Charcoal HQ

Mithrandir

11:00 AM

@ArtOfCode could also superping

doppelgreener

!!/watch hypez\.com

SmokeDetector

@doppelgreener 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.

Glorfindel

@doppelgreener Petter was faster than Smokey.

ArtOfCode

@Mithrandir eh, those can get lost too. Easier to just go do it manually

SmokeDetector

[ SmokeDetector | MS ] Bad keyword in title: How can i detect acne in images using image processing, opencv? by Omer Javaid on stackoverflow.com

doppelgreener

11:00 AM

!!/test hypez.com/buy-instagram-likes

SmokeDetector

> Url in title, potentially bad keyword in body, potentially bad keyword in title, potentially bad keyword in username
----------
Title - Position 9-18: hypez.com
Title - Position 1-18: https://hypez.com
Body - Position 9-18: hypez.com
Username - Position 9-18: hypez.com

oh good

fp- by Glorfindel

Heh

My flag condition was disabled. What was that about?

Cerbrus

11:02 AM

Actually, maybe users not updating their conditions after a superping is a good thing

angussidney

My 190 weight reason is 0.06% more accurate than my 265 weight condition

Cerbrus

It could be in indication of abandoned accounts

angussidney

@NVZ we've upped the min accuracy to 99.75% from 99.5%

SmokeDetector

[ 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: That is the point at which I helped myself by hogumnode on askubuntu.com

tpu- by Cerbrus

NVZ

@angussidney Okay. Anything I should do to restore my flagging feature?

angussidney

11:02 AM

Yours fell under and thus was disabled. You'll need to readjust it back up

@NVZ just edit your conditions

NVZ

Okay. Thanks.

angussidney

Night all o/

Cerbrus

@NVZ Here: metasmoke.erwaysoftware.com/flagging/conditions

If your autoflagging was disabled, update your flagging conditions here. Your conditions should have a min. accuracy of 99.75%

6

Mithrandir

@angussidney g'night!

SmokeDetector

[ SmokeDetector | MS ] Bad NS for domain in body, bad keyword in body, pattern-matching website in body: contribute a noteworthy measure on acquiring by Swf tfgr on apple.SE

tpu- by Cerbrus

Cerbrus

11:05 AM

Too many pins :P

ArtOfCode

I nixed one as well

Cerbrus

Maybe andy's link should be formatted, by the way

SmokeDetector

[ SmokeDetector | MS ] Offensive answer detected: Connect an Arduino directly to RGB LED strip by slidkjfpqouhg on arduino.SE

tpu- by iBug

NVZ

I've re-enabled my magic flags.

iBug

Am I the one with the most contribution to code among all non-developers?

NVZ

11:08 AM

@iBug You forgot to add the meme "Am I the only one".

iBug

@NVZ I know. You're the only one who ranked higher than NVZ in the SO rep all time league.

doppelgreener

Do we have a list of commands and how to use them anywhere?

angussidney

Yep

Mithrandir

@doppelgreener charcoal-se.org/smokey/commands IIRC

angussidney

Under ‘commands’ in the wiki

SmokeDetector

11:11 AM

[ SmokeDetector | MS ] Few unique characters in answer: When should I not use IIS 8.5's automatic cert rebinding feature? by Palabok de Ensaymada on stackoverflow.com

NVZ

@doppelgreener You can't just use them commands "anywhere". Pick a place.

SmokeDetector

tpu- by iBug

naa- by Cerbrus

iBug

@Mithrandir 404

angussidney

I expect something to blow up tonight in relation to the meta post.... so have fun everyone and don’t do anything stupid :)

Cerbrus

@SmokeDetector retracted

angussidney

11:12 AM

Actually going to sleep now :P

iBug

@angussidney we're going to submit it?

