« first day (1552 days earlier)      last day (2251 days later) » 

2:04 PM
[ SmokeDetector | MS ] Offensive answer detected: Node throwing Unexpected strict mode reserved word "yield" by lulliezy on stackoverflow.com
fp- by iBug
[ SmokeDetector | MS ] Username similar to website in answer, blacklisted user: How to buy using Kucoin? by TradingPeek on bitcoin.SE
tpu- by iBug
 
Anonymous
sd - k - f
 
mego k
 
[ SmokeDetector | MS ] Link at end of answer, username similar to website in answer: How to I enable or disable notifications for comments on Facebook ads? by Faizan Athar on webapps.SE
tpu- by iBug
 
2:28 PM
tpu- by iBug
Conflicting feedback across revisions: current, #1
fp by iBug
Conflicting feedback across revisions: current, #1
Conflicting feedback across revisions: current, #1
 
2:45 PM
I've gone through the discussion of the change we want to make and come up with a few questions we need to answer in our meta post. I'm sure there are more, but these seem to be the big themes.
Concerns to talk about in meta post:

- Why should this community trust your bot?
- What does this change accomplish?
- What are the goals of this change?
- It sounds like this removes community involvement in an aspect of the site. Is this good? Is this bad?
- Adding more automatic flags makes organic flagging harder. How does this community handle that?
- How does Charcoal mitigate the risks of reducing human involvement? Right now we require 3 additional people to confirm our flags to remove a post. If we drop that to 1 or 2, how do we ensure we aren't increasing the risk associated with
Remember, numbers rarely convince people that one way is better. There are concerns about automation here. It's stuff we've run into before. It's stuff I ran into with my comment bot. We need to address those concerns. Not just throw numbers out and mention our success rate
 
@Undo @ArtOfCode All righty we're in cash, we got about 7 skipped last night
 
@Andy Your comment bot?
 
There are follow ups to that, but that's the announcement post
 
@Undo latest one is 106826, so start there?
 
I have like literally nine minutes to look at this before I go skiing, but here goes
 
2:51 PM
That one was an FP but it should still have useful logs
 
[ SmokeDetector | MS ] Pattern-matching website in body: Woocommerce product description full width? by H Robbo on stackoverflow.com
fp- by iBug
 
skiing? season must last longer than I thought
cool stuff (no pun intended)
 
@quartata America has some pretty tall mountains and stuff
and we're having some weird weather. Season didn't start till like early Jan
 
@quartata It's also still winter :)
 
[ SmokeDetector | MS ] Bad keyword in body, blacklisted website in body, pattern-matching website in body, repeated URL at end of long post: Figure out how to the probability of lasers by user185074 on physics.SE
 
2:53 PM
Oh right, and winter has been weird this year.
 
tpu- by Tinkeringbell
 
@Andy groundhogs are awful classifiers
 
o.O
 
To be fair "6 more weeks until spring" puts it at the start of Spring. The ground hog isn't doing anything that standard orbital mechanics can't predict
 
W, [2018-02-16T13:17:25.257131 #29084]  WARN -- : [autoflagging] 106826: after_create begin
W, [2018-02-16T13:17:25.257490 #29084]  WARN -- : [autoflagging] 106826: thread begin
W, [2018-02-16T13:17:25.257540 #29084]  WARN -- : [autoflagging] 106826: Post#autoflag begin
D, [2018-02-16T13:17:25.259097 #29084] DEBUG -- :   ^[[1m^[[35m (0.9ms)^[[0m  ^[[1m^[[34mSELECT COUNT(*) FROM `posts` WHERE `posts`.`link` = '//stackoverflow.com/a/35324752'^[[0m
D, [2018-02-16T13:17:25.260235 #29084] DEBUG -- :   ^[[1m^[[35m (2.9ms)^[[0m  ^[[1m^[[35mCOMMIT^[[0m
and that's it
 
