@JohnMiliter In general, you're not supposed to to give feedback on posts which you !!/report, unless obvious spam.
@JohnMiliter @DavidPostill @Videonauth @JakeSymons IMO, this is not actually spam, but was an attempt to provide an example for or ask another question about the user's other answer (with code) just above this one. The destination site shows transparent video, which is what their answer was about. The video isn't in English, so I'm not sure if it's actually advertising anything. Did anyone listen to the video who understands what was being said?
Instead of the giant post-for-each-meta operation, he's offered to [feature] something on mSE - which gets us exactly what we want and more, in one place.
(2b) He also pointed out that we could get these ourselves in the future through SEDE. It's kinda painful due to site switching, but totally possible. I completely forgot about PostsWithDeleted, so that's helpful
Why can't we do a real statistics test of how autoflag counts affect post deletion timings? We could let metasmoke cast rand(1, max) where max changes to 4 or 5. Within a week or two we could easily have statistically significant results with confidence bounds and all kinds of fancy stuff.
My pull request was merged on Mar 27, 2015 and that was pushed, together with the recent changes needed for the Google Authentication, to production. Fixed and live in rev 2015.4.22.34
so I wasn't hallucinating a SEDE API, but it's been broken since 2015
Those two sites that do that are outliers; I haven't thought about them much. If we go to 4 on the 6-flag sites, probably leave it at one. If we go to 5, 2 would make sense
@thesecretmaster Yeah, but the three 15 rep socks are not that effective. They can only affect the posts which were autoflagged, which would very rarely match up with posts which the puppet-master had created the socks to handle (unless you're countering your competitor's spam). If you're running sock-puppets in order to delete posts by spam flagging, then the goal is to delete the questions you choose, not just those which the autoflagging gives you the opportunity to delete a bit easier.
Also, I dunno if anyone can answer this for me, but I've been curious: How do y'all get TTD data? Because AFAIK deletions don't show up in the realtime socket, and polling would be expensive.
The Stack Exchange WebSockets are handy for various things (like handling spam). For example, try running this JavaScript in your browser's console:
var w = new WebSocket("ws://qa.sockets.stackexchange.com/");
w.onmessage = function(e) {
data = JSON.parse(e.data)
data.data = JSON.parse...
metasmoke went down, then came back up and did that thing where it immediately fails over a random instance (happened to be Lunar), then someone put solar into standby and then metasmoke went down again and we were stuck with an instance that wasn't listening to commands because it had that ancient bug
luckily it came back up about 30 minutes later and since someone had tried a !!/pull it managed to run that before it died from the inevitable broken pipe
I think it was because I basically duct-taped the failover so that it just ran init again and ended up opening two sockets, but without actually joining the rooms
How did this show influenced Chemistry studies ?
Why is Meth considered to be the "highest" chemical drug ?
Does Meth cooking is done correctly in the show ?
@thesecretmaster You already can't create a flag condition that's under 99.5% accurate or hits fewer than 1000 posts - effectively, the worst you can do is 995/1000