Charcoal HQ

12:00 AM

When it's coming fast, getting six people to open the same post can be hard - it's all scrolling and no one is coordinated. Autoflagging means fewer people have to open the same post and flag it.

Catija

And peak spam time is one of the quietest times for Charcoal HQ.

Rand al'Thor

@Undo For autoflags? I dunno, man - honestly, any nonzero number of FPs for autoflags makes me feel :-/ about the whole thing.

quartata

then try a zero number for 5 flags

Rand al'Thor

Explaining to someone that their valid post jumped to -2 because a trigger-happy user flagged wrongly (and has been reprimanded accordingly) is one thing.

Undo

@Randal'Thor I'm trying to bound the problem - people here see "no false positives" and say "that's just statistically impossible"

Rand al'Thor

12:05 AM

Explaining to someone that their valid post jumped to -6 because a bunch of users' robot thought it was spam ...

Undo

Suppose I could say "We will delete all spam on your site instantly" (benefit) "with one false positive per decade, on average" (risk)

quartata

also I'll let you in on a secret: we set up autoflagging so that people who are more likely to be around to retract get picked first

Undo

ummmm we talked about doing that

quartata

Core

Undo

True

quartata

12:06 AM

(is what I'm referring to)

it's definitely correlated

Undo

Especially now. I don't have core :P

quartata

anyways this isn't productive if we're talking about autoflagging in general

Undo

@quartata No, it totally is. There's a compromise to be had here.

Catija

@Randal'Thor It's not going to jump to -6... it's going to jump to -3 (which it already does) unless it looks extremely spammy at which point it will jump to -4 unless it looks like there's no possible way it's not spam ... at which point it will jump to -5 ... and will still need someone to judge it and actually manually flag.

Undo

See, what bugs us (me, at least) is people looking at historical data, seeing one false positive, and declaring that it's fp-prone. The data and how it's presented invites those kinds of conclusions.

Shog9

12:08 AM

@Randal'Thor how often have you had to do this in practice?

quartata

@Undo we did all this over a year ago and it was fine. people backpedaling (and even claiming we never said we were going to do 3 flags) kinda gets on my nerves is all

Undo

don't worry about it

quartata

and yes, I got slammed on SFF for pointing out that we never had an FP for generalizing to the population. so that's a sore spot too

we can't really dispel that though

Undo

If we could go the other direction - you/meta define some risk threshold, and we see how much benefit we can provide the network at that risk tolerance, get back together and negotiate - I think that'd make everyone here far happier. We like numbers.

Catija

@Undo Not to mention that the flagging itself adjusts and improves over time... when we get a FP, we adjust and learn... we don't keep doing the same stuff hoping that next time it will be better.

quartata

12:09 AM

the trouble is that we already have quantified the risk in terms of accuracy. 100% from our data isn't good enough

Rand al'Thor

It's probably hard for you guys to step back and see CHQ the way people see it who haven't been so deeply involved.

Imagine you're a mod on some site and you find a bunch of random users getting together to cast coordinated flags on specific posts in order to get them deleted, organising their efforts through a chatroom not associated with that site. When you put it that way, it sounds deeply suspicious, like a voting ring/clique.

Undo

Definitely.

Rand al'Thor

Now I know you guys run a tight ship here, and you're endorsed and overseen by the CMs. But not everyone does.

@Catija -6: three downvotes from users, three auto-downvotes from spam flags.

quartata

not sure how else to get the word out then, this is our second meta post flexing our muscles about it

Catija

But does anything, anywhere, say "we learn from past mistakes"?

@Randal'Thor ... um... no. Why are good posts getting manual downvotes? That doesn't make any sense at all.

Rand al'Thor

12:13 AM

@Shog9 There's only been one autoflagged FP on my site; I don't remember exactly happened there, but I don't think a disgruntled user was involved. Then there was also a case when someone got blacklisted wrongly (and all their subsequent posts reported), and I had to come here to deal with that.

quartata

also I'm not sure why we specifically have this stigma but SOCVR gets a pass... Both of us have very strict rules, but I guess us being network wide gets more attention

Rand al'Thor

So yeah, not much. I think @terdon had an experience something like that the other day though.