2:55 PM
although tbf it's not like I get much of a winter. shivering in my boots at the astoundingly cold 17C
@Undo so it croaks when checking for the duplicate. that's the first SQL query
and it dies without throwing an exception
 
Yep
Also interestingly, it looks like it's running in parallel with a web request
 
[ SmokeDetector | MS ] Few unique characters in answer, repeating characters in answer: Can I access a nested dict with a list of keys? by KenB on stackoverflow.com
 
wait wait hold up I made a mistake that one really is a duplicate
 
[ SmokeDetector | MS ] Bad keyword in username: Changing the Index of a StackLayout from another object by Bioflex on stackoverflow.com
 
@quartata whew, I was thinking that'd be hard to track down :P
 
2:58 PM
let me find an actual one....
 
@Undo scratch this, it was the post request. 201 created
brb, one minute then probably thirty seconds to look
 
@Undo if you still have time, 106801. TP, but below autoflagging
 
.... weird
@quartata ends at post#autoflag begin
D, [2018-02-16T10:53:50.613335 #29315] DEBUG -- : TopbarChannel transmitting {"review"=>1} (via streamed from topbar)
D, [2018-02-16T10:53:50.618689 #29315] DEBUG -- : TopbarChannel transmitting {"review"=>1} (via streamed from topbar)
W, [2018-02-16T10:53:50.619005 #18283]  WARN -- : [autoflagging] 106801: after_create begin
D, [2018-02-16T10:53:50.619492 #29315] DEBUG -- : TopbarChannel transmitting {"review"=>1} (via streamed from topbar)
W, [2018-02-16T10:53:50.619834 #18283]  WARN -- : [autoflagging] 106801: thread begin
Sure it's not a dupe?
doesn't look like it
If we knew when this started happening, it might be possible to correlate to a Ruby version or something.
aaaand out. I'll probably be back in a few hours (like four); should have WiFi there during lunch
 
3:08 PM
Have fun!
 
I didn't know that dogs could ski! I guess you've gotta get 4 mini skis, one for each paw.
 
[ SmokeDetector | MS ] Few unique characters in answer, no whitespace in answer, repeating characters in answer: Force of a Train by Sumner Menold on physics.SE
tpu- by ByteCommander
 
and the train goes EEEEEEEEE...
 
[ SmokeDetector | MS ] Bad NS for domain in answer, potentially bad keyword in answer: What mean ERR_SPDY_PROTOCOL_ERROR in nginx? by ayeshaji on stackoverflow.com
 
....ok
 
3:20 PM
[ SmokeDetector | MS ] Offensive answer detected: Admin Functionality visible to basic users by Robert Thiessen on security.SE
 
So two possible fixes: replace after_create with after_save, on: :create (the proper fix, if it's what I think) or change unless count == 1 to if count > 1 (probably should be done regardless). Also we need to do send_not_autoflagged when it's a dupe
I think the thread is being context switched in before PostsContoller#create has finished, so when you do the count it returns 0 and not 1
 
3:41 PM
[ SmokeDetector | MS ] Offensive body detected: Washing clothes after pre cum by user26385 on islam.SE
naa- by Makyen
[ SmokeDetector | MS ] Bad keyword in body, bad pattern in URL body, blacklisted website in body, link at end of body, repeated URL at end of long post: And now more than one phrases by ganna wllyz on drupal.SE
tpu- by Tinkeringbell
fp- by Tinkeringbell on Washing clothes after pre cum [MS]
[ SmokeDetector | MS ] URL in title, bad keyword in body, bad keyword in title, blacklisted website in body, blacklisted website in title, +2 more: www.supplementoffers.org/renuvaline/ by hereynaked on graphicdesign.SE
 
4:07 PM
Always these weird posts on islam.se...
 
tpu- by ByteCommander
 
Time to stop looking for ejaculate
Mostly FPs. The TPs are real spams instead of offensives.
 
