« first day (1277 days earlier)      last day (2560 days later) » 

3:06 PM
I wonder if autoflags should be R/A instead of spam. I mean, no one is ever going to dispute a spam post being "abusive," but people might dispute an offensive post being spam
 
do we have any offensive posts triggering autoflags?
 
@quartata ...I think I said that recently
 
I believe so
Like "offensive title, offensive body" or "offensive body, blacklisted user"
@Mithrandir Oh, haha sorry
 
@tripleee applies to fire flags as well
 
May 8 at 14:40, by Mithrandir
Wouldn't it make more sense to always flag as R/A?
 
3:07 PM
@quartata I went through the three Offensive xx reasons in Metasmoke with Autoflagged=Yes and none gave any results
for example
sorry, meh, I had forgotten to zap the previous search, my bad
 
Also gibberish posts get autoflagged often
 
at 156 and 220
though both are below the recommended autoflagging threshold
maybe we should simply have a different threshold for abusive?
 
do we really want those to be autoflagged?
 
Those? absolutely
look at them
 
3:09 PM
24 autoflagged with Offensive answer, top score 252 metasmoke.erwaysoftware.com/…
 
There are people who will dispute a spam post being abusive, too
 
CI on faba899 succeeded.
 
@quartata they should be flagged alright but should they be automatically flagged? I would be very hesitant
 
@ArtOfCode fewer
 
actually about the same
 
3:10 PM
@ArtOfCode Really?
 
yup
 
Well then.
People are picky huh
:/
 
That's when we pick a volunteer to come and complain on meta
 
we already did that
 
CI on ae7e352 succeeded.
 
3:13 PM
@quartata it's because mods are picky about all the other flag types, so unless someone tells them otherwise, they're picky about spam vs abuse too
 
the focus here is on spam anyway -- if we want a change, my proposal would be to exclude abusive from autoflagging
 
in that case I want Fire to support R/A flags
 
that requires MS changes anyways though
 
which is currently blocked by Metasmoke refusing to cooperate
 
Good.
 
3:14 PM
It's just we don't have a route for it
 
The goal of this project is SPAM detection
Finding abusive posts is a side effect of that because those types of posts trigger common spam patterns
 
The room description says "smoke", not "spam"
 
that's just to fit with the charcoal theme
 
I think abusive posts are our purview
 
If r/a was just a nice side effect, should I nuke gasmask?
 
3:15 PM
> Headless chatbot that detects spam and posts links to it to chatrooms for quick deletion.
 
Not that anybody still uses it anymore anyways.
 
They're very similar to spam in that they require instant nuking and that they're very easy to detect with patterns
 
@JanDvorak no it's not, that could be done with the SE API
 
> Charcoal are the people behind SmokeDetector, a community bot to detect spam on the Stack Exchange network.
 
@ArtOfCode but it's easier if MS cooperates
 
3:16 PM
We can argue semantics all we like but essentially abusive posts are spam and spam posts are abusive
so this discussion is pointless
 
@JanDvorak actually I do
 
@JanDvorak not much, actually
 
Shog himself clarified that the two flag types are only there because users have gotten accustomed to the UI and that the 2 flags are treated equivalently by the backend
 
@Magisch No, spam is spam. It is distinctly different than abuse. It serves a purpose - one we don't support - but generally doesn't target a single person. It is trying to make some money
 
@Magisch I think it's a problem that people are running with autoflagging thresholds set so low that abusive posts get incorrectly flagged as spam, it could affect our credibility
 
3:18 PM
@Andy Spam is defined as any unwanted message transmitted digitally
And even that definition is quite narrow
 
@Magisch no, it's not
 
not on SE
 
If we're talking SE, then this point is moot because on SE spam and abusive are the same
 
@Magisch Here's my suggestion: Go post on Meta stating that you are going to start flagging all spam as abusive and that everyone else should too. Link to Shog's message about clarification. Let's see how that works and how the community responds.
 
and the only reason 2 flags exist is because of legacy UI design
 
3:20 PM
@Andy on every meta?
 
