« first day (1459 days earlier)      last day (2354 days later) » 

4:12 PM
@SmokeDetector why
 
@JakeSymons That's not a report.
 
@SmokeDetector why
 
@JakeSymons That's not a report.
 
@SmokeDetector why
 
@JakeSymons Body - Position 875-893: tcl.tk, Position 946-964: tcl.tk
 
4:13 PM
!!/test tcl.tk
 
> Would not be caught for title, body, and username.
 
...huh
all .tk sites are supposed to be pattern-matched
 
Evidently they are because it was caught
 
but why didn't the !!/test work then?
also hi
 
!!/test somesiteontk.tk
 
4:14 PM
> Would not be caught for title, body, and username.
 
haven't seen you in a while
 
o/
 
Yeah things have been pretty busy and I measured my time spent on here during work hours and well...
 
[ SmokeDetector | MS ] Link at end of answer: why we need VLAN in any network? by SteveH on serverfault.com
 
4:15 PM
Lets say it was a lot more than I realised
 
naa- by CalvT븃
 
o_o
 
!!/test tcl.tk
 
> Url in title, pattern-matching website in title, pattern-matching website in username
----------
Title - Position 1-19: http://www.tcl.tk/
Title - Position 1-18: http://www.tcl.tk
Username - Position 1-19: http://www.tcl.tk/
 
!!/test www.tcl.tk
 
4:17 PM
> Would not be caught for title, body, and username.
 
Ok, so .tk domains need the full http:// etc to be caught
 
huh.
that may need to be fixed
 
@Undo or we could just use GH pages :P
 
!!/test tcl.tk
 
> Url in title
----------
Title - Position 1-14: http://tcl.tk
 
4:18 PM
no sense in wasting compute time for something that doesn't need a server
 
 {'regex': r'(?is)\w{3}\.tk(?:</strong>)?\W*</p>\s*$', 'all': True,
         'sites': [], 'reason': 'pattern-matching website in {}', 'title': False, 'body': True, 'username': False,
         'stripcodeblocks': False, 'body_summary': False, 'questions': False, 'max_rep': 1, 'max_score': 0},
 
Revision - need full http://www. to be caught
 
@angussidney woot we're up to 54%
thanks
 
4:32 PM
Hey, @CalvT븃! How are you doing? :)
 
@Henders Hey :)
Not too bad thanks
 
Awesome! Good to see you around
 
Just got back home after a 6 week stint in UK
 
Great country, right? ;)
I hope you enjoyed the rain :)
 
Well it's ** cold, but being english myself, the rain doesn't bother me
But at 6am last Wednesday it was -2C, Friday I was back home enjoying 32C :D
 
4:37 PM
Woah! That's not bad :P
 
4:47 PM
[ SmokeDetector | MS ] Offensive answer detected: Installing can I install gstreamer 1.4 on ubuntu 14.04? by mevsme on askubuntu.com
tpu- by CalvT븃
 
@CalvT븃 not generally, it's jut that the .tk regex is special
I tried to work on a fix for this recurring FP but it's not working yet
oh I see you discussed that already, nvm
 
[ SmokeDetector | MS ] URL in title, bad keyword in body, bad keyword in title, blacklisted website in body, blacklisted website in title, +2 more: allsupplement4u.com/vars-performance-us/ by Wilberweisz on graphicdesign.SE
tpu- by CalvT븃
 
@SmokeDetector why
 
Title - Position 28-44: vars-performance
Title - Position 8-27: allsupplement4u.com
Title - Position 11-26: supplement4u.co
Title - Position 1-27: http://allsupplement4u.com
Body - Position 185-192: jelqing, Position 1002-1018: vars-performance, Position 1077-1093: vars-performance
Body - Position 138-143: penis, Position 163-168: penis, Position 225-230: penis, Position 581-586: penis
Body - Position 982-1001: allsupplement4u.com, Position 1057-1076: allsupplement4u.com
Body - Position 985-1000: supplement4u.co
 
@SmokeDetector that's a bad post
 
4:57 PM
@JakeSymons Allow me to point you to the MS link in each of these reports. Clicking that will provide you will the full why information, so that you don't need to ask why for every report
 
@JakeSymons those are obvious spam just flag them as such
no need to do a 'why' on those, those supplement site links and stuff are evil
also 100% GUARANTEED spam
trust me i squish about 20 of these a week :P
 
and flags squish a whole lot more :P
 
^ this
 
you'll notice there's a fair amount of overlap, the reasons find that the link is blacklisted in the title and the same link is blacklisted in the body and so forth so the same piece of text is detected in different contexts, bringing up the score
 
^ this
 
5:12 PM
v this
 
hullo
 
^ this
goes back to doing something useful...
 
not this >
 
Is there any more bots? Like one that flags for duplicates or things that Smokey doesn't deal with
 

 SOBotics

This room is dedicated to moderating content on Stack Overflow...
 
