« first day (3163 days earlier)      last day (232 days later) » 
00:00 - 14:0014:00 - 00:00

12:16 AM
Are those kinds of bots disallowed, if they DV/flag based on valid criteria?
 
@forest you can't really auto downvote
 
huh
 
I mean a machine can't effectively decide if a downvote is warranted
 
Honestly I sometimes think it would still be better than many humans...
 
And smokey aside most auto flag bots end up being a drain on resources.
Cause it's too tempting to use up your 50 daily votes at once...
 
12:26 AM
Damn, and here I was hoping I could write a bot to DV any answer that links to GRC's (a crackpot security "researcher" with a big following) website...
Thankfully I've already gone through and done it manually. :P
 
Or in the case of comment flags make things vanish without oversight enmass if you naively keyword search
 
heh
That's clever
 
@forest eh. More fun to also rip the crackpot theory to shreds
 
Well I always link to the attrition website in comments which does that for me. :P
 
@forest that's also not the intent of it
 
12:29 AM
Still a clever observation. It seems a lot of SE's design is fragile to silent misuse.
Single-handedly taking something off of HNQ by adding MathJax to the title, etc.
 
It assumes good intentions initially
 
Sure, but most systematic abuses can be detected quite easily.
There are just too many that are (unnecessarily) hard to detect.
Wow why do I have to disable the April 1 theme on each site individually??
The snowflakes are slowing down my browser.
 
1:05 AM
@forest so, a bot that just auto-votes on stuff (every new question, every new answer, every post in a certain tag, keyword, etc) isn't kosher. Same for flagging, close-voting, etc.
OTOH, an app that implements, say, a bespoke review queue might be.
bots that vote, flag, etc. based on learned heuristics tend to fall into the grey area in between those two - think, Andy's comment-flagging script, Smoke Detector, etc.
Generally, my approach to these things is... To treat them the same as we would any other user: if they're causing harm it doesn't matter what mechanism implements that harm, if they're not causing harm then who cares
 
@Shog9 Ah that's what I was thinking.
So a bot that auto-flags, say, one-word posts that contain nothing but the n-word would certainly be acceptable (although likely completely redundant).
But not a bot that auto-downvotes every post tagged with $language_I_hate.
 
yeah
now, this still gets weird in some cases
got a user on one site who has apparently been downvoting as many answers as they can for a few weeks now. Got a whole bunch of folks irritated by it. Does it matter if they're scripting that? The site is small enough you could do it over your lunch hour and still have time to eat one of those massive muffuletta sandwhiches.
 
Bad users are bad users, whether the brains behind the account are carbon or silicon.
 
OTOH, we got Slim here apparently using some Google Sheets contraption to trigger downvotes on questions that are effectively abandoned, some of which would be deleted anyway in a few months, some of which wouldn't due to having 2 comments.
 
I don't know how Google Sheets works.
 
1:16 AM
It's like Excel but with more JavaScript
Much more selective, but now it falls into that grey area where folks start to worry it's doing something they can't see.
 
It's capable of triggering downvotes?
 
Sure. Can do all that stuff with the api
 
Rob
Apparently there's a lot to headhunting but part of it's likely the usual: discussing salary, relocation, and parachute, along with clearly defining/agreeing on the responsibilities, the fit and finish.
 
1:34 AM
@Shog9 0_0
 
Yes, that's roughly the sentiment I've seen expressed lately
 
@Shog9 the thing with both SD and andy's script is though, I recall there's a ton of oversight.
 
SD, yes. Andy's script, yes - after he went public with it.
 
and with SD at least, lots of complaints aside, they're pretty responsive
@Shog9 ah. um... ;p
I'd assume at least the mods were told. Else tsk tsk.
 
393
Q: Can a machine be taught to flag comments automatically?

AndyTL;DR: Yes it can. Background On June 27, 2014 Skynet awoke. It looked at Stack Overflow and thought "Why are all these people being so chatty and talking about obsolete things? I should nuke them all!" Fortunately, Skynet was a baby and only had access to my 100 comment flags a day. Prior ...

Andy's script, for folks who weren't around in '14
 
2:08 AM
@Shog9 Reminds me of a case where a user had upvoted every question on a site in order to try and get an Electorate badge.
 
could a mod please give me a hint why answer was deleted? i know it was lame, but was it that wrong? stackoverflow.com/questions/55465509/…
 
@ocæon errrrr
Not always the best place to find SO mods...
I don't even have 10k there
 
actually, no need mods... only need 10k+ rep users on SO, but looks like most of them were Away From Tavern
 
@GetAnswerWizard technically, they can tell what the answer is, but not why it was deleted
 
well, if it was lame, then probably that's the reason...
 