4:29 PM
@Andy I'm not sure we should bring up the 'trust' aspect ourselves. There's some trust on the part of communities, but it's nothing compared to the trust of those few hundred people who gave us their flags (and gave me their tokens)
We should definitely have arguments (bad word, but best I can come up with) ready for the trust thing, but only if it's brought up.
Might be able to head that off with "Here's where you can get all the data we have"
(automated dumps would be nice, not required)
Yeah. Here's all the data we have, here's the code, you can audit us whenever... And please do, because more eyes is good
I'm pretty much the only one capable of hiding anything, and if I went evil... There are bigger problems to deal with.
 
@iBug maybe we should just whitelist it from the religion sites (Buddhism, Judaism, and Islam) those are the only ones where we really see it (fp)
 
@WELZ We haven't seen real TPs.
 
@SmokeDetector n
@iBug 22
 
@WELZ 22 "real"?
Those are really spams, not "offensive".
 
you're right that they are all caught by other reasons, though it helps to give it more weight
I wouldn't put it in offensive, just a bad/blacklisted keyword (on all sites besides for a very few, if it's there it's either spam or rude)
 
4:42 PM
\o/ SE stickers got here!
 
91087 and 44291 are the only two "real" TPs.
 
@iBug 2 offensive here and here
 
We can safely remove ejaculate.
 
that makes 4 then lol
 
@WELZ Both would be caught for other words like nigga or shit.
More words don't make a single reason used multiple times.
You won't get 2k+ weight for testro t3 tst 11 alpha boost pro muscle erx pro tryvexan blank atm cards essay writing service...
 
4:49 PM
@WELZ offensive
 
@WELZ k
 
tpu-
 
welz 2- k
(Maybe another userscript would do that: Easily create a fake report)
 
That would be awesome!
 
That would break more userscripts.
 
4:55 PM
 
@WELZ It's iBug's fault.
 
!!/blame᠎ ‌‌‌⁠⁠⁣
 
Have fun.
I'm going to sleep now.
 
This is interesting, stats on the SD shortlink clicks.
 
5:12 PM
[ SmokeDetector | MS ] Few unique characters in answer, no whitespace in answer, repeating characters in answer: Record video while playing music by Robin Buchanan on apple.SE
 
@Undo I wouldn't phrase it was "you should trust us because X, Y, Z, <data>, <pretty picture>". We can imply trust though by showing what we've already done for the community - Deleted X spam in an average of Y seconds/minutes over the past Z. Of these, there are approximately W users for $community that have already joined and are involved in autoflagging. Define W as users above 101 or even 200, so we aren't counting just users with association bonus.
Then when it's asked, we can reraise those points - we are already involved in your community - and provide the more detailed explanation
 
tpu- by Jake Symons
 
But, "trust" is one of the big things we need to communicate. Especially in combination with the "you're taking away an aspect of community involvement"
Basically, why should $community trust this algorithm over the community of users that seem to be doing fine now? Even if that community is already augmented by Smokey
Another thing we can mention as 'already done for the community' is some of the existing moderating tooling we've built plus anything we plan to build/upgrade prior to this meta post. "We've taken feed back from your moderator team to help improve our transparency and help your community by doing A, B, C." Extra points if we can show some results from those things too
 
5:44 PM
@Andy Maybe it would also be smart to explain why our system is more accurate without audits than the review queues with audits
 
I'm still doubtful how much community involvement is actually there... but I'm pretty sure there's not a way to attribute the non-autoflagged flags to actual "community" members... meaning people with rep higher than 101.
 
I agree. We should only be looking at what we've contributed to the community. We mention that there are so many users of the site over 101 that have contributed autoflags. Maybe mention how many flags too.
 
@Catija We could get basic data on that
Most Charcoal people use FIRE, and FIRE uses the metasmoke API to spam-flag, which creates database records in metasmoke. We can add those to the autoflag logs to find a subset of posts that were deleted entirely by Charcoal members.
 
I'll admit I'm confused about the desire for community involvement in spam hunting, but... looks around the room ...we are a community. I do understand how some of our actions can look like we're imposing though.
 