5:20 PM
^ a large number, there
some communities doubtlessly have their own little private bots, too
 
I need 20 rep on Stack Overflow to post :(
 
(to chat, you mean?)
 
@tripleee yeah
 
try sobotics.org for a start, they probably have links to archived chat sessions too
some cool stuff in the Twitter feed on that page
(maybe you need privileges for that, too?)
 
@tripleee I can open it
One of my flags on meta stackexchange got rejected :(
 
5:25 PM
yeah, you want to be careful with flags especially when you are struggling with a low rep
 
@tripleee yeah I thought it was alright.
flagging it for not an anwser
is there a way of disputing a flag that's disputed
 
@JakeSymons The only way is to write a post alerting a mod about it but a) I wouldn't recommend it and b) it'll probably get downvoted meaning you'll lose the little rep you have
What answer was it anyway?
 
@CalvT븃 yeah, i'll just leave it
 
@JakeSymons Which answer? Link?
 
-1
A: Synonym / Retagging request: [mathjax] and [latex]

200_successI suggest that latex be made a synonym, with mathjax as the master.

It put it didn't answer the question. I was probably wrong though
 
5:32 PM
It does give a suggestion for an answer though
 
@CalvT븃 yeah the flag was probably wrong
 
Meta sites work a bit differently to main sites - often the answers will be suggestions, then the one with the most votes gets implemented
 
@JakeSymons Answers on meta are different to answers on the main sites (as is voting). People can propose possible solutions to proposals and then people vote on the answers to agree or disagree. This applies particularly to tag suggestions.
 
And I quote from that:
> Unlike normal Stack Exchange sites, Meta invites the community to discuss, debate and propose changes to the way the community itself behaves, as well as how the software itself works. On posts tagged feature-request, voting indicates agreement or disagreement with the proposed change rather than just the quality or usefulness of the post itself.
 
@CalvT븃 so the flag was wrong
oh, its only one wrong, I've got four right so far
 
5:36 PM
@JakeSymons Correct. You need to learn more about how individual sites works before you start flagging.
 
@JakeSymons That's correct yes. Just a little tip (which I had to learn once I started helping here) is to stay away from flagging until you understand how a site works. If it's rude or blatant spam, that's one thing, but NAA and VLQ and the like - be careful.
 
@CalvT븃 so I gotta wait for Smokey to pick some spam up to flag
 
Or go looking for them - simply participating in a site is a surprisingly easy way to find things. It's also not that hard to gain rep, especially on the smaller sites, and once you have a certain amount of rep you'll probably know how the site works
 
!!/test www.goggle.es
 
> Would not be caught for title, body, and username.
 
5:51 PM
@JakeSymons Please stop abusing the bot.
 
6:04 PM
0
Q: Suggestions to replace 50rep-barrier

FabianNew users have to get 50 reputation points to be able to comment. I disagree with this design choice, because: If you are mostly just reading instead of writing, it takes ages to get 50 reputation, even if you are a normal, non-spammy person. If you find an answer that you can't execute because...

 
6:14 PM
@ProcessedMeat hi
I haven't seen you before
Why is your logo different
 
@JakeSymons it's another bot
or rather, I think it's part of the chat infrastructure
 
@tripleee what does it do and why has it made a post
 
it posts a notice here when something (on meta only I think) is tagged "spam"
it's configured via the room's feeds, every chat room can have these
 
yeah it's an RSS feed
 
The site it posts from depends on what site the chat room is tied to I believe? As a general rule
 
6:18 PM
@CalvT븃 actually you can set its avatar to anything
 
@CalvT븃 it's all configurable.
 
you can create a feed for anything with RSS, some have XKCD in their feed
 
Is that post spam
 
@JakeSymons no, it's (vaguely) about spam
 
[ SmokeDetector | MS ] Bad NS for domain in body: gmap3 infowindow background coustomization by Muzahid on stackoverflow.com
[ SmokeDetector | MS ] Bad keyword with email in answer, email in answer: Does opening credit cards at different banks impact credit score? by victor on money.SE
 
6:20 PM
@SmokeDetector f
 
tpu- by Henders
 
@tripleee Looks a bit spammy? o.O
 
@Henders I think that's the site he's working on
 
@CalvT븃 nope
 
Sorry, the first one is a link to an image...
 