2:20 AM
well i literally just answered "perhaps include formnovalidate in the date's input tag"
 
ah
that feels like it would have been better in a comment
 
i don't have the rep to comment there yet lol
 
Throwaway answers are like the dark side. Its tempting, seductive, and might actually have raisin cookies.
 
potential not-an-answer (NAA) without explaining any details. guessing game is discouraged on SE...
 
@ocæon considering its 20 rep - do you think you can flesh that out into a useful answer?
 
2:22 AM
it wasn't throwaway, but i was basically trying out some low hanging fruit to get started
 
you can then flag that post for reopening, and probably get that 20 rep
heh. There's no low hanging fruit on SO
 
oh i can edit while it's deleted?
 
yup
then flag it with a custom flag.
"Hey, this wasn't really an answer before but I've fixed it up to be a real answer. Could you take a look please and undelete?"
works
make it good though
(actually my first and only answer on SO is low hanging fruit....)
but I did get the association bonus
 
@ocæon some tips to write a good answer, basically if you want to post an answer, then write it the best you can. If you don't feel the effort compared to the question, then find other questions that you can write a good answer...
 
contrariwise. the best way to get rep fast is to post good answers to posts with terrible answers ;)
so your answer looks better.
 
2:28 AM
As controversial as it is, we don't write an answer just to help the asker, but we write an answer to help future viewers when facing the same problem...
 
lol @JourneymanGeek thanks, infact i think i'll let this one slide, i havn't done html forms for a while and perhaps there's a django related issue i wouldn't know about
 
@GetAnswerWizard Its only controversial cause people want answers NOW
I realised, eventually that every so often, I come across a problem I can't solve
only to find I solved it a few years ago.
so I post answers to help the asker, other people and myself
 
@GetAnswerWizard i see this more now i'm signed up, my first answer was.. well, not even an answer, so i went back and polished it up, though it had already hit 12 ups!
 
@ocæon I've been on SE almost a decade
I still do that ;p
 
@JourneymanGeek yeah, just stating my observation because some regular users were surprised when I stated that on many meta outside of SO and SE...
"helping the OP is secondary goal"
 
2:33 AM
@GetAnswerWizard eh, kinda sorta
We want to help OP, but also others in future with the same problem
 
but it always conflicts when you feel the question won't be useful to others
Primary example: PSQ on Maths.SE
 
well, you solve the problem first - then you build on it
 
oh answering poor questions doesn't seem to go down well at all
 
Well not always
Unless there's something neat the OP missed
or the OP has the wrong manual...
which has happened
 
Certainly: stackoverflow.com/q/27297113 (shameless plug, I answered it)
 
2:41 AM
huh.. so the bug got reported to ms but the question still got stomped on?
 
@ocæon OP assumed it was a bug
 
There's a history on that question. The "stomping" came shortly just after it has been posted, then the edit by OP... didn't really help the situation...
 
yah, difficult to see the chronology from the surface, but it is a complete mess the way it is now
 
Rob
3:01 AM
@SonictheWizardWerehog I have reviewed the link that you provided, the edits you made are fine. A better place for what we were discussing is here: meta.stackexchange.com/a/135361/282094 and that information is already present. So, nothing further.
 
@JourneymanGeek Yep, one of the last rants I've seen posted on MSE was that "people don't have time to do research if they need to get their answer immediately"
 
@SonictheWizardWerehog snort
@SonictheWizardWerehog If it was easy, its already out there.
 
hmm.. the only flag available on my deleted answer is needs moderator intervention, no custom flag, so i just label my edit and flag for an so mod?
oh right it has an explanation box . thanks all for the input!
 
@oc├Žon yup, we call it "custom flag" because we can customize the message, and posts deleted by mods can only be undeleted by mods
 
Heh, the Spanish description of the "spam" flagging option directly translates to "unwanted content"
 
3:19 AM
loool, got 'citizen patrol' badge for flagging my own deleted answer
 
@AskQuestionWizard They got to keep their Area 51 Discussions diamond (though they lost their staff bit and are thus listed as a moderator there).
But that seems typical for former staff. E.g. Bret Copeland and a few others got to keep their A51D diamonds for a small period of time after losing staff moderator rights and the staff bit.
 