Mithrandir

github.com/Charcoal-SE/SmokeDetector/wiki/Commands

NVZ

Anyone noticed Google's page on Henry Cavill? Prank? i.stack.imgur.com/ZmCSH.png

doppelgreener

@NVZ not the sense of "anywhere" i meant. :)

Cerbrus

@angussidney Aww Okay... *disassembles bomb*

angussidney

11:12 AM

@iBug submitted*

Cerbrus

nite

7 hours ago, by Andy

Big Fancy Meta Post

iBug

I see..

NVZ

@doppelgreener I think the joke missed someone. :P

doppelgreener

:D

SmokeDetector

[ 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, +5 more: www.oralhealthplus.com/dream-fit-garcinia/ by Shereds1964 on askubuntu.com

tpu- by iBug

doppelgreener

11:14 AM

@Mithrandir That's it. This one doesn't mention/describe !!/scan. Could someone who knows how it works document it? (just above/beneath !!/report would probably be a good place, if it is a privileged command.)

iBug

sneaky change :)

Art just increased the threshold from 9950 to 9975

@ArtOfCode What are the top 10 sites in terms of FP count?

Or just 6th to 10th

SmokeDetector

[ SmokeDetector | MS ] Blacklisted website in body: How can solve remote issues? by techelpp on askubuntu.com

tpu- by iBug

iBug

ah hah

I just added information about !!/scan into the wiki.

SmokeDetector

[ charcoal-se.github.io ] continuous-integration/travis-ci/push success on e3a1da9: The Travis CI build passed

Mithrandir

@SmokeDetector uuuhhh

That's from 3 days ago

iBug

11:30 AM

@Mithrandir 3 days ago? I just edited the commands

SmokeDetector

[ SmokeDetector | MS ] Blacklisted website in body, link at end of body, link following arrow in body, pattern-matching website in body, repeated URL at end of long post: What is Enhance Mind IQ? by Doduchis 45 on workplace.SE

tpu- by iBug

Mithrandir

@iBug not on Charcoal-SE.org though

!!/feedback workplace.stackexchange.com/questions/107679 tpu-

SmokeDetector

@Mithrandir No such feedback.

Magisch

I went from 180 weight to 185 weight for flagging

Went from 99,74% -> 99,76%

SmokeDetector

tpu by Jake Symons on Text Annotation tools [MS]

Conflicting feedback across revisions: current, #1

tpu by Jake Symons on Are there any tools to make text labeling faster? [MS]

tpu by Jake Symons on Text Annotation tools [MS]

Mithrandir

11:39 AM

I've adjusted my conditions to be 99.95% accurate.

...up 0.02% from what it was before :P

SmokeDetector

[ SmokeDetector | MS ] Bad keyword in link text in body, blacklisted website in body: How can solved sound issues? by techelpp on android.SE (@iBug)

tpu- by Glorfindel

iBug

PR #1677

SmokeDetector

[ SmokeDetector | MS ] URL-only title, bad keyword in body, bad keyword in title, blacklisted website in body, blacklisted website in title, +3 more: www.facts4supplement.com/total-radiance-teeth-whitening-pen/ by JorgeTyler on askubuntu.com

Glorfindel

@SmokeDetector k

SmokeDetector

[ 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: The essential settling, ID-alG, generally called Brown by Fisir19 Virk on superuser.com

tpu- by Glorfindel

SmokeDetector

11:55 AM

[ SmokeDetector | MS ] Bad keyword in body: Got a low credit score? by Leahy Petra on money.SE

tpu- by iBug

Jake Symons

@SmokeDetector k

@ArtOfCode why the policy change?

SmokeDetector

[ SmokeDetector | MS ] Few unique characters in body, mostly dots in body: how to make create a random combination of fixed lenght using python by Ewout Pollaris on stackoverflow.com

Glorfindel

@SmokeDetector v

iBug

er?

Where can I see the flag count?

Glorfindel

Which flag count?

