« first day (1570 days earlier)      last day (2250 days later) » 

11:00 AM
@BhargavRao we've been telling angus for years now to get a diamond
 
New metasmoke user 'Francesco' created
 
@terdon Eh... I just don't see how it really affects anything given how fast it's nuked anyway. Normally it's already off the front page by the time anyone could downvote it.
 
@tripleee Yeah, piling on is the issue. Something which ALL chatrooms face to some extent. The SOCVR case where they piled on a meta post, the Lounge case where they piled on a mod. Heck, the TL faces it too, one link and you have 50 mods running on it.
Also you guys speak a lot, already tired of scrolling back and forth :p
 
@Mithrandir Sure, it's about as minor as point as it gets. Just, you know, every little helps and since there's already a fear of these strange charcoal people coming in to steal our flags, you may as well not add to it.
 
a VLQ post without downvotes is jarring, "why is nobody doing anything" so I tend to downvote just to signal "this is bad" regardless of which specific reason I have to do that
 
11:04 AM
@Glorfindel True. I remember declining David Postill's 3 wrong flags in a row and getting them flag banned. But after that I've never seen even one instance where they've flagged wrongly.
 
conversely, a post with a lot of downvotes and no close votes is jarring, too
 
[ metasmoke ] ci/circleci success on 9e8a567: Your tests passed on CircleCI!
 
@Magisch yeah, this is what I want to prevent. We shouldn't be getting till the site moderator message level. Users must be taught beforehand.
 
I'm not about to propose we stop logging for autoflagged posts entirely but perhaps we could at the very least tweak them so they are harder to manually flag (use a format which FIRE won't pick up, for example?)
maybe just "autoflagged [link to metasmoke]"
 
Perhaps take every mistake a user does very seriously, and teach them what's the correct way.
 
11:06 AM
@tripleee bleh
 
before you can say "problem solved" there's a userscript which works around that
 
@tripleee VLQ isn't spam, that's a different issue.
 
@tripleee I don't think changing our format is the way to go here, FIRE already knows that a post is autoflagged, what if it only adds feedback, and doesn't flag the post if it sees the post has autoflags, it should just popup a notification that you should manually flag, works great for the roboreviewers who don't pay attention
 
[ SmokeDetector | MS ] URL-only title, bad NS for domain in body, bad NS for domain in title, bad keyword in body, bad keyword in title, +6 more: top10sideeffects.com/total-radiance-teeth/ by Explesse89 on askubuntu.com
 
but anyway what I'm getting at is presentation tweaks to reinforce the "think before you click" rather than technical obstacles
 
11:08 AM
tpu- by Henders
 
@Magisch Uh no. I think I'll need to correct you regarding this again. (something which Martijn did a few years ago). Flag bans ≠ You suck, get lost. Flag bans = You didn't get this right, but before you continue doing so, let's remind you that what you're doing is wrong.
 
@terdon I'm lumping them all in the same pile here for the sake of the perspective of a naive user
 
@tripleee Why? There are many cases where a question is crap but on topic. Those should have many downvotes but not close votes.
 
@Ferrybig that's a good idea
 
FIRE makes it easy to forget how powerful a spam flag can be.
 
11:09 AM
@Ferrybig I like this
 
@Magisch nah not your fault, there's a lot of other reasons internal to only Stack Overflow mods.
 
I like that a lot
 
@ArtOfCode Good luck! :)
 
[ SmokeDetector | MS ] URL-only title, bad NS for domain in body, bad NS for domain in title, bad keyword in body, bad keyword in title, +6 more: healthflyup.com/vedda-blood-sugar/ by barryogorm on askubuntu.com
tpu- by Henders
 
In other news, I've made a bunch of performance improvements to metasmoke, so that should be speeding up a bit. Notably, the query for review posts is down from 500ms to about 0.5
 
11:11 AM
@terdon yeah there are such cases, but more often than not, I see a lot of posts which should have close votes which don't, even though a lot of people clicked the down arrow, and more rarely the opposite
@ArtOfCode yay!
 
@BhargavRao dunno if I'm running yet :P
 
@ArtOfCode Oh no, that's just additional work. I'd advice against it. Let's focus on the future. Smokey does have a useful feature that reports conflicting feedback, just make sure that everyone is on the same line.
 