@Andy should I buy some popcorn?
 
traditionally email spam was carefully defined as unsolicited bulk email (where some people tried to go for unsolicited commercial email, but that would exclude e.g. religious and political messaging) ... the bulk nature and the unsolicited restriction are significant
if you receive a message from your mother that you did not want, it's not spam
 
@Mithrandir Of course not. Go put it on MSE
 
less traditionally, operators had to budge when people kept on clicking spam on e.g. newsletter they no longer wanted, so the operational definition is less strict these days
but if you want to have a definition, it needs to be stricter than what you were trying
 
@Magisch If you want that post to be more constructive, I recommend asking if the spam/abuse distinction is utilized by SE anywhere. Is it used for any type of analysis or any planned analysis in the short term (slight sarcasm thrown in due to SE's "Ability" to estimate development time)
 
3:26 PM
Although in all seriousness
 
One thing to really consider with this recommendation: Getting users to think of spam as "abusive" is going to take time. A lot of time. The youngest users here are 13 the oldest are multiples of that. Almost everyone knows what "spam" means. Not everyone understands what "abusive" means. Some think swearing is abuse. Others disagree that cat on keyboard is not abuse.
 
we already did this on MSO and it turned out fine
 
I don't think there's even a need for the users to know this
Shog's post is out there, and it serves as guidance for mods
So we can just use one flag and point to that post when a mod declines it
Not that we need to care one bit, they can keep their 2 historical flag types
 
so really what we need is a pinned TL message
 
@Magisch Bullshit. We posted on MSE that was featured network wide for two weeks that explained that we flag spam
That post has been viewed 13K+ times.
Users know what SPAM is
 
3:28 PM
@Magisch getting a lot of declines would be suboptimal
 
They don't know what "abusive" means.
 
viewed mostly by civilians, presumably
 
@tripleee red flags can be changed to 'disputed' after either outcome, pretty sure
 
As an example...go get a cat, put it on your keyboard, and post it's poetry. You'll generate spam flags, abuse flags, NAAs, VLQs, and probably a custom flag too.
 
that's abusive unless you get the consent of the cat
 
3:32 PM
@Andy this isn't a good thing
although it's not really our problem
 
@quartata I agree, but there is a huge difference between "lksadghfoi7asedr09v 6 kjldf" and "CHINA UNIVERSITY FREE DEGREE FOR $3.50" in terms of what users see. One is clearly spam. The other is not welcome, for sure. But, do they know to flag it as "abuse"? If anything, I'd flag that first one as spam rather than flagging the second one as abuse
 
It's more important if mods know to handle the flags
 
If you get a declined flag raise it on the meta where it was declined, and the mod learns. We were considering having templates for stuff like that even/
 
Mods should know how to handle it, but the community issues and deals with the majority of spam. They should know how to handle the above two post types
 
3:47 PM
[ SmokeDetector | MS ] Bad keyword in answer, bad keyword in link text in answer, username similar to website in answer: Pitching Three Startups by Secure Web Technologies on startups.SE
 
sd k
 
4:22 PM
[ SmokeDetector | MS ] Bad keyword in answer, offensive answer detected: Why the units of surface tension are force per unit length? by user156424 on physics.SE
tpu- by Cai
 
4:57 PM
Neat! The SO app question is now the 4th most downvoted on Meta
 
What is it good for?
 
Oh wait. Just noticed this:
@ShadowWizard I'm not going to lie and say our Android app has gotten the love and attention it needs in the last year. But, this app is built in a shared codebase with the SE app (albeit UI changes still need to be ported) -- the biggest changes made to the SE android app have been secret early access programs which I can't wait to talk about after I/O :) — Kasra Rahjerdi ♦ yesterday
after I/O would be Google's I/O which is today
Sounds like some new features they can't talk about?
@JanDvorak I dunno
 
[ SmokeDetector | MS ] Bad keyword in answer, bad keyword in link text in answer, blacklisted website in answer, pattern-matching website in answer: Schema First WCF Development by neiljakson on stackoverflow.com
tpu- by CalvT
 
[ SmokeDetector | MS ] Manually reported answer: Could I save printer's ink/toner by changing the font? by Cade on superuser.com
 
5:07 PM
sd k
 
[ SmokeDetector | MS ] Few unique characters in answer: What is our skin made up of? by user32477 on biology.SE
 
sd k
 
[ SmokeDetector | MS ] Repeating words in body: How to export c# sql excel? by Sviatko Sandor on stackoverflow.com
 
sd why
 
@CalvT [:37466049] Post - Repeated word: marpicsaba
 
5:16 PM
sd - - f
 
Conflicting feedback on How to export c# sql excel?.
 
5:55 PM
[ SmokeDetector | MS ] URL in title, bad keyword in body, bad keyword in title, link at end of body, pattern-matching website in body, +2 more: worldmuscleking.com/phallyx/ by Extrelun on drupal.SE
 
sd k
 
@SmokeDetector K
 
@SmokeDetector N
 
sd n
 
6:17 PM
[ SmokeDetector | MS ] Blacklisted user: Capitalize! Statement by Riyas Randhava on codegolf.SE (@quartata)
 
^ seems legit
 
sd fpu-
Was blacklisted for a garbage post
 
@CalvT you don't need the u on the end of that
fpu- is fp and whitelist user, not remove from blacklist
just fp- will remove from blacklist
(it doesn't actually do any harm in this case because it's only the username that's whitelisted, not post content, but just something to be aware of)
 
Should this system be modified to be more intuitive?
 