Shog9

@Randal'Thor I ask, because I've had to do it a fair bit. Due entirely to human error. If you saw that list I posted above... Folks misuse spam-flags on SO quite a bit without any assistance.

quartata

62% right

was surprising

SmokeDetector

API quota rolled over with 7336 requests remaining. Current quota: 19999.

quartata

12:14 AM

sd poof-force

I'm walking here

ArtOfCode

We're hardly idiots

quartata

I think part of the problem here is blackstone's formulation

Shog9

@Randal'Thor that's really out of line. I don't think you intended it to be.

Rand al'Thor

@ArtOfCode Didn't say you were :-) But a bot has no intelligence.

Shog9

I'm going to delete that comment and give you a second try at it.,

quartata

12:17 AM

that is, it seems that meta thinks it's better to let 30,000 guilty posts linger than 1 post end up with downvotes

that sounds extreme but it is the philosophical difference

ArtOfCode

The bot itself has no intelligence, no. But the humans running it are quite good at getting it to do what it should be doing.

Undo

It's valid, though - an insight into how folks can see us.

Rand al'Thor

@Shog9 Sorry, that came out way differently from what I intended.

Shog9

It happens

quartata

maybe autoflagging has made it easier to focus on the posts that get caught in the crossfire

now that the spotlight is no longer on the crap that doesn't get deleted

@Undo I think some pretty vocal users would attest to the literal interpretation of that.

Shog9

12:20 AM

Now... The nominal purpose of this MSE post is to get some consensus on 5-flag auto-flags. But as I see it, the larger purpose is to get past a lot of the bad feelings that have arisen over how this project and folks on various sites have communicated over the past year or so.

To that end, it's going really well and I'd like to continue for another day or two.

Andy

Me too

Rand al'Thor

@Shog9 There are always idiots who can misuse pretty much any bit of the SE system. But hypothetical coordinated idiots all flagging the same post would be a lot more serious - and when an unthinking bot coordinates several people's flags, the effect (until a human steps in to cancel those flags) is pretty much the same.

better? (sorry again)

quartata

don't think a day or two is enough to hammer this out

Shog9

If it ends with a much larger population of folks on the network able to explain what SD does and how it does it, that'll be a win, even if y'all never move beyond 3 auto-flags.

quartata

we seem to be losing the PR battle (one that shouldn't be a battle in the first place, not because we're always right but because we should all have the same goal)

Catija

12:22 AM

@Randal'Thor Calling Smokey a "dumb bot"... is a bit rude to the people who have spent so much time teaching it. Could you be... at least a bit cognizant of where you're speaking?

Rand al'Thor

You say part of the purpose of autoflagging is to cover those hours when there aren't enough people here to flag manually. Conversely, that could mean there aren't enough people here to cancel mistaken flags either.

@Catija I'm British - "dumb" doesn't mean "stupid".

Catija

@Randal'Thor You're also a grammar pedant who knows that isn't the case everywhere.

quartata

ok back it up

we've all had enough bombs hurled about this for one lifetime

Andy

@quartata You're throwing out other's opinions because they disagree with your own

quartata

no? Calling someone a pedant is out of line in this context. I'm assuming some level of good faith because of diamonds but

Catija

12:25 AM

@quartata He calls himself that all the time. Would you prefer I put it in quotes?

Rand al'Thor

@quartata I'm a proud pedant!

Andy

@Randal'Thor and other mods in here, Shog mentioned something above about improving communication between CHQ and your communities. A yearly (or less) meta post isn't enough. What would help with just the communication front.

quartata

thanks

we have the RSS feed in a PR

so you can feed that into your mod room

Andy

Preferably without the associated pokey sticks (unless it's deserved)

quartata

speaking of which @thesecretmaster is that good to merge yet

Rand al'Thor

12:28 AM

@quartata Re PR, I want to come back to this again. You know that you know what you're doing. But if you worded your meta posts more as if you had to convince people that you're not a vote-ring clique, you might get less pushback. Remember, the people you need to convince aren't those like me who already know what you're doing; it's those who've never heard of you and are suspicious of the whole venture.

thesecretmaster

Just fixed @ArtOfCode's changes, waiting for CI

Sorry it took so long, I was fighting with webpacker again (and lost)

SmokeDetector

[ metasmoke ] ci/circleci success on 537ee6a: Your tests passed on CircleCI!