That might be useful. I would be interested to see how many of the three remaining flags are cast by active site users who see the posts and how many come from Charcoal... I know that those numbers aren't going to be mutually exclusive but it'd say something.
 
5:50 PM
Good point. We saw X spam posts. We cast Y out of Z flags needed to remove those.
 
[ SmokeDetector | MS ] Manually reported answer: Login to a vbulletin forum using Java by dishwasher reviews on stackoverflow.com
tpu- by teward
 
Whoever posts that is going to get a big boost of rep on meta :)
I think it will work, my auto flagging hasn't had a single fp yet. I wouldn't mind an increase in flags, get my total past 10k eventually
 
5:58 PM
FYI, I'm going on a skiing trip so I won't be around for a week or so.
 
@Glorfindel nice, going to compete in the Olympics?
 
I wish ...
 
@ArtOfCode for the posts that we wrongly flagged, maybe it would be smart to link those posts, even if they are deleted, just to show to other what kind of false positives they can expect
 
@Ferrybig But we don't want to add too many fps, at least less then the tps listed
Also @ArtOfCode you're probably the best person to ask, can I be added to the really cool people list please
 
@JakeSymons We could link to the metasmoke search page: metasmoke.erwaysoftware.com/…
 
6:04 PM
@ArtOfCode "That's Dangerous!" -> "But, That's Dangerous!" but it's so minor a change it's just a passing thought :P
 
@Ferrybig the tp/fp graph probably shows best how fps are expected, but we get lots more tps
 
@Ferrybig Expectations is good, I think I'd link to what they can actually expect, though - as in FPs for 250 or 280 weight, rather than FPs for autoflagging in general
because for 4-5 flags there will be a lot fewer FPs than 65
@JakeSymons bit young for an OAP ain't you? ;)
got a GH account?
 
@ArtOfCode what about Mith?
 
@ArtOfCode Yeah I got one here
 
@WELZ it's a terrible pun
@JakeSymons in your inbox
 
6:13 PM
[ charcoal-se.github.io ] continuous-integration/travis-ci/pr failure on 30f96ce: The Travis CI build failed
 
@ArtOfCode thanks :)
 
CI failing for some reason
that's one for @JF
 
I've added my comments to the gist
 
6:18 PM
[ charcoal-se.github.io ] continuous-integration/travis-ci/push success on f6a6a2c: The Travis CI build passed
 
6:33 PM
[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title, blacklisted website in body, link following arrow in body: Alpha Pro Testobuild Patients can remove by ashlucas on astronomy.SE
tpu- by WELZ
 
needs one more flag (just use fire)
 
6:56 PM
@ArtOfCode I haven't actually read it yet, but first impressions are "seems long".
 
@Undo sure is
 
Would be nice to fit it in a laptop screen. More likely that people would read it, then yell at us, instead of the other way 'round ;)
(but that might not be possible. I wouldn't know because I haven't read it)
On phone at ski place.
 
@Undo there's a lotta stuff we need to get across
 
Yeah, that's probably a root of our issues. This is such a context-heavy domain; it's understandable that some folks are nervous about us asking permission for something they don't understand fully
 
Aye. I think I'd rather err on too much information than not enough
 
7:07 PM
Works for me
If we're going to do this, we probably should go back and get permission to take those concerns out of TL... otherwise, this is going to be a ridiculously complex information silo thing
 
I'm happy just using the risk and community involvement concerns. They're generic enough that they might not have come from TL.
 
trying to get token
 
@YvetteColomb sounds like JS is blocked
 
oh god - thanks -
 
Anyway, here's my proposal for a timeline: (1a) start drafting meta post, (1b) fix deletion logs, (1c) pull together backdated deletion log info. (2) Run experiment, make fancy graphs. (3) Identify sites, users, etc
 