A J
@SmokeDetector Such spams should get 5 auto-flags.
@ArtOfCode great
 
That brings me to another issue, what if a Charcoal member is flagging spam, but not feeding back to Smokey? (Something which looks real bad on the mod side, and regarding which nothing can be done on Charcoal side). :|
 
@tripleee Oh, I see the opposite far too often. People tend to use close votes instead of downvoting.
 
11:14 AM
[ SmokeDetector | MS ] Bad NS for domain in body, bad keyword in body, blacklisted website in body, link at end of body, pattern-matching website in body, +1 more: up the period of time of your service by Shon mon on drupal.SE
tpu- by A J
 
@BhargavRao the regulars like to make sure we have Metasmoke up to date but of course, it's possible that somebody could go stealth
 
@BhargavRao as long as they're flagging spam correctly, not an issue
 
tpu- by Henders
 
A J
@BhargavRao that is possible.
 
if they're not it becomes a moderation issue
 
11:14 AM
[ SmokeDetector | MS ] Potentially bad keyword in answer: I want to learn some android architecture by Appbazooka App on stackoverflow.com
 
@ArtOfCode yeah, that's what I'm looking at.
 
@BhargavRao yeah, in that case it's down to site moderators
 
Okay, then that's perfect and clear cut out.
 
@SmokeDetector @ArtOfCode What happened here? There's a TPU for me but it's not showing on the post and I got a server 500 error...
 
I can say "flag better n00b", but I don't have the big red buttons that stop them
 
11:15 AM
@SmokeDetector I'd n that.
@ArtOfCode true that.
 
@Henders I'll have a look... most likely explanation is gremlins
 
A J
@ArtOfCode what is/are gremlins?
 
@ArtOfCode You didn't feed them after midnight again, did you..?
 
@AJ google 'em
 
11:18 AM
@Mithrandir Regarding Alibaba spam, better be careful. There are a few instances where Alibaba ≠ Alibaba spam. One way to make sure is to look if the post is plagiarized. Alibaba spam posts are plagiarized, but the legit Alibaba posts aren't. From the mod data I can clearly say what is Alibaba and what is Alibaba spam, but that's not possible by normal users.
4
 
@BhargavRao I've actually done it in the past (before I had FIRE)
 
I spoke to Andy about this earlier. Be very very careful while flagging on Alibaba posts. Try not to hit the legit Alibaba ones.
 
A J
@Henders got the joke now.
 
@SmokeDetector Yeah, this is a legit Alibaba spam. I had nuked that user earlier, I have no idea why I did not delete the post after closing it. (I'm getting old)
 
11:21 AM
@BhargavRao @Mithrandir and all the other people who say I should get a diamond, here’s the reason why I’m not going to try anytime soon: these days I only really visit SE for Charcoal. I may or may not be mod material, but I’m simply not active enough in actual Q&A to justify it. I know ya’ll saying it as more of a long-running joke, but I thought you might want to know why
 
tpu- by A J
 
@BhargavRao thanks for letting us know about that, I’ve been meaning to ask for clarification
 
@angussidney then you need to git gud at Q&A.
 
I've been meaning to add some crufty plagiarism detection hacks to Halflife but it probably needs to be a separate development effort
 
[ SmokeDetector | MS ] Pattern-matching website in answer, username similar to website in answer: How to merge local and live databases? by BestDesign2Themes on wordpress.SE
[ SmokeDetector | MS ] URL-only title, bad NS for domain in body, bad NS for domain in title, bad keyword in body, bad keyword in title, +6 more: healthflyup.com/vedda-blood-sugar/ by mkgumkgu on askubuntu.com
tpu- by WELZ
 
11:23 AM
@angussidney I've mentioned this to Andy in the mod room and iBug in one of the posts which they had flagged. I was not that aware that most of you didn't know bout this. Sorry for being a bit late then :(
 
A J
@SmokeDetector okay this now needs to be flagged.
 
@Henders so the technical explanation for that is that a small green thing stamped really hard on metasmoke's nose and metasmoke lost your feedback while it was trying to swat it
 
@BhargavRao well, I’ve got the ‘git’ side of things all ready, where can I install ‘gud’? :P
 
@tripleee You can look into Guttenberg which does plagiarism detection on Stack Overflow.
 