Rand al'Thor

Now it's past midnight here and I've made enough faux pas for one evening, so I'm dropping out :-)

Shog9

@Randal'Thor So, lemme try to come at this another way.

quartata

@Randal'Thor Not going to take responsibility for the wording of the meta post. All I can tell you is that we kept it general and flexible because we were concerned if we slightly tweaked our approach we'd get bashed for doing something meta didn't agree too

thesecretmaster

12:29 AM

Should be good to merge now (cc @Undo @quartata)

Shog9

We get folks who contact us all the time. Via email, via meta, via twitter... Who think that the folks who use these sites are just the world's biggest jackasses.

quartata

@ArtOfCode

Shog9

Little hitlers, out for nothing but to cause the world pain.

And I gotta say, it really hurts to see that.

Because I think these sites and the people who built them are really wonderful.

SmokeDetector

[ SmokeDetector | MS ] Username similar to website in answer: How can I check if an Ethereum address is valid? by Balidator on ethereum.SE

Shog9

But... These folks don't see that. They don't see the hours of selfless work, of patient responses and refinement, that've gone into just about everything here.

quartata

12:31 AM

sd f

(disclosed)

Shog9

All they see are folks telling them that they can't ask big vague questions, they can't vote without demonstrating knowledge, they can't go on for hours about nonsense in the comments, they can't promote their product...

So in effect, there's two Stack Overflows:
- the one we built to help folks share their knowledge that's been wildly, wildly successful
- the one folks think exists to answer questions, that's full of pedantic jackasses

SmokeDetector

[ metasmoke ] ci/circleci success on 6c02632: Your tests passed on CircleCI!

quartata

oi

he's in the middle of something here

Shog9

And right now, that's where we're at with Smokey, @Rand: there's two Smoke Detectors...
- ...the multi-year project to identify and classify spam involving numerous skilled volunteers and endless painstaking research and care...
- ...the group of bored idiots who want a script to flag for them for TEH BADGES & POINTS

In both cases, the second one doesn't actually exist.

WELZ

all teh badgez and pointz

Shog9

12:35 AM

But it's what folks see, because we've done a shitty job of communicating the one that does exist.

So that's where we need to do better.

Literally nothing else - not opt-out, not opt-in, not endless dashboards and tools and arguments on every site now or in the future - will help without that.

quartata

Thanks. That was relatable to say the least

Shog9

This MSE featured post is the start of that. Not the end. But hopefully it's a good start.

@Randal'Thor that make sense?

K.Dᴀᴠɪs

\o\ . |o| . /o/ . |o| . \o\ . |o| . /o/ . |o| . \o\ . |o| . /o/ . |o| . \o\ . |o| . /o/ . |o|

Undo

@Shog9 This is where automated meta posts on a schedule would help a lot.

ArtOfCode

I think part of communicating is tooling - like you were suggesting, perhaps casting a custom flag. There may be other tools we can build, too. But... you're right, people want to talk to people about it like people, not people mindlessly backing the bot. Need to be careful we don't come across like that.

Shog9

12:38 AM

I've spent years being an apologist for human flaggers, who as a rule are often far less predictable and excusable than the folks using SD. I'm often kinda... discouraged... about my ability to help folks see SO for what it really is. But I think for SD y'all are capable at succeeding where I've failed.

thesecretmaster

@ArtOfCode I don't know if this idea has gone anywhere (I haven't been able to read the transcripts), but couldn't room meetings help to actually communication to people what CHQ is and what y'all do?

ArtOfCode

They might do, yeah, but people need to know they're happening

They haven't happened yet because we've been focusing on the meta post, but I'd like to.

quartata

Yes. SOCVR does these. The trouble with them is that they're on way too strict of an intinerary IMO. We need something more casual.

Undo

But a small enough number of people that it's not unmanageable

thesecretmaster

@ArtOfCode Sure. There's no silver bullet that'll solve the problem. We need to try to reach people through many different methods.

quartata

12:43 AM

like with the SOCVR meeting you have to submit a question ahead of time as a GH issue. It's only really for internal use I feel

this needs to be open to all like a Town Hall, but also moderated like a Town Hall which is rough

since we're not CMs

thesecretmaster

There was talk of asking 2 non-CHQ mods to moderate?