7:16 PM
AR SQL queries are gold
SELECT `sites`.`id` AS t0_r0, `sites`.`site_name` AS t0_r1, `sites`.`site_url` AS t0_r2, `sites`.`site_logo` AS t0_r3, `sites`.`site_domain` AS t0_r4, `sites`.`created_at` AS t0_r5, `sites`.`updated_at` AS t0_r6, `sites`.`flags_enabled` AS t0_r7, `sites`.`max_flags_per_post` AS t0_r8, `sites`.`is_child_meta` AS t0_r9, `sites`.`last_users_update` AS t0_r10, `posts`.`id` AS t1_r0, `posts`.`title` AS t1_r1, `posts`.`body` AS t1_r2, `posts`.`link` AS t1_r3, `posts`.`post_creation_date` AS t1_r4, `posts`.`created_at` AS t1_r5, `posts`.`updated_at` AS t1_r6, `posts`.`site_id` AS t1_r7, `posts`.`u
 
... @Art did you nuke that previous message?
 
@Undo Yours?
 
Yeah
Attributed to feeds. Never seen that.
 
Not me guv
talking about 42900521, right?
 
Just above Yvette's screenshot.
Don't think that crossed any lines.
 
7:19 PM
I... see no message there
 
521, yeah. Sorry
 
gotcha
yeah, not me
if someone wanted that redacted it didn't work
 
Well that's interesting. I cleared history
But it was a feeds deletion. I'm going to take that as an indication that we need permission to even talk about that.
 
No CM in the room
 
Restart: API quota is 13535.
 
7:23 PM
@ArtOfCode well... I don't have any explanation other than lurking staff. Which isn't very convincing even to me
 
[ SmokeDetector | MS ] Bad keyword in link text in answer: why my apple id isdisable.l can not install app by kolka on apple.SE
 
@Undo could always ask. Lurking staff seems kinda unlikely
 
fp- by Yvette
 
@ArtOfCode Do you think that should be deleted? If it's crossing the line, I'll just accept it as someone trying to help out.
... oh. Can you pull up that flag view? Betcha that's it. Would make sense.
 
@Undo I don't, no. It's a generic "might be a concern someone raises", no mention of TL content
@Undo yep, got it in one. One offensive flag.
 
7:26 PM
A unilateral mod flag is the only way I know of to get those results.
 
so a mod
I wouldn't put it past me hitting buttons accidentally on a touchscreen
 
I'm gonna drop it, then. Either it was you hitting buttons, or someone doesn't want it out there. Either way, not worth chasing
We do need clarity. A quick "hey, we can share this specific concern you raised anonymously and text-anonymously, right? Thanks :)" would make me more comfortable
Anyway, back to skiing
 
[ SmokeDetector | MS ] Few unique characters in body, repeating characters in body: Generating a page with php from a single header by POPCORNS on stackoverflow.com
[ charcoal-se.github.io ] continuous-integration/travis-ci/push success on 7a96cda: The Travis CI build passed
[ charcoal-se.github.io ] continuous-integration/travis-ci/push error on 5b3c482: The Travis CI build could not complete due to an error
 
7:54 PM
@WELZ You really shouldn't delete something just because it was status declined... it can be of value to people asking in the future... also, Shog may have been in the middle of writing an answer and I'd like to see it.
 
@Catija ok I've undeleted, you're right, but then after he posts his answer I won't be able to delete it and I'm just gonna keep on collecting downvotes -_-
 
You could delete it before he gets any upvotes. :P
 
finger on the delete button lmao
and too late 🤣
 
[ SmokeDetector | MS ] Bad keyword with email in body: How to extract gmail email addresses from a string by Boston Howie on stackoverflow.com
fp- by WELZ
 
@SmokeDetector fp-
at least for that case
 
8:05 PM
@Catija like I said, collecting downvotes 🙄
 
@WELZ You got an upvote, too.
 