Rob
@angussidney You don't necessarily need to be active in the Q&A areas, in fact, as far as I can see, lots of the mods mainly deal with things behind the scenes
 
11:24 AM
@BhargavRao ooh
 
What if FIRE refused to flag if you had more than 10% spam flags marked as declined? That would mean that our tools basicly "enforce" our policy that you are supposed to only flag true positives
 
@BhargavRao if it happened in the past two days I wouldn’t be surprised if I missed it amongst all the noise
 
[ SmokeDetector | MS ] Bad NS for domain in body, bad keyword in body, bad keyword in title, blacklisted website in body, link at end of body, +2 more: Total Radiance Teeth Whitening Pen FAQ's by Vance19 Virk on superuser.com
 
@Ferrybig Then I would go into FIRE and disable it :P
 
@Ferrybig Oooh, perfect. +1 from me
 
11:25 AM
tpu- by WELZ
[ SmokeDetector | MS ] URL-only title, bad keyword in body, bad keyword in title, bad keyword in username, blacklisted website in body, +1 more: junivivecream.fr/tryvix/ by TryVix on askubuntu.com
tpu- by WELZ
 
I dunno if pure bad luck or something else (which I pinged @Mithrandir in another mod only room), but the number of wrongly autoflagged posts increased in the week upto posting the meta. I'm really happy that Art pushed it to 99.75. That would shut down most of the fps.
 
I'd been thinking about it for a while, so had other people, Mith suggested an actual number so I did a kinda "¯\_(ツ)_/¯ okay"
 
[ SmokeDetector | MS ] Bad keyword with email in body: 200 Hour Kundalini Yoga Teacher Training Rishikesh by kanerich89 on superuser.com
 
A J
@SmokeDetector lacks flags.
 
tpu- by A J
 
11:29 AM
@ArtOfCode That darned @Mith shuda told you a week back.
 
@BhargavRao and then I probably would've kicked it for debate rather than just doing it. Just happened to be right place, right time, right words
 
!!/watch shivatattvayoga\.com
 
@WELZ Added shivatattvayoga\.com to watchlist
 
Ah! Cheeky mith
 
@ArtOfCode what would happened if you raised it to a cool 100? Would that eliminate too many tps?
 
11:30 AM
yes, nothing is 100%
 
oh yes it is
270 up is 100%
 
Apparently, many things are here
 
oh, you're right.
 
@BhargavRao ^ graph of post volume to accuracy
 
CI on 79b6c2b succeeded. Message contains 'autopull', pulling...
 
11:32 AM
Yes. This. Is. what I wanted
 
Restart: API quota is 15147.
 
@ArtOfCode What's the x-axis?
 
Thank you!
 
@terdon Reason weight
 
@BhargavRao We'd lose about 3k in post volume for a move to 100%, in other words. Not huge, but not huge benefit either.
 
11:33 AM
Meh going for 100% is useless, 99.75 is fine. But at 5 flags, 99.99 would be better.
 
oh yeah, of course
not planning to do 5 flags at 99.75, that'll be at least 99.99 if not 100
 
@ArtOfCode The benefit would be in silencing dissenters, not any signiificant improvement in stats.
 
Remember 1 or 2 declined set of 3 flags isn't an issue, the issue is 1 flag above a set of 5 wrong flags.
 
[ SmokeDetector | MS ] Bad keyword in body, blacklisted website in body, link at end of body, link following arrow in body, pattern-matching website in body, +1 more: If you want to get young by Huntionly 89 on drupal.SE
 
@terdon yep, well said.
 
11:35 AM
But keeping the 100 for 5 flags would probably work as well.
 
I've so far gotten 2 pings on fp autoflagged posts, and both turned out to be tp (someone accidentally chose fp)
 
tpu- by WELZ
 
@terdon yup, but dissent for the current system is minimal (which is what the 99.75 threshold is for, not for the 5 flag proposal)
 
I put the auto flagging threshold on Natty at 99.95 for the reason that I was too lazy to log in and retract the flag everytime it flagged wrongly.
 
@ArtOfCode Yes, I was thinking 5 flags.
 
11:36 AM
@ArtOfCode might be a bad idea, but have you tried flagging from accounts which are active? Atleast the 4th and 5th ones?
 