@ArtOfCode ah ok fe, I misunderstood
 
6:24 PM
define intuitive?
 
@CalvT User removed from whitelist (68954 on codegolf.stackexchange.com).
 
por las dudas
 
@ArtOfCode something such that you don't have to repeat the rules because people thought the rules were different
 
 
@CalvT User removed from whitelist (7404680 on stackoverflow.com).
 
I did it earlier too
 
@JanDvorak okay, so what should it be instead?
 
Pretty clear to me - I didn't read it properly before
> Marks a reported post as false positive. Additionally removes the user from the blacklist, if that was the reason that the post was reported.
And fpu
> Marks a reported post as false positive and adds the poster to the whitelist.
 
The whitelist is such a rarely used feature. Have u toggle the blacklist while fp doesn't and make whitelist only accessible via *wlu
 
6:31 PM
[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title, link at end of body, pattern-matching website in body, repeated URL at end of long post: Phallyx It will stop the debilitating procedure by Evesce1978 on superuser.com
 
sd k
 
[ SmokeDetector | MS ] Bad keyword with email in answer: Kazoo installation from source code by Cloud Pacific IT Solutions on stackoverflow.com
 
waffles
sd k
 
Waffles indeed
 
@JanDvorak Why not get rid of u on fp responses entirely? An fp response removes from the blacklist. Whitelist is only accessible via *wlu.
 
6:34 PM
what if I want to fp without unblacklisting?
 
@ArtOfCode then you'll still get people using fpu as it follows the tpu format
 
@JanDvorak why would you want that
 
@ArtOfCode their last post seemed fine, but I still suspect they might fall to the dark side again soon
 
then that should be a rare enough case that it can be done with fp and addblu - in keeping whitelists being rare and shifted off to addwlu
 
Worth trying out I guess.
Should fpu trigger a deprecation warning?
 
6:37 PM
I'd just remove it and let it respond with the usual "wth are you talking about human"
I don't think we actually have a mechanism for deprecation warnings
 
deprecation warning = direct reply without further action
 
dat shouting. I'm deaf now.
 
sd f
 
@JanDvorak WHAAAAAT? CAN'T HEAR YOU
 
 
1 hour later…
8:33 PM
Note to self: when rage-quitting Charcoal, be sure to make library, name it after one of the bigger pharma spammer products and then popularize the library.
 
Restart: API quota is 7858.
 
Hello!
 
 
1 hour later…
10:10 PM
[ SmokeDetector | MS ] Pattern-matching website in answer: On new iPhone, when I call people no one can hear me but I can hear them by Jonathan Hawkins on apple.SE (@CalvT)
tpu- by DavidPostill
 
10:49 PM
[ SmokeDetector | MS ] Blacklisted website in answer: How to use iOS 10.3 devices with Xcode 8.2.1 by Michael Revlis on stackoverflow.com
fp- by NobodyNada
 
...I think
 
^ Checking website history on MS
 
sd - f - k - k - - - f f - k f
4
Hmm imobie.com has 5 tps but 6 fps
 
@CalvT Yeah
Better unblacklist it
 
[ SmokeDetector | MS ] Blacklisted username, offensive answer detected: Scientific findings of strobe lights and the brain by GAY nIGGA on cogsci.SE
tpu- by NobodyNada
 
10:55 PM
sd - k
Ok PR #761 submitted to remove imobie.com from blacklist
@SmokeDetector f
^ has been member for over a year, 2 badges, although only post
 
11:30 PM
API quota rolled over with 6485 requests remaining. Current quota: 19999.
math: 281
stackoverflow: 240
codegolf: 69
superuser: 62
physics: 51
serverfault: 48
apple: 33
english: 31
unix: 24
salesforce: 22
askubuntu: 22
security: 22
es.stackoverflow: 19
electronics: 15
dba: 14
scifi: 13
meta: 13
wordpress: 12
drupal: 11
travel: 10
arduino: 10
gamedev: 10
gaming: 9
crypto: 9
android: 8
mathoverflow.net: 8
sharepoint: 8
bitcoin: 8
rpg: 8
mathematica: 8
magento: 8
history: 8
engineering: 8
graphicdesign: 8
webapps: 7
softwareengineering: 7
gis: 7
webmasters: 7
worldbuilding: 6
 
11:43 PM
[ SmokeDetector | MS ] Offensive body detected: looking for 16 volt 20 amp diode by frank 85 on electronics.SE
CI on f690fc9 succeeded.
 
J F
sd - k
 

« first day (1277 days earlier)      last day (2560 days later) »