@Mast ah, minimal change from yours is Sonic\Wthe\W\b(\w*([A-z\-])*\w)\b\W([A-z\-]{3,7})[Hh]og - case insensitive H for WizHog, and "Monica-" is 7 characters, not 5
@JeffSchaller 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.
Question: I brought up the proposal on Teams for the 1 fp for old posts idea and it seems to have a few upvotes. I've seen people use open issues for this type of proposal as well. Is that something I should consider doing here at some point? Not trying to rush things but I'm just completely unfamiliar with this kind of workflow
@JeffSchaller That pattern looks like it's already caught by Potentially bad keyword in answer and Potentially bad keyword in body; append -force if you really want to do that.
@thesecretmaster I think a combination of both would be most effective. My main thrust after having reviewed around 8,000 of these is that old low weight posts are overwhelmingly, obviously fp. There is little value in requiring a second fp at that point (IMO) for tp and NAA obviously 2 would still apply. Even pushing them to the back of the queue just means that these posts will wait longer and have even less value when (if) they are eventually handled.
Note that with the rate of review before the Barbequeue it would mean that those posts never get reviewed and thus never add any value to the system.
@Mast Yes, but I specifically chose a class of question that is almost guaranteed to be fp even before one trusted fp vote. And failing to an fp state is probably the least problematic failure mode if anything actually is missed (by my understanding do correct me if I'm wrong!). But after 6 months or so, the vast vast majority of spam has been detected already.
Don't forget we are talking about only 1 weight reports here only.
I think 1-weight posts are around 15% tp. The last 18.6k are 14% tp with 15% unknown. Unreviewed posts are also less likely than average to be tp, since they're easier to identify.
@RyanM also note: that I'd expect the combination of age, 1 fp, and low weight to mean that the tp rate would be significantly below 15% though admittedly I haven't thought of a way to test that.
Maybe we could look for 1 weight posts where a tp comes after a fp for historical data? No idea if it is even possible and not sure about including age as well.
@Mast Sure we can. The problem I'm trying to solve is people spending so many man hours on things with so little value in return. Wouldn't we rather people spend time on posts with higher impact?
@thesecretmaster I think a combination of both would be most effective. My main thrust after having reviewed around 8,000 of these is that old low weight posts are overwhelmingly, obviously fp. There is little value in requiring a second fp at that point (IMO) for tp and NAA obviously 2 would still apply. Even pushing them to the back of the queue just means that these posts will wait longer and have even less value when (if) they are eventually handled.
@Rubiksmoose I kinda think Undo have a point here: "It's a good thought, and would be perfect in a lot of applications, but we value correct data a lot (helps with community trust, etc.) It's better for us to have a big queue and good data than a small queue and questionable data (even if it's right, it's still questionable from the outside)."
I think the risk of things being falsely reviewed this way is miniscule. But yes, if we are myopically focusing on accuracy here, the best thing to do is carry on I guess. Say if the risk of fp from this was 0.01% would it be worth spending 70 man hours to gather that 0.01%?
> Probably. 5% of 20,000 is still 1000 spam posts that you are probably marking as not spam. I don't like probably. Probably means guessing and guessing means inaccuracies.
as a very sporadic reviewer, I don't mind seeing/reviewing fp posts -- it's a nice mental balance for all the true spam we see. Is there a handy-dandy list or queue of low-score 1-fp posts that I can tackle as time permits?
@JeffSchaller This query only shows old posts with 1 fp and <=5 weight. Though currently does not filter out ones you already reviewed (though uncommenting one line will do that)
@SmokeDetector k Translation is: "When we go on trips, one of the things we often think about is accommodation. Do we stay in a hotel, campsite or will we have our holiday apartment? The real estate developer Kahloun Immobilière has an article that will answer this question."