ArtOfCode

11:58 AM

@Mithrandir latest repo commit gets rebuilt when the wiki gets updated

@JakeSymons read March 1 transcript

iBug

Total autoflags by MS

Magisch

@ArtOfCode I think the debate is going to be opt-in vs opt-out

Not sure how to adress that concern

JAD

@iBug metasmoke.erwaysoftware.com/flagging

under stats

> 223 active participants
81023 spam flags total cast over about 1 year

iBug

No post count.

JAD

you said flag count :o

Jake Symons

12:03 PM

@Art is this the right issue - github.com/Charcoal-SE/SmokeDetector/issues/1665

iBug

my bad

Henders

Bah, the star board is busy these days...

ArtOfCode

site_name, TPCount, FPCount, TPPercent
Stack Overflow, 16771, 7463, 64.6232
Super User, 8428, 700, 90.6139
Ask Ubuntu, 12391, 476, 95.5727
Mathematics, 685, 346, 60.8348
English Language &amp; Usage, 804, 312, 67.1119
Ask Different, 5846, 274, 94.0779
Android Enthusiasts, 827, 257, 67.9540
Magento, 207, 196, 48.3645
Server Fault, 374, 193, 62.3333
Information Security, 2070, 193, 90.0000
WordPress Development, 313, 186, 58.2868
Arqade, 1526, 168, 85.4423
Travel, 850, 160, 80.0377
Movies &amp; TV, 420, 159, 70.0000

@iBug ^

A J

!!/blame

SmokeDetector

@AJ It's John Dvorak's fault.

ArtOfCode

12:04 PM

@JakeSymons yes

A J

@Henders ^

Henders

@AJ Any opportunity ;)

A J

@Henders No. I haven't used it in a long time. ;P

Glorfindel

@Henders $("#starred-posts").hide()

iBug

@ArtOfCode Thanks. Can you take a look at PRs 1676 and 1677?

SmokeDetector

12:05 PM

[ SmokeDetector | MS ] Few unique characters in answer: Concentric Squares in Python with Turtle by yeboihmmbsudsao on stackoverflow.com

tpu- by Glorfindel

ArtOfCode

@iBug urgent?

iBug

@ArtOfCode not urgent, but I think 1676 can be merged directl

Magisch

Any reason it's not featured yet btw?

ArtOfCode

Shog hasn't done it yet

probably asleep at the moment

@iBug okay, I'll look later

iBug

wtf is that error???

terdon

12:09 PM

@ArtOfCode what are those numbers? The tp rate is way, way higher than that, isn't it?

Jake Symons

@ArtOfCode can the auto-flags just be increased, is it the social aspect we are tacking with mods and stuff via the meta post?

ArtOfCode

@terdon Those are post counts per site for all posts Smokey's caught, ordered by highest FP count.

Smokey itself ain't actually that accurate, about 80% overall. Smokey aims for zero false negatives.

Autoflagging is way more accurate, aiming for zero false positives.

iBug

finally got that CI working.

terdon

@ArtOfCode Ah, right, smokey. Fair enough.

ArtOfCode

@JakeSymons there's probably some technical stuff to do if we do increase them, but it's mostly the people stuff at the moment

Mithrandir

12:12 PM

@ArtOfCode where do NaAs fall in there?

ArtOfCode

@Mithrandir they don't

SmokeDetector

[ SmokeDetector | MS ] URL-only title, bad keyword in body, bad keyword in title, blacklisted website in body, blacklisted website in title, +3 more: maximizedmuscleideas.com/lyaxtin/ by chaicoffee on askubuntu.com

tpu- by Glorfindel

[ SmokeDetector | MS ] Bad NS for domain in body, bad NS for domain in title, bad keyword in body, bad keyword in title, blacklisted website in body, +5 more: www.supplementschoice.com/dream-fit-garcinia/ by rbhrrbhr on askubuntu.com

tpu- by iBug

[ SmokeDetector | MS ] Blacklisted website in answer, blacklisted user: How to use a smart TV as a monitor via built-in Wi-Fi adapter? by techelpp on superuser.com

tpu- by iBug

iBug

Improvement on the meta post is very discouraged?

tripleee

@iBug there was two weeks of encouragement to review the draft, maybe now at least discuss here first before making nontrivial changes

ArtOfCode

@iBug sorry, hadn't got round to feeding back to you here yet. I rolled that one back because it was mainly thousands separators, which are done that way deliberately.

iBug

12:23 PM

my edit was "non-trivial"?