Undo

Maybe a secret here is that not all concerns need to be addressed. When someone responds to a meta post, we don't have to convince every person. Consider everything, don't engage everything.

quartata

Ooh. We can call them "Charcoal Fireplace Chats" hahahaha

~~no gas grillers allowed~~

thesecretmaster

@Undo I think that's how you make people feel like CHQ is just defending their bot and will do whatever they want without regard for any objections... etc etc

Catija

New Chat room : The Grill

K.Dᴀᴠɪs

12:45 AM

Charcoal Grill, duh

face palm

thesecretmaster

@Undo And all concerns can't be addressed, but that doesn't mean it's not worth trying.

quartata

@Catija even better

WELZ

The roasting room.

where we roast all the haters

quartata

go ahead and grill us, we'll sort things out

@WELZ now now

Catija

@WELZ I think you have it backwards... we're supposed to be listening to them, not mocking them.

WELZ

12:47 AM

> in order to listen... you have to mock (said no one ever)

Undo

@thesecretmaster That's how we come off when we answer concerns with numbers, though.

Because... for some reason, the numbers all support us.

Rand al'Thor

@Shog9 Yeah, it does. I think your analogy is a good one: the "good" and "bad" views of SD just like "good" and "bad" views of SO. Of course, the people who already hold, or are predisposed to hold, the "bad" view are the hardest to convince. If you can bring them round to the "good" view, that's a real victory.

Undo

@Undo It's like we've spent the last three years making that happen. And it's a little inconvenient when you can't tell the truth without seeming partial.

quartata

We'll have to think about the format more. I kinda like the idea of first muting the room (gallery + write access) to let someone present a question, we'll give a succinct answer, then we open it up for comments

Rand al'Thor

Really really going this time. Goodnight all :-)

Undo

12:49 AM

night :)

quartata

@WELZ stop. stop. no politics. not here.

K.Dᴀᴠɪs

o/

Catija

@quartata So... like the Tavern...

If we wanted to shut the mods up, we'd have to get a CM and do it on MSE.

quartata

I don't know if that's how they do it but yeah sure whatever

Catija

Or, if we can trust the mods to shut up, we can do it here.

thesecretmaster

12:50 AM

@Undo I think that when people are seeing the numbers and aren't saying "Wow, I now have total faith in CHQ", they can still have valid concerns. Everything isn't numbers, and PR especially isn't numbers.

quartata

Tell you what

give me a bit here and then I can try and make a mock one

Catija

For the town hall chats, they'd put the room in timeout for three minutes and talk about stuff, then they'd open it to questions.

quartata

I think we can do this where it's regimented but not ornery

Undo

format.rss

well that's convenient

WELZ

@Catija and the room wouldn't explode?

thesecretmaster

12:52 AM

@Catija You could also make a bot that's an RO that'll delete all messages posted in that 3 minute period not by users x, y, and z.

ArtOfCode

@thesecretmaster too much effort

Catija

@WELZ Well, it was MSE... so the turnout was sort of hit and miss.

thesecretmaster

@Undo Doesn't every mime-type have a format.something?

quartata

@thesecretmaster this is kinda what I'm writing right now. basically I want to set it up so people get in a line to present their questions

Catija

@thesecretmaster Nah, I think we can trust the mods to be quiet if it's not their turn.

WELZ

12:52 AM

@thesecretmaster sounds like he's volunteering @ArtOfCode

thesecretmaster

Uh oh

quartata

first we answer with a quick response, then the room opens up for anyone

then once that's over we go to the next

not sure whether the person themselves should state the question in chat or the bot says it

Catija

We'd talked about this before and even picked people to manage it...

WELZ

Maybe there should be Charcoal Chat, so that SD reports and other related stuff doesn't get bumped in.

quartata

Yes. Can't be here.

Catija

12:54 AM

Hence "The Grill"...

quartata

Yep. Definitely The Grill.

Yvette Colomb

@Undo if they're used to them, they'll eat if skin and all! So do sheep! I've hand fed them to both :) It's fun - it feels so nice when sheep nuzzle your hand

Undo

Metasmoke deploying

quartata

d'awhhh

thesecretmaster

YAY

ROs: Anyone object to testing the RSS in this room (as a ticker, not posts)?

SmokeDetector

12:55 AM

[ SmokeDetector | MS ] Bad keyword in answer, bad keyword with email in answer: What is the best way to compute trending topics or tags? by rolandanna on stackoverflow.com

ArtOfCode

ew tickers

but yeah sure

SmokeDetector

tpu- by K.Dᴀᴠɪs

Undo

Why not another room?