 
1 hour later…
4:44 AM
How does the weekly roomba actually work? Is it runned on a very specific time (e.g. 3:00 UTC), or every minute for a whole hour (e.g. from 3:00 to 4:00 every minute), or... anything else? (ignore the time if it's wrong, it's arbitrary)
 
 
2 hours later…
7:04 AM
@GetAnswerWizard 9 days is run everyday at 03:00 UTC, 30 days and 365 days runs in the weekend, on Saturday IIRC at 03:00 UTC
 
7:26 AM
15 messages moved to Chimney
 
 
1 hour later…
8:40 AM
@rene Actually 0:00 UTC for the weekly checks in my tests
 
fair enough
 
Have we ever had cases where users have attempted to counter the Roomba by deleting their own post just before the weekly deletion scripts run, then undeleting it, so that their post is not caught by them?
 
8:55 AM
I think downvotes should not be automated
 
upvotes shouldn't be either.
Though I might write my own answer there, as the one from MadScientist doesn't suggest strongly enough for me that the roomba deletions should be more easily triggered
 
flags, okay. Edits, maybe. Pointing posts out to users so they can act themselves, fine. But votes are inscrutable by anyone but the CM team
Not a great plan
 
@Tinkeringbell I've written an answer there that proposes altering the Roomba criteria to reduce the ability for users to unilaterally delete posts just by downvoting, while adding checks to make sure that said change doesn't cause a lapse in curation and to ensure that tactical voters aren't gaming the process.
 
9:32 AM
@SonictheWizardWerehog That's just a caching issue, it'll be gone in a couple weeks, if it works properly
 
@SonictheWizardWerehog I'm not a fan. I generally want to see more stuff deleted, and lowering thresholds for automatic deletions by roomba to -3 makes a lot more stuff linger. On very low traffic sites it might even be hard to even get to that. So if anything, if this happens I'm tempted to go script downvoting myself, to get stuff cleaned up.
3
 
@Tinkeringbell Well, under my proposed plan, questions reaching merely -1 and meeting the existing criteria for RemoveDeadQuestions would be raised for moderator attention, and the mod would be expected to verify that it's not a tactical vote that caused the flag (e.g. by checking if the date of the last downvote is a long time after it was originally posted) before deleting it.
So those questions would still end up deleted anyway.
 
10:29 AM
@SonictheWizardWerehog Then we can just entirely disable roomba and always leave stuff up to moderators. I don't want extra work, but I do want more stuff to be deleted.
 
!!/watch-number 919876751387
 
@Olivia 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.
 
ninja?
ninja. xd
 
 
1 hour later…
11:41 AM
restrict access meta.stackexchange.com/questions/326214/… @Bart @rene @Glorfindel
 
The Parrot deleted it under 6 to 8 minutes ...
 
Arr! It's tinker, the scourge of the 7 seas
 
@rene usually the case, yeah. Parrot and dog are faster than most of us. :D
 
11:57 AM
@rene Sorry?
 
@Tinkeringbell I was called to action but you beat me to it.
 
Awww.
I don't have 'better' stuff to do, my gradle build is borked and it takes a long time to reach the point where it fails so I'm wasting 5 minutes or so everytime I try a new 'fix'.
 
I just pushed my 8th commit for the day
 
I spent a week on a story that we're now not going to do :/
A week and assorted hours.
 
I join you as I'm looking into a race condition that only happens when I don't have debug statements all over the place
 
12:04 PM
Commit a debug/log statement and call it fixed.
 
Myeah .... no ...
 
@rene smells like concurrency errors
 
Or currency
e.g. client doesn't pay enough
:D
 
Food for thought: When two cars collide on a race track, it's because of a race condition and would be avoided by proper use of a semaphore.
 
This here is like the nerd chat to rule all nerd chats
SE is a nerd site, this is a nerd profession, meta-chat is extra nerd. We're basically peak nerd
 
12:15 PM
@Magisch Are you calling me a nerd? :(
 
@JohnDvorak Concurrency type stuff is always pretty mind warping
@Tinkeringbell we all are :D These days, it's more used as a term of endearment
 
@Magisch there's a game called Seven Billion Humans, worth checking out
Also an inspiree by Manufactoria, but I don't remember that one's name.
 
Okay, I fixed stuff. Deleting everything sometimes works :D
 
12:39 PM
7
Q: Redefinition of pH

Karsten TheisFundamental SI units like the kilogram and the mole change definition this year. We should not stop there but redefine some quantities describing aqueous solutions of acids and bases as well. A better scale for acidity and basicity of aqueous solutions There are two aspects of the current defin...