6:22 PM
Also, technically, it's a feed, not a bot
 
The title of the site is gmap3, and the title of the question is gmap3 infowindow background coustomization
 
I voted to close as unclear but I guess he is trying to ask about the mapping service he is linking to
 
is this the list of allowed websites?
that won't flag as spam
 
no, it's just the ones we tagged as "not usually or exclusively spam sites" to keep them sorted separately, the entire spam domains section is just collecting an archive
it's a fairly new feature so we are mainly just collecting data at this point
though you might find something worth analyzing in e.g. the #drugs tag
but the tag is just an annotation, it doesn't do anything ... at the moment
 
@tripleee so if someone entered one of them would it come up on here if there was nothing else in the post
 
6:31 PM
like I said, it doesn't affect anything, it's just a way for us to organize the ones we have already collected via the Metasmoke spam archive
once you have a Metasmoke account you should be able to add tags as well I believe
like if we had a false positive on a popular domain which you recognize, you could go ahead and add that to the #not_spam set
 
@tripleee I've got an account so I will have a look and see if I can get on it
 
Facebook and Twitter are already there but we haven't seen any FPs on Baidu.com or vKontakte.ru -- if we ever get one, those would be the kind which we would add to this set
 
@tripleee I've got the list of tags up
 
some domain which is used a lot in examples on Stack Overflow would be good to add as well
tcl.tk for one!
go and tag that
 
@tripleee how do I add tags? Do I need privalages
 
6:34 PM
I guess you don't but I'm not sure
if you click on www.tcl.tk at the bottom of that page, you should see a list of the hits and a box where you can enter tags
I guess you see the "no tags <text box> [add]" as well?
 
yeah
 
so you'd type not_spam in the text box and click Add
 
how do you shreenshot
 
on my Mac, the secret keystroke is cmd-shift-4 <click> <space>
it saves the screenshot on the desktop
on Linux it depends on your distro but IIRC alt-prtSc on some Gnome
 
Yeah I've done it but that comes up; no privileges :(
On my account it says my email is 12007967@se-oauth.metasmoke should I change it to my actual email
 
6:39 PM
[ SmokeDetector | MS ] Link at end of answer: Removing a line from a PDF document by jay jay on superuser.com
 
too bad, really not sure which privilege this is associated with
@JakeSymons I don't think that matters
 
> You need to have the Core role to view this page
 
@JakeSymons No - it is tagged spam.
 
bah, added the tag myself
 
naa- by Glorfindel
tpu- by Glorfindel
 
6:41 PM
!!/watch eudormail\.com
 
@Glorfindel Added eudormail\.com to watchlist
 
you guys can do all this cool stuff
I can just be like -why
and that's it
 
patience, young padawan
 
@SmokeDetector lol Redirects to a bible study site ;)
 
if you're any good with computers, you know, you could do all kinds of off-line analysis with the data we have collected
 
6:43 PM
@DavidPostill now that you're around, could you please check the SU moderator archives for a potential spammer for Copytrans Contacts?
 
CI on e917876 succeeded. Message contains 'autopull', pulling...
Restart: API quota is 11771.
 
@Glorfindel Wait a minute ...
 
Yep, core for spam domains and domain tags.
 
sure, no hurry.
 
gargle, I can't for the life of me repro the tcl.tk bug, it should be easy to fix but I wanted to create a test case and I just ... can't
 
6:44 PM
He's like Neo before the kung fu training. You telling me I can dodge bullets?
 
the test suite runs on questions but changing it so it thinks all the test cases are answers doesn't help
 
@JakeSymons 12007967@se-oauth.metasmoke is the fake email that metasmoke created for you since you signed up with SE OAuth. You can change it if you want, but it's not necessary.
 