@WELZ don't worry too much. I've asked some of the most unpopular questions on both MSE and SO meta. They've since been dissociated from my account - that's a long story
 
I don't really care about the rep or even the downvotes, just it feels that everything I suggest on meta gets crazy downvotes and then it looks bad on my profile as "Worst Newtork Posts" (and I see this with so many things on meta, not just my own)
 
@WELZ trust me, I've been there. lol
 
@YvetteColomb Which posts, just curious?
 
8:09 PM
it takes time to work out what meta responds to
I'll have to go dig them up
 
I'm almost gonna be on the Worst voted questions on meta
 
@WELZ It's important, when you make suggestions on MSE to think about everyone affected by the question. It's important to support your suggestion by explaining the benefits. What were you hoping to get out of making mods pingable in the main chat on their site, for example?
 
@Catija I wanted the mods to give an opinion on a certain (controversial) meta post, weather an answer, a comment, or even a chat message - just to see what they thought etc. but it doesn't really matter anymore.
 
OK, someone needs to explain to me why Smokey can't give me booze :P
 
8:23 PM
@Catija SE chatrooms can be populated by everyone older than 13. Underage drinking shouldn't be encouraged?
 
@Tinkeringbell Eh, it's digital and one of the rooms I use is a Tavern :P
 
Drinking age is 13 in some countries
 
@Catija Taverns can serve milk, right? :P
 
@Tinkeringbell yeh, a large portion of Charcoal Members is under 21
@ArtOfCode SE is a US company lol
 
Large portion of the Charcoal members are in countries where the drinking age isn't 21...
 
8:25 PM
@WELZ Yeah, I even saw a 5 year old in your statistics :)
 
@Tinkeringbell 1? 🤦
 
@WELZ I don't know, last time I looked there was 1 five year old, a six year old and 1 seven year old... do you still have the statistics? Where did the links go??
 
@Tinkeringbell They're there, I just haven't been able to filter out the crap.
 
@WELZ Ah... that works ;) I found the link again. There's still the 5, 6, and 7 year old :P
 
@Tinkeringbell and I deleted some
 
8:31 PM
API quota rolled over with 12842 requests remaining. Current quota: 19999.
math: 140
stackoverflow: 99
physics: 37
superuser: 30
codegolf: 28
askubuntu: 22
english: 19
apple: 17
serverfault: 14
electronics: 12
travel: 11
interpersonal: 11
mechanics: 9
islam: 8
wordpress: 8
unix: 8
mathoverflow.net: 7
es.stackoverflow: 7
worldbuilding: 7
judaism: 6
raspberrypi: 6
meta: 6
drupal: 6
gaming: 5
magento: 5
salesforce: 5
bitcoin: 5
rpg: 4
money: 4
sitecore: 4
android: 4
tex: 4
graphicdesign: 4
dba: 4
webapps: 4
arduino: 3
ell: 3
softwareengineering: 3
emacs: 3
writing: 3
puzzling: 3
 
so @Catija @Andy about that data on how many of us flagged a post... it's woefully inaccurate
 
@Catija well, here it's infinity
 
@M.A.R. No alcohol or no wine?
 
apparently there are only 140 posts that were completely nuked by Charcoal, ever
 
Alcohol is haram in Islam
Which is Arabic for bad
@ArtOfCode that is, underwhelming
 
8:34 PM
@M.A.R. mainly because it's wrong
there are far more than that, I just don't have the data
 
I was thinking 4000 or something.
At least 10 percent of the questions Smokey catches
 
@ArtOfCode you may end up stuck without my payment until the 20th - stupid paydays being stupid.
wrt the stickers
i'll add in a couple extra pounds as compensation
 
@ArtOfCode Really? How does it determine that?
 
Eh, no worries. As long as I get it at some point, it's all good. We've got enough to order stickers with
@Catija I ran a query against the MS database
only 140 posts have 6 flag logs on them
 
@ArtOfCode But if I flag something on the post, that wouldn't count... right?
So, when I can't get FIRE to respond, then I resort to going to the post directly...
 