@ArtOfCode May I re-apply it without the thousands separator change?

ArtOfCode

@iBug that makes it just "increase that number to 5 flags" vs "increase that number to 5", which... eh

iBug

@ArtOfCode I'll drop that change, too

ArtOfCode

there were no other changes...

so yeah you could re-apply that edit but it'd be an edit of nothing ;)

PeterJ

I think it should be five not 5 :p

ArtOfCode

we had that discussion on the gist too :P

Magisch

12:26 PM

@ArtOfCode Did we decide on opt-in vs opt-out yet?

Tinkeringbell

five/5... that'll help both people with dyslexia and dyscalculia :P

iBug

OK, I re-applied the edit. Now there shouldn't be any visual changes.

ArtOfCode

yup, works for me

Glorfindel

TBH, using spaces as thousands separator in normal text s***s (in tabular data, it's OK). Here's why:

iBug

We'd like to increase that number from III to V, :)

Glorfindel

12:29 PM

BTW, that's something which couldn't be spotted in the draft.

Magisch

There's a long debate going on under Journeyman's answer

iBug

Looks like Glorfindel's using iPad Pro 12.9"

ArtOfCode

@Glorfindel should be fixed, I made 'em non-breaking

Glorfindel

> MacBook Pro (Retina, 15-inch, Late 2013)

@ArtOfCode thanks. Still, it's silly to use 6 characters where a single , would suffice.

ArtOfCode

but i18n

Glorfindel

12:35 PM

The post is in English, people like me who live in a country using . as thousands separator are used to ,.

tripleee

@iBug looks like somebody is assuming everybody runs their browser full-screen

Glorfindel

Especially when dealing with discrete entities like posts.

Cerbrus

Let's use | as separator :D Then everyone will have to get used to it as much as everone else.

iBug

@Cerbrus Good suggestion :P

Glorfindel

Bad idea. Everybody knows 29|592 is 605.

tripleee

12:37 PM

the Swiss (I think it is?) convention of 1'000'000 is kind of unintrusive even to the unfamiliar

John Dvorak

Let's useas the separator. Works just fine with thousands.

tripleee

that looks like a newbie trying to post HTML without markup (-:

John Dvorak

@tripleee until you get to representing the decimal period

tripleee

"I mean I'm typing this --> but I see nothing"

John Dvorak

How about 1*000*000 ?

ArtOfCode

12:39 PM

=0

John Dvorak

/me curses markdown

1<i>000</i>000

iBug

1$000$000

John Dvorak

just stick to spaces...

tripleee

or the inevitable 1¯\_(ツ)_/¯000¯\_(ツ)_/¯000

WELZ

Why was autoflagging min raised to 99.75% ?

Cerbrus

12:41 PM

Because higher is better :D

WELZ

😈

Cerbrus

Seriously though, I think it's along the lines of: There's no reason not to. Let's enforce higher accuracy

John Dvorak

Looks better in stats

Mithrandir

Higher accuracy = less autoflag fps = people trust us more

Cerbrus

"higher is better" ^

Mithrandir

12:44 PM

Autoflags aim for 0 false positives. So the higher accuracy we enforce the closer we come to that goal.

Magisch

@Cerbrus 99,75% seems ok, doesn't it?