@ArtOfCode okay i'll leave it then
So if I emailed it nothing would happen
 
@JakeSymons .metasmoke is not a valid TLD so you would very quickly receive a non-delivery report
 
6:47 PM
@tripleee just come through
 
@Glorfindel There are some very old posts mentioning Copytrans, nothing in the last 3 years ...
 
#mailer-daemon
 
@Glorfindel Do you have a particular reason for asking?
 
@DavidPostill thx. I've got a flag accusing a user of being a spammer of that product, but they haven't been seen for a while.
 
PR#1230 tcl.tk FP fix
 
6:50 PM
^ that user.
 
I was just going to click on that
 
@JakeSymons sorry, I forgot it oneboxed.
 
Last seen 2014 on SU
 
Reminder that this isn't TL, please keep private stuff there, including suspension information
 
Thanks. I'll take no action, then.
 
6:53 PM
[ SmokeDetector | MS ] Offensive answer detected: Count matches at the end of a string by John Eickson on stackoverflow.com
tpu- by Glorfindel
 
@Mithrandir I didn't mention any user names ...
 
@DavidPostill it was linked earlier
 
@SmokeDetector why
 
@JakeSymons Post - Offensive keyword: fuck
 
why didn't it get the other words
 
6:54 PM
And people can see the history of deleted messages.
Maybe only ROs, but that still includes some non mods.
 
@JakeSymons You have been asked to stop using "why"
@Mithrandir OK
 
reminder, clicking the [MS] part of the message takes you to metasmoke.erwaysoftware.com/post/93670 where you can review the Metasmoke report in full
 
@Mithrandir Just ROs, but yeah, I'm a RO and a non-mod
 
me too
 
@JakeSymons The other words aren't in the offensive keywords list
 
6:56 PM
@Mithrandir ah, sorry, forgot about that.
 
Maybe too many FPs, maybe nobody thought to add them
 
@NobodyNada should they be??
 
@Glorfindel Ask in TL next time :)
 
if you find that we miss some favorite words of yours, and that this actually causes us to miss offensive messages, maybe propose adding them
we don't particularly strive to catch offensive messages, though -- there are other bots for that too
 
@tripleee yes we do
 
6:57 PM
@tripleee yeah maybe we could add that one to the list
 
The only other bot that I'm aware of is HeatDetector on SOBotics, and that only searches comments
 
@JakeSymons the cat one?
 
as in, we don't prioritize work on that but if you create a PR I'm sure we'll be happy to take it
 
@NobodyNada and only on SO
 
Indeed
 
6:58 PM
@Mithrandir yeah
 
And SO has 62,000,000 comments
 
mm I guess I was thinking of HeatDetector
 
@Mithrandir wow that a lot of comments!
 
Yup
Only two other sites have even hit the 1,000,000 mark
 
@Mithrandir wow
 
7:03 PM
@Mithrandir hmm...SU and SF?
 
SU is one of them.
(with 1,300,000 comments)
 
Oh, is AU the other?
 
Wait, did I write a billion instead of a million? Whoops :P
No, AU has 945k
 
Oh, that seems a bit more reasonable :p
@Mithrandir :/ it's not MSE is it?
 
Nope
(if you mean Meta.SE)
647k for Meta.SE
 
7:10 PM
Arqade?
 
Math SE
at least that's my best guess
 
@NobodyNada 288k
@quartata ding ding ding
 
ha
 
Whoa, really?
 
people underestimate its size
it's actually got the second most number of questions IIRC
 
7:11 PM
@quartata Indeed we do :)
 
All those numbers came from Data.SE
 
I figured; SQL's pretty powerful
Also, did somebody fix the bug where Smokey would shut down after a couple minutes if it doesn't have a MS token?
 
doesn't ring any bell, when was this?
 