 
Nice April Fools post ;)
 
On an unrelated note GDPR Art 13 (2) lit d) is stupid and inane
Why do I have to inform users that they can complain about me at the regulatory authority
 
That's super weird. 3 downvotes disappeared here. o_O
It was at -6 already.
 
Mod cleared spam flags.
 
ohh
Nice.
Yeah it's not spam.
 
12:49 PM
got autoflagged
 
That's bad
And why I really don't like bots flagging stuff.
 
thats a poor example
 
Not relevant.
If bots flag incorrectly even one post, it's bad.
No matter the reason.
So when three more people will install whatever runs those auto flag bots, it means off topic posts on MSE will be auto nuked as spam. Anyone really think it's a good idea? Sorry, I don't.
 
@ShadowWizard That's not how SmokeDetector works as far as I know.
 
Hopefully...
 
12:57 PM
@ShadowWizard they don't "install" bots, they give MS (and therefore Undo, but you can trust he won't chew) privs to their account so that it can be used for flagging: the responsibility is yours
 
Anyway, time to delete the fancybox question... spam or no spam... :)
@EKons but how it choose what account to use and how many? Is three the upper limit?
If so then it's not really that bad, just bad.
 
there's no upper limit, unless you use sockpuppets to evade flag bans for that :P
 
@EKons I mean, can it cast six auto flags?
 
No.
 
currently, no
 
12:58 PM
good.
 
the maximum number is (number of flags necessary for red-deletion on site) - 2
and that's not necessarily even the set number
so at least two humans review each report before it gets red-deleted
(or one site mod, that is)
 
It's also part of Shog's new meta discussion...
23
Q: What limits should we place on automated voting?

Shog9This has come up a few times now in moderator circles, but with the exception of Andy's Robot there hasn't been a lot of public discussion that's applicable. That feels kinda negligent to me, so... Here goes. The setting For the past five years, it's been trivially easy to script requests to th...

Auto flagging by bots is by all means auto voting.
 
No?
 
IMO should still be removed altogether, but guess it won't happen.
 
Flagging is much less subjective.
 
1:02 PM
@Mithrandir it'a automatic. If affects the site. So why not?
 
@ShadowWizard Voting =/= flagging
 
Flagging is much less subjective, and votes have much less oversight.
 
Voting (up and down) has implications that make policing and tracing the automated activity harder
Whether automatic flagging is allowed is not in dispute
 
Those accidental three autoflags from earlier? The oversight system kicked in - at least one out of many - and no harm was done. That's it. With automated voting causing the deletion of posts... that's much more subtle, hard to catch, and has much less oversight.
 
I agree that bots should never +1 or -1 automatically, the end
however, flags aren't votes, per se, and they're entirely different
 
1:07 PM
Though, roomba-deleted posts are still shown on 10k+ tools, right?
 
recently deleted posts? I think so
 
@ShadowWizard flagging has a record of who did what and is traceable by moderators with enough free time. voting is not
 
yeah, the roomba'd one are still shown, everyday there are many questions deleted at around 3:00
if only the "recently deleted posts" can be filtered like: mod-deleted, community-deleted, self-deleted, roomba'd...
 
yeah, I don't think so
 
yeah, but no one can see who caused it to be deleted
 
1:12 PM
mods can see the accounts who flagged
but there's no trace of if the flag was automated (not that it changes anything, since autoflags are to be dealt with just like normal flags)
 
I mean with roomba
 
oh, you mean who downvoted?
 
yes
 
I don't really like roomba working with votes anyways....
What should score matter?
If it's not locked for historical significance... get rid of it. :P
 
score does matter a lot everywhere
too few score consistently and you get banned
 
1:14 PM
well... -100 means the question isn't very well-received... ;) :P
 
@Magisch Oh, I mean for cleaning up closed stuff.
 
ah that
 
Only risk I can see there is that people start asking the same stuff over and over because a search didn't show them the closed question... then again, I'm not that faithful in humanity to think that one might search first if we're talking about people asking close-worthy questions.
 
that's simply a case where somebody doesn't read the tour properly...
 
@EKons This is not true, in the case of SmokeDetector. SmokeDetector provides it's own flag on the post. This was done at the request of moderators to more easily track where flags came from. In general, though, you are correct. It's not easy to determine automated or not via the moderator tools.
 