Yeah. 5 flags'll be a much higher threshold :)
 
[ SmokeDetector | MS ] Few unique characters in answer, repeating characters in answer: Displaying an image from a json file in Java by user9450942 on stackoverflow.com
naa- by WELZ
 
@BhargavRao we don't have data on which those are
nor is it particularly easy to get
 
@SmokeDetector k
 
@BhargavRao: It seems this is your first time sending feedback to SmokeDetector. Make sure you've read the guidance on your privileges, the available commands, and what feedback to use in different situations.
2
 
11:37 AM
bahahaha
 
@SmokeDetector no wonder you don't store my feedback :|
 
@BhargavRao it doesn't associate your feedback because your MS account isn't authenticated with SE
 
@ArtOfCode Hmm, true. The API doesn't return this data.
@ArtOfCode sssh, you'll spoil the joke which I've been using all these days. :p
 
@ArtOfCode question on some of the above - I don't use FIRE because I've never really liked the idea, I normally like to do things on-site because some things that look like spam might be more of a problem with the question. When I flag spam manually would you prefer if I always give MS feedback as well? I don't mind doing that but always thought it was a bit redundant if someone had already given feedback via FIRE.
 
11:40 AM
@PeterJ Isn't there another userscript that gives feedback when you flag manually?
 
FDSC, yes
but... eh
 
Ahh OK thanks I'll take a look at that
 
Rob
@ArtOfCode Question about that page, do the access tokens need to be stored if they're just used to link the MS account with an SE account?
 
@ArtOfCode So more nets to protect metasmoke from small green things then...? Okay cool :P
 
You can use our product github.com/SOBotics/AdvancedFlagging ;)
 
Rob
11:40 AM
Couldn't the access token be used for a /me request and thrown away?
 
I tend to agree, @PeterJ, if there's already feedback on it and you're not using FIRE, it doesn't matter too much
 
@Rob If you authenticate read-only, that's exactly what we do with them
If you authenticate with write access we store it
 
@ArtOfCode OK will just stick with what I'm doing and only bother with feedback if a few have mounted up when things are quiet
 
JAD
@Ferrybig From BhargavRao's post on meta, isn't the problem partially with retracted flags as well?
 
11:42 AM
[ SmokeDetector | MS ] Bad NS for domain in body, bad keyword in body, blacklisted website in body, pattern-matching website in body, repeated URL at end of long post: It might at first appear to be confounded to apply on your by Vance19 Virk on graphicdesign.SE
tpu- by Cai
 
Having extra feedbacks (to me at least) seems like a good thing, it shows that it was unanimous and not just one users decision.
 
@PeterJ yup, works for me :)
 
Rob
@ArtOfCode Ah, perfect
 
@Ferrybig Oi, it's a very bold statement at the start. That particular user was wrong, not entire charcoal or any userscripts. :D
@JAD nah, I have no issues with retracted flags. My issue is with the first FGITW flag.
 
JAD
hmm
 
11:44 AM
@BhargavRao Should I just remove the line from the issue? Or do you have a better idea how to clearly state the problem?
 
New metasmoke user 'Katu' created
[ SmokeDetector | MS ] Bad NS for domain in answer: How do I post a photo on a Facebook wall? by omejames on webapps.SE
tpu- by Glorfindel
tpu- by WELZ
 
What does Bad NS actually mean?
 
nameserver
 
[ SmokeDetector | MS ] Manually reported answer: Comment on a Facebook page item owned by me, as myself by omejames on webapps.SE
tpu- by Glorfindel
 
JAD
11:46 AM
@Ferrybig I think the problem is that getting in a flag before it is deleted is turning flagging into a race, partially helped by the gamification through badges
 
@WELZ if you do a "dig -t ns (domain)" and it returns Namecheap, that's a bad NS
 
!!/watch papacookinggames\.com
 
@Ferrybig upto you, but that's a misinterpretation of my statement. There was some context on the meta, which isn't present on the issue. The line sounds weird to me, that's it.
 
@WELZ Added papacookinggames\.com to watchlist
 
I've been meaning to extend the list to cover more name servers which are predominantly or exclusively used for spam but I got sidetracked
 
11:46 AM
!!/watch deletebinghistory\.com
 
@Glorfindel That pattern looks like it's already caught by Bad ns for domain in answer and Bad ns for domain in body; append -force if you really want to do that.
 
!!/watch-force deletebinghistory\.com
 
@Glorfindel Added deletebinghistory\.com to watchlist
 
I saw one bad NS... with like 80 tp, and no fp, so it should also be blacklisted, no?
 
Should definitely be.
 