Thats one FP for every 400 TPs

much better than normal users in their own ways

John Dvorak

this sounds awfully close to the speedup loop

Glorfindel

@Magisch 399 </nitpick>

Magisch

@Glorfindel shakes fist

Cerbrus

@Magisch Waffles

Glorfindel

12:48 PM

yummy

Henders

ymmuy

Magisch

too many carbs

25g carbs a day max for me a waffle exceeds that

Cerbrus

Killjoy -.-

Magisch

@Cerbrus my turn

Tinkeringbell

@Magisch There's special waffles for people like you :)

I found a recipe with one that has only 1,5g carbs a piece ;)

Magisch

12:52 PM

interesting

SmokeDetector

[ SmokeDetector | MS ] Offensive answer detected: What makes substances explosive by user59877 on chemistry.SE

tpu- by Glorfindel

SmokeDetector

1:31 PM

[ SmokeDetector | MS ] Few unique characters in answer, repeating characters in answer: vba Outlook: Sync happens on all the folders even if I choose only one. Why? by Riccardo La Marca on superuser.com

Glorfindel

@SmokeDetector n (bump)

Andy

I have a disappointingly low number of notifications. Anything I should be aware of before I start responding?

Glorfindel

notifications on your meta post?

Andy

Yeah

Glorfindel

Why is that disappointing?

Also, the post hasn't been [featured] yet. Only 267 views so far.

Andy

1:34 PM

I was expecting at least a double digit number

A J

Poor Journeyman Geek. ;)

Glorfindel

Yeah, most of the discussion is taking place on his answer.

Andy

We're not being mean are we? (I'm reading the two new answers since I went to bed)

thesecretmaster

@Undo What do you think of giving mods the power to customize the number of autoflags that get cast on their site, while having a global default?

Glorfindel

@Andy Some of the discussion took place in the Tavern. Nothing special, but you might want to read that as well.

SmokeDetector

1:38 PM

[ SmokeDetector | MS ] Link at end of answer: codeigniter car rental management software by Nepal Car Rental on stackoverflow.com

tpu- by chade_

Glorfindel

@SmokeDetector k

!!/watch nepalrentcar\.com

SmokeDetector

@Glorfindel Added nepalrentcar\.com to watchlist

CI on 902d1ff succeeded. Message contains 'autopull', pulling...

[ SmokeDetector ] SmokeDetector started at rev 902d1ff (SmokeDetector: Auto watch of nepalrentcar\.com by Glorfindel --autopull) (running on teward/Solar Flare)

Restart: API quota is 14196.

Magisch

@Andy The debate is centered on opt in vs opt out at the moment

I don't think we can hash this out to any degree - some people want it to be a strictly opt in system

John Dvorak

In before "why is [minor site] covered in spam?" "well, we wouldn't mind cleaning up there, but they didn't ask for our services"

terdon

@JohnDvorak Are minor sites ever covered in spam? I remember the days before smokey, and spam was only a big issue on the big sites.

But if and when a small site is covered in it, then they can always ask for help, right?

Andy

1:51 PM

@terdon Drupal definitely was.

I don't know it's size off hand though

Mithrandir

@terdon Astronomy.SE also has a big spam problem.

JAD

@terdon Astronomy gets more than it's size would suggest

ArtOfCode

I'd call it moderate

Magisch

@terdon my 600 helpful flags on drupal.SE say yes

terdon

@JAD Um. Please do not confuse those two :)

JAD

1:52 PM

I know

terdon

@Magisch How small is drupal?

Magisch

very

JAD

I know the difference, I didn't remember which was which though :p

John Dvorak

webapps, tex

terdon

There doesn't seem to be a direct correlation oddly enough. For example, Unix & Linux never got much spam, while Ask Ubuntu got metric craptons and they're not thaaat different in size.

Magisch

1:53 PM

@terdon 75k questions, so probably not that small

John Dvorak

webapps is tiny

Magisch

graphic design gets hammered too

terdon

@Magisch Yeah. Doesn't make much sense to spam tiny sites.

