.
@double-beep and @terdon I do not know if you need to reread what I wrote but the rules regarding feedback are clear and I have both followed them correctly and written unequivocally above. GTG back to work.
@Rob I've read it multiple times and you say "I agree with the flag being FP (and not NAA, nor TP)" and then "I agree with the Autoflag." So you both think that the flag was a false positive and you agree with the flag. That's very far from unequivocal :)
whoever has privileges in CH in GH, then this probably needs to be modified:
> In these cases, it’s recommended that you pick the feedback type that feels most appropriate (if you’re still unsure, err on the side of caution and choose f). Then, leave a comment on the post indicating what is wrong with it, and why you chose that feedback.
@terdon "I agree it was a false positive, but I also think it should have been flagged for human review anyway." which I kinda get because an algorithm can't really tell the difference between "needs an edit" and "needs to go"
but it still kinda rubs me the wrong way ¯\_(ツ)_/¯
@double-beep 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.
@Seth (@terdon) I agree, the post was an answer and should have just been edited to remove the first sentence. Unfortunately, the post included wording that triggered multiple detections. For this post, those detections were based on the first sentence, which is R/A, even though it was directed at the poster themselves. For the characteristics of that post: 161 weight, 3 detections, a user with 8 reputation, and on Ask Ubuntu, flagging is correct 99.95% of the time, based on historical data.
I wasn't around when that was detected, but from the transcript here, it appears it was recognized as salvageable by a user within 15 seconds of the report being posted in here. However, even by that time the post was already deleted. Given that there were 3 autoflags cast and no manual flags, it was not deleted solely based on our flags. From the image you've provided it appears it was deleted as the result of the question being deleted, most likely by the OP (I presume).
Unfortunately, the person who recognized that it was salvageable 15 seconds after it was reported here in Charcoal HQ did not give feedback on the post at that time. However, at least part of their choice not to do so was that it was already deleted ("salvageable but deleted now"). If the user had given FP feedback, which they were likely to do, based on their history, then the people who had flagged would have been notified at that time about the FP on a autoflagged post.
If the post was not deleted, that notification would have likely resulted in at least some of the flags being retracted. How many would have depended on the number of users who were active at the time. Unfortunately, the SE API doesn't have the capability to retract flags, so retractions can't be done automatically.
Had the user given FP feedback 15 seconds after reporting, the system is set up such that a custom mod-flag should be raised on the deleted post asking a moderator to review the post. This is done when there is FP feedback after a post is deleted when there are flags that are cast through metasmoke. The intent is to make sure that the site mods are able to review the situation.
HTTPSConnectionPool(host='metasmoke.erwaysoftware.com', port=443): Max retries exceeded with url: /status-update.json (Caused by SSLError(SSLError("bad handshake: SysCallError(-1, 'Unexpected EOF')",),))
Matched by the following regexes: amazonhealthstorediet\.over-blog\.com on line 6151 of watched_keywords.txt amazonhealthstorediet on line 6154 of watched_keywords.txt
@Seth Ah! Gotcha. Yes, that does indeed make sense. I thought that @Rob mean the post should be flagged and also _shouldn't_be flagged. Never mind Rob, I get it now.
@Makyen thanks for the detailed explanation. I was just curious because I had misunderstood Rob's message.
@double-beep Yes, that is exactly the crux of the matter. Regardless of whether the algorithm made the correct decision or the wrong one, I agree with the result - I'm not saying it should have flagged it TP and then we should have reversed it, I'm saying it flagged it FP (right or wrong) and that is what I agree with. --- @terdon OK. --- The guy is saying "I f-ed up" and so it's FP and an edit, not tp (with a blacklisting). The section of the 'How to give Feedback' offers that
Unless you are the one reviewing it, or are going to write a PR for that, then nothing to see here. --- We do appreciate people asking about these things, and thank you for your question.
@Makyen It is a continuation of the prior comment. There isn't anything rude. They asked about a reply that I made to the computer, a conversation that I was having with someone else (that they didn't understand), and I took the time to explain it. I added that we appreciate people asking about how such decisions are reached and encouraged them to ask such similar questions in the future.
Matched by the following regexes: keto\W?lean on line 1606 of bad_keywords.txt keto(?:nes?)? on line 4694 of watched_keywords.txt
Matched by the following regexes: purefit\W?keto on line 1276 of bad_keywords.txt keto(?:nes?)? on line 4694 of watched_keywords.txt shark\W*tank on line 7456 of watched_keywords.txt
@Machavity 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.
@Machavity I may be missing something, but I only see 2 TP posts on MS. That is not sufficient for blacklisting. For blacklisting, 5 TP and no FP are required under most conditions. Please see the Guidance for Blacklisting and Watching.