11:48 AM
[ SmokeDetector | MS ] Bad keyword with email in answer, email in answer, pattern-matching email in answer: How do I assume title as the co-signer on car loan after the borrower's death? by Compact Loan on money.SE
 
But don't just check metasmoke data. See on SE.
 
tpu- by Glorfindel
 
@JAD we could exacerbate that in FIRE too, let people flag after it's deleted if they want to (in as much as experiencing a button being clicked counts for them, which I guess will do it for most people)
 
JAD
@SmokeDetector k
 
Searching spam amongst spam and saying "oh look I found spam" is pointless. :p
 
11:48 AM
CI on 4e63fa1 succeeded. Message contains 'autopull', pulling...
Restart: API quota is 14967.
 
@BhargavRao I removed it, because if the user wasn't associated with Charcoal, it doesn't really help for supporting a change in a Charcoal product
 
It's funny, when I search it on SE... deleted spam posts come up -_-
!!/blacklist-website cluesforhealth\.com
 
@WELZ That pattern looks like it's already caught by Bad ns for domain in answer, Blacklisted website in answer, Bad ns for domain in body, and Blacklisted website in body; append -force if you really want to do that.
 
@tripleee ._. (I use namecheap ;p)
 
!!/blacklist-website-force cluesforhealth\.com
 
11:50 AM
@WELZ Already blacklisted on line 782 of blacklisted_websites.txt
 
@Ferrybig the user was associated with charcoal, his actions weren't. What I want to say is, let's groom these users into model SE users.
They will certainly get better. That I'm sure of.
 
[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title, bad keyword in username, blacklisted website in body: TryVix Cream Review: How does it works? by TryVix on superuser.com
 
eh
 
tpu- by WELZ
 
@BhargavRao I'm fairly sure the guy in your Meta answer is me :)
 
11:50 AM
Anyway, I think I'll get going. Have an entire night left to complete my submission for tomo. Do let me know if I can be of any help.
@iBug why'd you say so?
 
@BhargavRao if I do a quick check of the product, it seems like it by defaults suggests to add a comment, isn't this against the new policy? (just to inform you, not to force you to change it)
 
@BhargavRao I checked the image you posted and my flagging history, and found exactly one post that could match.
 
@Ferrybig cc @Rob, our CUSO.
Well if you've found a post that matches my description, then hey, you need to start being careful.
 
New metasmoke user 'Adam Wheater' created
 
I wonder if it's this one.
@SmokeDetector Sudden increase in new MS registries.
 
11:53 AM
I wonder why....
 
My general advice would be, don't jump on Smokey reports with a spam flag. They are reports (i.e "check if this is spam"), not instructions ("spam flag this immedaitely goddammit").
 
Yeah I know that. The only thing I'm confused is, if a retracted spam flag once a week is counted as "a lot"
 
@angussidney ikr, it's not like we've posted a meta or something. :p
@iBug then it might not be you.
 
@angussidney Obviously the meta post attracted some unfamiliar users.
 
But if you're in a position where you need to retract and then reflag a few seconds after seeing others, then you're doing it wrong.
12
 
11:56 AM
I tried to make that comment about ‘I wonder why’ as sarcastic as possible but obviously that’s didn’t come through :)
 
@BhargavRao Problem is, I do that every time.
@angussidney The comment about "you also can if you want" with a link to the MS register page?
 
@iBug that's what he's saying
 
New metasmoke user 'Sean Houlihane' created
 
@iBug Then that's exactly where you're faltering.
 
@BhargavRao OK, I will be more careful from now on.
 
11:57 AM
Why do you feel that there'a need to spam flag first and then retract?
 
Rob
@BhargavRao CUSO?
 
I'm not familiar with the link (precisely, the domain), so I kept suspecting.
 
[ SmokeDetector | MS ] Bad NS for domain in body, bad keyword in body, bad keyword in title, blacklisted website in body, link at end of body, +3 more: What are the benefits FlexinAll? by clairmack on askubuntu.com
 
Rob
@Ferrybig Indeed, it could be toggled. However, comments are only enabled for SO
That said, it'd be good to have it configurable. I'll add an issue for it
 
Can I suggest a mindshift from ‘lets increase my flag stats’ to ‘This flag will give the user -100rep and an IP block’
10
 
11:59 AM
I don't remember where, but there was this policy called "guilty until proven innocent". A man was put to jain for 63 years for an offense he didn't commit. We can equate the act of flagging a report first and then retracting to that.
 

« first day (1570 days earlier)      last day (2250 days later) »