Add a site to it, then we can easily comment & point to it on answers to that meta post

ArtOfCode

If we do a town hall chat, we need to (a) collect questions beforehand, (b) specify designated people to answer them in that first 3-minute response so that we don't fall over each other

thesecretmaster

@Undo OK

Yvette Colomb

12:56 AM

@Shog9 you've had a tremendous influence on shaping the site - not to be underestimated. There will always be people who are contrary/oppositional and then there's the ones who continually fail to understand

Undo

@ArtOfCode I think it's a good idea, but it seems like a high-effort way to start. Starting with some simple stat posts might be the best ROI

ArtOfCode

I'd also wonder if it'd be worth doing the first hour formatted like that, and having a second optional hour for more freeform discussion.

@Undo Yeah, agreed

Undo

Then - once things are a little less explosive - think about chats.

thesecretmaster

RSS Test Room

WELZ

@ArtOfCode and the third hour for direct war?

Undo

1:00 AM

@WELZ Profanity only, and ideally some projectile warfare.

WELZ

Lol!

Catija

@ArtOfCode ouch... that's long. Half hours to start?

ArtOfCode

@Catija I don't think we'll get enough in. Recall that the March 1 discussion ended up lasting 5 hours...

Catija

Mashed potato cannon at the ready!

quartata

@ArtOfCode My plan is that the bot will collect them from a little Bottle app as the chat goes on (with a time limit obviously), mute the room and post them, one designated person answers quickly (this may take some outside coordination), then x amount of time for free form discussion before moving on

I think that's a good compromise between structure and easy-to-ask

ArtOfCode

1:02 AM

I think that's overkill for what we want, but eh. Implementation details.

Catija

@ArtOfCode there's nothing saying that it has to stop after an hour but I think 30 minutes of planned content is sufficient.

ArtOfCode

shrug we can try

Catija

Lots of people will be at work...

WELZ

We will have renowned speakers from across the globe. (who sit at their desks all day and control online forums :P) [crawls into corner]

quartata

@ArtOfCode I'm bored and I can do it fast so

thesecretmaster

1:05 AM

@quartata This reminds me -- do I owe you any ChatX changes?

quartata

I'll write it then do a mock one to test the waters

We can tweak it later

@thesecretmaster no you're all clear for now

thesecretmaster

\o/

quartata

I may want some EM integration but I suspect I'd do that myself if we need it

we probably won't -- I'm already deferring posting the messages, and it needs to be in order anyways so threads are better suited

cause critical regions

Mithrandir

(why do the interesting conversations always happen when I'm asleep? i can't always be awake until 3)

Yvette Colomb

@Mithrandir tell me about it - I'm in the opposite time zone

Mithrandir

1:09 AM

It's 3AM here now, I woke up because notifications and phone buzzing and decided to peek in

SmokeDetector

[ SmokeDetector | MS ] Offensive answer detected: Difference between Steam and Fog / Mist by bol on physics.SE

tpu- by K.Dᴀᴠɪs

WELZ

@SmokeDetector still alive folks

thesecretmaster

Huh. Where'd everyone go? That's been alive for 10 minutes now...

WELZ

!!/scan physics.stackexchange.com/a/390618

SmokeDetector

[ SmokeDetector | MS ] Offensive answer detected, blacklisted user: Difference between Steam and Fog / Mist by bol on physics.SE

WELZ

1:23 AM

oooooooooooor not

quartata

Nope. We only autoflag the first time

Otherwise we'd cast too many

WELZ

we didn't autoflag at all the first time.

sd k

@ArtOfCode @Catija any update on US bound stickers?

ArtOfCode

should be with Catija by tomorrow

a couple of Europe people have got theirs

WELZ

Stickers Tip: Make sure to put it on from one side at a time and sorta press down and drag, otherwise you'll end up with air bubbles.

their stickers (in particular) will create air pockets if you don't

thesecretmaster

It appears that !!/scan made a second post in metasmoke

ArtOfCode

1:28 AM

indeed

thesecretmaster

That is not intended behavior, is it?

Because have double posts can skew stats.

ArtOfCode

intended as long as it doesn't autoflag it second time round

WELZ

it's still there !

thesecretmaster

1 more flag...