8:38 PM
@Catija correct. MS only has data on what you flag through the MS API, i.e. FIRE
it'd be neat if we had a userscript that reported when you spam-flag a post that Smokey caught without going through FIRE
 
Yeah, that was my concern. And (even on the moderator side) who spam flagged something isn't really visible, so you can't do it from the site, I'd think.
 
still wouldn't get complete data, but it would be better
@Catija you can get it out of the flag history, I think, but it's not obvious
 
Yeah, that's why my idea was to go through the site rather than through MS... but I don't think that'd be very easy to do.
You'd also have to interpret the data somewhat, since some people are both... so if I mod spam flag something on A&C or IPS, how would it count, in that case...
 
Nope. It'd need mods on every site to run a userscript... or Shog to give us the data, but that'd be a breach of flag confidentiality
 
Ugh. Yeah, I figured it'd be difficult.
 
8:43 PM
@Catija single post timeline scraping as a moderator, but that's not trivial to obtain, and borders on the violation of the confidentiality world if we try and use any of that data
 
Ah, I see. Yeah, you can see the flaggers in the timeline.
 
@Undo you around?
@ArtOfCode oh right, deletion watcher
not quite ready for merge yet
 
@Catija FWIW, gist.github.com/BrockA/5761948/raw/… gives me fast access links to Timeline and Revisions button (but not necessarily on all sites I don't think), but it's still not trivial to get the data out in a usable format for Smokey's purposes (just in case someone wants to use that person's script)
 
because there's a bug
 
@ThomasWard I don't know if that's the one I have but Art did make one that does that, I think.
github.com/ArtOfCode-/Userscripts/blob/master/stackexchange/mod/… I think is the one I'm using... and it says not to use it with SEMTI but it actually seems to be perfectly compatible ... I think it got fixed
 
8:52 PM
remind me what SEMTI is?
 
Stack Exchange Moderator Tools Improved. animuson's script.
 
ah right
 
Have you tried Shog's improved Flag Filter, by the way?
 
@Catija that script specifically is compatible with SEMTI - if SEMTI is present, my script just doesn't run and lets SEMTI add the links
 
@ArtOfCode Right, but the note on the script says not to use it with SEMTI :P
 
8:54 PM
[ SmokeDetector | MS ] Bad keyword in answer: Possible cheating in Online Poker (by the dealer) by 2SloJo on poker.SE
 
@Catija no I haven't, but perhaps sharing that is better done in a different location?
i also like SEMTI now takes
 
@Catija that's saying the same thing I just said - it's compatible, you can use 'em together
 
@ArtOfCode Oh, I get it... it's saying what the fix did... I keep reading it as "don't use this script if you're also using SEMTI".
 
aye
 
[ SmokeDetector | MS ] Repeating characters in answer: How to secure client-side anti-cheat by asjwajpow on stackoverflow.com
 
J F
9:01 PM
?
 
tpu- by teward
 
9:23 PM
@quartata On phone, so kinda
@ArtOfCode It would be neat to have accurate numbers on that. We could throw a userscript in the sidebar for core people to run and xref their flag history against MS, but it'd be too much work. Especially right now
 
yup
 
@ArtOfCode We used to have that. Forgot what it was called, but it did that.
 
FDSC could do it, actually - it already picks up on Smokey-caught posts you spam flag to send feedback, so it can just create a flag log too
that might be the one you were thinking of
 
Might be. I think it did do that
 
might've done
 
9:33 PM
Reading Andy's messages earlier, agree with most everything there. I could be wrong, but I don't think we'll hit much pushback on the "community needs to be involved in spam" front. I kinda understand it, but it breaks down under scrutiny IMO... and is unlikely to come up organically if we don't specifically bring it up
Also, we're going to have to figure out how to not be a vote cartel on these meta posts
 
Tell members that if they don't have X rep on the site, stay away from the vote buttons...
 
J F
^ :+1:
 
I do fear that many of them won't get much attention on the smaller sites (if we choose to post there). We'll want to get a default stance, which might be "apparently our services are useful here"
@Catija that'd work. Including comments.
 