1:23 PM
@Andy well... actually, the SD flag alone isn't enough to determine if all (n - 2) flags were automated
there have been a few cases of, well, ninjas :P
 
No, but it's enough to say that some automation has occurred. Smoke Detector does not flag enough to delete a post.
 
um... what kind of chat delay was that
sure, the SD flag alone is automation ;P
 
A smokedetector flag existing on a post means that there are autoflags cast on the post. If I care enough to look, I'd need to use metasmoke to confirm whether or not the other flags were automated. As an example - this is what I see on a randomly selected Smokey flagged post: i.imgur.com/u0xPH1n.png
Experience tells me that 3 of those are automated based on the time stamp. The others flagged very quickly afterward but are not automated
All of this is accessible to be though because they were flags, not votes.
 
um... where did you learn to black text out so quickly? :)
also, yeah you can tell from MS ("Success"es)
 
But, Smokey is designed for transparency too. Everything it's done is accessible on metasmoke. I'd say it's well behaved in that aspect. Shog's question is about the scripts and applications that aren't so transparent in their behavior.
I wasn't trying to derail that conversation. Just providing context and a little clarification :)
 
1:30 PM
eh, Shog's question really refers to "dumb heuristics", Smokey's reasons are very intelligent compared to what he probably refers to
 
I do a similar thing for my serial voting flags. The flag text is standardized, so mods know it's me flagging based on my script, even if it's not strictly speaking a bot
 
custom flags aren't really the focus here
 
transparency is important. You can automate moderation but the moderators need to be kept in the loop and you need to be aware it's very much at their discretion in the end that you continue
 
@Magisch Also, it has your name on the flag :D
 
@Andy That may change sometime
 
1:33 PM
custom flagging for every downvote to roomba something... defeats the purpose though...
 
@Magisch "Totally not Magisch" has flagged this post :)
 
evil cactus grin
 
If you're in the SOBOTICS team you know why :p
 
I'm not.
 
It's nothing concrete yet
But if you're interested I can screencap the convo for you and send it to you seperately
 
1:36 PM
@forest More precisely, it's like a Time Travel button for JavaScript. A very old, frozen, implementation of ES 5.1.
 
he isn't in the SOBotics team? oh, he probably won't know anything about our big plans then... :P
 
rumba. it's always rumba.
 
Not to be confused with numba.
 
... weird that I could guess it's a Python framework only by its name
 
Oh well, let bots flag.
Let flags bot.
Now, where's my key?!
:P
 
1:40 PM
Did you mean: create a bot to flag other suspicious bots?
 
A meta-bot
 
Create a flag that will flag other flags.
Back to asking a new LH question... perhaps. :D
(I ditch 80% of the question I plan to ask anyway lol)
 
I still have my pending questions hidden on my About Me, but never asked them anyway...
 
> restrictions must be applicable to anyone voting via any normal mechanism
So, the question is actually "What limits should we place on automated voting, and have them apply to any kind of voting?"
Which could be simplified by removing "automated" from the title.
 
Actually, yeah, because SE mobile apps are also using the API, and I don't think they have differentiate between pure API and the apps
 
1:52 PM
If someone is still using those apps, that's suspicious activity already.
 
@AskQuestionWizard assuming they vote several thousand times using it in a short period of time...
reads up
 
@JourneymanGeek rate-limited, and out-of-vote already...
 
@GetAnswerWizard rate limiting would be one option
but... erf
Just because you could dosen't mean you should
 
Why are we even able to vote on closed (not on-hold) posts?
Up or down?
 
Good signal for Google
 
1:54 PM
@Tinkeringbell actually, that's a pretty logical question. Though I don't think there's any mechanical difference between them.
@Undo does google care about votes? 0_0
 
yes
 
@Tinkeringbell Some of them may be very useful, closed by mistake, and should be reopened after editing.
Or just reopened.
 
can't remember how I know that, and no time to dig it up, but yes
 
Hey undo
 
@AskQuestionWizard which shog's hypothetical scenario... if someone went and downvoted closed nearly roomaed questions that should be fixed...
 
1:56 PM
@Undo If you do have the time, I'd love to read up on it :) A hint as to where to start searching (is it on MSE/MSO, likely?) is okay too :)
@AskQuestionWizard it takes quite a while before something is 'closed' instead of just on-hold. Now I want to know how many actually closed questions are edited and reopened before further commentary :)
 
On the other hand, those questions that don't get so much love is also removed by 365-day roomba everyday...
 
@Tinkeringbell html microdata exposes a bunch of stuff to Google in nicely-tagged format. upvoteCount is one of those, see meta.stackexchange.com/a/233858/215468
 
@Undo Thank you! I'll go catch a train and read up while on it :)
 
00:00 - 14:0014:00 - 00:00

« first day (3163 days earlier)      last day (232 days later) »