WELZ

@thesecretmaster unless someone downvoted

poof

thesecretmaster

1:30 AM

Ah

quartata

1:40 AM

I forgot how much of a pain SE OAuth is...

maybe I don't need that

can't use OpenID after july 1st eh? So I have to do it this way

thesecretmaster

Ah darn it. OpenID not being supported will break all my chat bots :(

Grant Garrison

Ok, so I'm reading about this awesome anti-spam bot, but I'm not finding the install/download link. Am I just stupid?

WELZ

Makes me rethink their real reason for discontinuing it...

@GrantGarrison There's no installation, it all runs server side.

Mithrandir

@GrantGarrison There is none... it's running in this room.

!!/alive

SmokeDetector

@Mithrandir Watching this endless list of new questions never gets boring

Grant Garrison

1:49 AM

So you just flag anything it says?

Or is there an automatic system?

quartata

@thesecretmaster Oh... Uh oh. That will break us too....

Uhhhh

thesecretmaster

Well, time to update chatx

quartata

@Undo @ArtOfCode You know CE better than I do, no OpenID will be a big problem right?

WELZ

@GrantGarrison I would read through some of this training

Mithrandir

@GrantGarrison No, we don't just robotically flag.

Every post is reviewed by humans.

Undo

1:51 AM

@quartata I don't know how CE works, actually. Just that it usually does

WELZ

and the wiki

ArtOfCode

@quartata no

thesecretmaster

I believe it'll break. I stole that bit from ChatX, IIRC

quartata

are you sure

WELZ

@Mithrandir and/or dogs

ArtOfCode

1:52 AM

openid.stackexchange.com is staying

quartata

last time I had an account with no OpenID login it didn't work

Mithrandir

@ArtOfCode that's not what I saw

quartata

@ArtOfCode that's not what it sounds like. Sounds like we can't have any OpenID logins.

Mithrandir

Looks like it'll be deprecated soon

thesecretmaster

@ArtOfCode "deprecate openid.stackexchange.com "

quartata

1:52 AM

Pretty sure we and all of SOBotics are screwed

SEDE will also break

thesecretmaster

Whee

Grant Garrison

@WELZ ok

WELZ

in Nuclear Experiments, Feb 23 at 22:07, by SmokeDetector

Goodbye, cruel world

quartata

hahahaha

ArtOfCode

@Nick_Craver Does that mean https://openid.stackexchange.com is going to die? I should probably start migrating accounts off it if so...

Tweeted by _ArtOfCode on February 14, 2018 at 2:43 PM in reply to Nick_Craver

WELZ

1:53 AM

@quartata I'm thinking they may not have considered ths....

Mithrandir

> July - End support for logging in via OpenID and deprecate openid.stackexchange.com (note: only 0.2% of authorized sites in the last year have been non-Stack Exchange sites. Running some additional analysis and will update when we have more info.)

quartata

@ArtOfCode ...really? their new meta post sounds nothing like this

they even said that SEDE would break actually

thesecretmaster

And regardless you won't be able to log in to chat with your SE openid

Catija

@WELZ They're here. :D Will pull stuff together and get them out tomorrow or Thursday.

quartata

yeah ok we're boned, Tim Post's answer even says so

ArtOfCode

1:56 AM

I sent Nick another tweet

quartata

This is kind of a nuisance

@JeremyBanks you around?

Shog9

Nick's kinda out of town right now. If this is going to break stuff, please make sure you write an answer on that meta post so we can plan accordingly.

ArtOfCode

@Shog9 problem is we don't know if it's going to break stuff

Anonymous

@quartata sort of.

Shog9

@ArtOfCode oh?

thesecretmaster

1:58 AM

I'll fix ChatX, and maybe someone with python skillz can fix ChatExchange

ArtOfCode

ChatExchange logs in by logging into openid.stackexchange.com and using that cookie to log into an SE site. Is that a flow that's going to break or not? We don't know.

Grant Garrison

@Mithrandir What about this? charcoal-se.org/training/autoflagging

Shog9

@ArtOfCode assume "yes" and answer accordingly

ArtOfCode

@Shog9 copy, answer incoming

thesecretmaster

Wait a minute... log in with stack exchange posts to openid.stackexchange.com/affiliate/form/login/submit

WELZ

1:59 AM

@GrantGarrison autoflagging is only on posts that have a greater than 99.75% positive spam rate

thesecretmaster