I remember it happening around May or so, but it was around for a long time before that
 
7:15 PM
@NobodyNada it's right there on the front page
 
@tripleee I don't suppose there's any easy way you could send something down the Halflife websocket is there?
too impatient to wait for a post :P
 
not impossible I suppose but why don't you set it up to listen to a recorded transcript?
by the by I changed the format slightly earlier today, make sure you git pull
 
oh that's not what I meant
I mean I have something that's listening to your instance of Halflife
I want to make sure the websocket bits work
 
yes, I understand, but can you make it listen to your own recording instead?
set up a websocket server with a few messages, have it listen to that
 
yeah I suppose I could
oop question answered
thank you
 
7:24 PM
typos and all! /-:
 
@DavidPostill do they have an ask a moderator for every site or is it just superuser and a few others. An are you the only mod that helps out in ask a superuser mod
 
@tripleee actually it's weird it looks like whenever I connect it sends the past x messages or so anyways
unless chrome somehow caches that
turns out by the way I had the html with the messages set to display: none :P
 
I think that's a feature of the websocket protocol, or possibly the websocketd server I use
 
well it's very useful for testing
 
@tripleee I always have typos, like all the time.
 
7:26 PM
in Charcoal Test, 2 mins ago, by PulseMonitor
Teset for quartata
I find that hard to compete with
 
@tripleee do posts that come to check chat.stackexchange.com/transcript/message/41170432#41170432 also come up on here??
 
@JakeSymons no, it's a separate room
once Halflife is stable and reliable some of that might spill over to this side somehow
 
@tripleee so are the posts on there flagged. When it says to Check A post is it because its bad
 
though I guess if we are confident to automate it, it might perhaps talk directly to Helios or something
@JakeSymons it picks up the same posts you see here, it listens to Smoke
so it's the other way around, more or less
though it's not physically listening to this channel, there's a websocket behind the scenes
 
@tripleee oh causae I was going to have that open as well so I could see some extra pots but if it all comes up on here then I can just have it all through here
 
7:29 PM
where Smokey reports its findings in friendly JSON instead of human-readable
 
@tripleee okay
 
@JakeSymons Halflife does perform some follow-up analysis, the purpose of that room is to see the results of that
so it follows redirects, checks the Metasmoke results for the URLs, etc
 
Quick Question: Before I report a bug on the StackExchange website have they stopped the drop down menu that shows you the sites. When I click on the button in the top left corner I am redirected to the homepage. Is this normal.
 
I used to do that by hand for every post during the day but I wanted to automate it
 
@JakeSymons Yep, they did that on purpose during the top bar redesign IIRC
 
7:31 PM
@JakeSymons I think they changed that recently, yeah
 
It worked earlier
today
 
I think the site switcher is on the right
 
stackoverflow.blog/2017/02/14/… explains their design goals and the rollout strategy
I guess you got rolled out during today
oh wait that's the February redesign
 
so they changed that in like a day.
I need to get on the blog more
 
@quartata It seems the NG chat system broke !!/test and !!/errorlogs?
 
7:45 PM
quick and dirty nameserver investigation gist.github.com/tripleee/2da65c11aaa274c7a992056865040f80
I think we have some support for blacklisting Reza+Lloyd and maybe Apollo+Liz and Lola+Chip but I ran out of API quota while testing so I will continue scanning new posts tomorrow
I have a vague hypothesis which might allow us to block part of Domaincontrol without incurring too many false positives
I'm speculating that these providers actually know full well the identity and purpose of many of these spam domains, and try to put them all in one place where they are easy to watch and easy to cut off if you want to
though I'd be surprised if Domaincontrol turned out to be quite so cooperative as to put all the bad eggs on a series of consecutively numbered name servers
 
@SmokeDetector why
 
@DrMcMoylex Body - Position 117-127: Beast Mode, Position 211-221: Beast Mode, Position 294-304: Beast Mode, Position 410-420: Beast Mode
 
@SmokeDetector f
 
whats wrong with the words beast mode
 
7:55 PM
@SmokeDetector ignore
 
@tripleee Post ignored; alerts about it will no longer be posted.
 
@JakeSymons search and ye shall find out
 
@tripleee I just got some amarican footballer
 
see the Metasmoke search link I just posted
 
@tripleee that might be a detection we have to have ignored for a specific site, last pos detection was 3 months ago
the rest were FPs from Anime.
 
7:57 PM
@tripleee oh, I get it now
 
yeah, I was thinking maybe we could retire it at least for now, or use the full phrase "beast mode boost"
the oldest hits are spurious anyway
it was just the one campaign three months ago
but seeing how they work in cycles I'd be hesitant to remove it completely
 
@tripleee maybe demote to watchlist?
 
^ this is one of the things we should have a better understanding of, anybody up for some analysis?
 

« first day (1459 days earlier)      last day (2354 days later) »