Mithrandir

65

Q: 2017: a year in spam

Last year, I posted 2016: a year in spam. Since that was pretty well-received, here’s the 2017 version. Ladies, gentlemen, aliens, and everyone in between: more statistics about spam that you never needed to know. Caveat/disclaimer: this data is from metasmoke, the web dashboard for SmokeDetecto...

JAD

@terdon the big ones get spam, that much is clear. The small ones seems a bit random

Mithrandir

1:53 PM

@terdon Actually, I'd say it does make sense.

Smaller site = stays up for longer = more Google indexing

terdon

@JAD U&L is relatively big and yet gets very little. There must be something else involved.

JAD

hmm

Magisch

@terdon I'm not an expert but from what I know about spam campaigns it probably has to do with keywords / topic categorizations of interest to spammers

John Dvorak

As soon as one piece of spam lingers a bit on a fresh site, I expect heavy hammering as spammers see grazing grounds.

JAD

do we have stats on spam relative to traffic?

terdon

1:55 PM

@Magisch Yes, presumably.

WELZ

@Magisch The Spam Blot room, usually get's about 10-20 posts a day.

Mithrandir

For some reason Puzzling gets a ton of profile spam for sports event and stuff like that.

WELZ

I think the weirdest one is Astronomy...

John Dvorak

and if we need to wait for the small site to ask for help, said grazing grounds will flourish with spammers the more.

Magisch

@terdon point is though sometimes small-ish sites (75k questions is still small if it's been around a long time) get hammered with spam - sometimes the spam stays up for upwards of half an hour then.

terdon

1:56 PM

@Magisch "Upwards of half an hour"? Heh, you say that as though it were days!

JAD

eons!

terdon

But sure, in those cases the community could ask for help, what's the problem?

Magisch

@terdon as far as spam is concerned, thats unacceptable though

Mithrandir

I've seen spam last for 6 hours occasionally

WELZ

We only have 24k questions on GDSE

Mithrandir

1:57 PM

on smaller sites

Magisch

The goal to disincentivize spam is to minimize views & uptime and prevent indexing by google

terdon

@Magisch Not really. But I know you have strong views on the subject.

Magisch

If spammers can get their spam indexed they've won. And that'll encourage more

WELZ

True.

ColleenV

So I'm seeing posts on the ELL dashboard as "undeleted" that were in fact deleted. How is the deletion detected? Does "undeleted" mean "not deleted" or am I misunderstanding the label? Example: metasmoke.erwaysoftware.com/post/108828

Mithrandir

1:58 PM

@WELZ and another ~5k deleted spam posts, at least in MS

John Dvorak

@terdon how long do you think it will take them to ask? Even if they know of Charcoal, if the mods are asleep, 8 hours+ of the front page covered with spam.

thesecretmaster

@ColleenV Some of the deletion data is broken.

John Dvorak

If they don't know of Charcoal, well... GG

ArtOfCode

@thesecretmaster eh, not broken per se

terdon

@JohnDvorak That seems exceedingly unlikely. Never happened before in the days before smokey either.

WELZ

1:58 PM

@Mithrandir that's a nearly 17% spam rate :O

terdon

And it's also a doomsday scenario.

ColleenV

@thesecretmaster Is there something we should do differently to help with the data quality?

John Dvorak

@terdon Webapps took a MSE post to notice

Magisch

@terdon We had some incidents with smokey before autoflagging where it was an off hour and the front page of a site was polluted quite a while

ColleenV

Or is it just a "that's reality" problem?

terdon

1:59 PM

@JohnDvorak And did they have their front page covered with spam for days?

ArtOfCode

@ColleenV We stop watching for deletion after a while. In other words, not every deleted post gets recorded as such in metasmoke, but those that do get recorded are accurate.

Magisch

Earlier in the project, when we had fewer people

terdon

@Magisch I'm an AU mod. Believe me, I know and remember.

Transcript for