Well, I was under the impression that the goal was to pick a few sites initially? I'd imagine that sites with more spam would be earlier targets?
Smaller target sites you may be able to contact later in the process, and, superping a mod in the TL to see what they think.
 
@Catija It is. Depending on how those go determines whether we bother with smaller sites. I don't know the conditions when we would, but it'll probably be easy to tell if we need to
171 sites is... a lot, and I'd be fine with throwing something on mSE for 150 of those with opt-out criteria
 
9:40 PM
The problem with using MSE is that a lot of small sites and small site mods don't use it.
 
That's the other thing. After doing a bunch of thinking on ski lifts today, I tend towards not linking these individual posts in TL, and only as necessary here. If the spirit is to get community involvement, dragging network mods in could easily sway results unnaturally.
@Catija It's the best we've got, outside of 171 posts. I honestly don't like the "site agency" thing - the network has to mean something outside of sharing server space
 
Sure. Lots of mods don't use the TL either... and it's not as if you, Art and Andy (among others) aren't mods...
 
nvm I misread that
 
@Undo Sure... and that's sort of why I was thinking that it might be worth talking with the CMs to see what they think. They may be unwilling to "rule" on the situation but they may have some ideas for how to address it.
 
@Catija Yeah, which is why I'd lean towards having one or two folks answer questions per post, with the rest of us supporting and maybe heckling from here. We easily outsize active meta crowds on even larger sites
@Catija Probably would be.
 
9:47 PM
I'm pretty sure the active meta community on Seasoned Advice is about four.
I mean... yes, the mods have agency over their sites... but the staff absolutely do (particularly when it comes to things like spam)... so if the benefits to the network are of value to them, then maybe they will green light it...
 
"staff liked it" doesn't really resolve disputes as well as "your community liked it, or was at least indifferent"
 
Sure, and I admit that's the case, but having their endorsement and their thoughts on it might help.
 
But it's a good thing to pursue first and have in our pockets. Best case is "y'all are fine, just do what you think is best", worst case is probably "yeah, good idea to ask".
In between that is "post on MSE again and call it fine"
 
My main concern in focusing on the high-volume spam sites is that they're often the most populous sites, so the volume is dealt with quickly by there being more people able to flag... but sites like A&C where there's not a lot of native usage... or Ask Patents... or some of the other sites that are more likely to have spam sit around for long periods of time... those are the sites that really need it more than others.
 
A long tail problem
Maybe this is a network issue that should be dealt with at that level
 
10:01 PM
Dunno... I feel like there's got to be a middle ground between posting on MSE and posting 171 posts all over the network... maybe an analysis of the sites with the combination of highest spam activity and lowest userbase?
 
[ SmokeDetector | MS ] Few unique characters in answer, repeating words in answer: How to query vsts artifacts with tags by spam spam on stackoverflow.com
tpu- by Jake Symons
 
@SmokeDetector I think they were desperate for it to be flagged as spam. Calling themselves spam spam and spamming the word spam. Too much spam!
They've edited it to "not spam" now :)
 
[ SmokeDetector | MS ] Offensive answer detected, repeating characters in answer: How do I alt+f4 in OS X? I want to close focused window by user276666 on apple.SE
 
Anyway, I will be back for some more flagging bright and early tomorrow, when that meta post goes up please pin it.
 
tpu- by Jake Symons
 
10:15 PM
Now my comment is being spammed my comment
@SmokeDetector goodnight
 
10:44 PM
@ArtOfCode that's disappointing. Do you know if there are any less than 6 flag posts where a charcoal member/site mod jumped in and spam flagged it too? That would give you less than 6 flags but still be only CHQ members involved
That's probably not an easy query though
 
@Andy 498, supposedly.
Either the data or my query is wrong.
 
I'm guessing bad data. Still seems low. That said, I'd be really happy to be wrong about that and see there really is that much human involvement still
 

« first day (1552 days earlier)      last day (2251 days later) »