« first day (1073 days earlier)      last day (3044 days later) » 

13:00
@Glorfindel Well, we hypothesized stuff not about the name, but about the water drop.
@SmokeDetector Not sure about this one
@DavidPostill sounds effy
I Flagged it. For me the username was the final nail in the coffin
But it could be a false positive. I'm not 100% sure either way
@DJMcMayhem Eh? What about the username is iffy?
13:06
It's the same as the site they're linking
I didn't visit the site
@DJMcMayhem I think you're referring to a different report ...
21 secs ago, by DavidPostill
@DJMcMayhem I think you're referring to a different report ...
Whoops, you're right
12 mins ago, by SmokeDetector
[ SmokeDetector | MS ] Bad keyword in body, blacklisted website in body: Control Dual Sim Mobile Data by Burks.bro on android.stackexchange.com
Facepalm 'effing appreciated'
13:09
@DJMcMayhem
@SmokeDetector f
Yeah, I thought you meant this one:
[ SmokeDetector | MS ] Bad keyword in title, pattern-matching website in body: Do Beard Supplements Like VitaBeard & Beardilizer Work? by Thaut1943 on workplace.stackexchange.com
tpu- by Jan Dvorak
@SmokeDetector k
BEARD SUPPLEMENTS
13:12
Can we get a room where only Smokey is allowed to talk?
(╯°□°)╯︵ ┻━┻
Y U NO LIEK MY FACE JAN
Your report is redundant and distracts me from watching videogames
@JanDvorak That Smokey school or whatever
Hello @JohnB
hey Magisch
13:16
@Magisch what d'you mean?
@ArtOfCode SE is talking integration
That integration will most likely not happen with us
But they'll just take the things of the bot they can use themselfes effectively
And that'll be it
I'm... not getting you. If they're integrating Smokey, they kinda have to include us.
I think what Magisch is saying is they're going to 'nitpick' features
and not fully integrate, ArtOfCode
[ SmokeDetector | MS ] Manually reported answer: Who copies app.config to app.exe.config? by 536546 on stackoverflow.com
(which has happened with other open source things I've had on the 'net with other 'companies' or entities)
13:18
tpu- by Jan Dvorak
SD K
@M.A.R. [:33117632] That message is not a report.
@ArtOfCode Not really
@SmokeDetector K
The part of smokey they'll be interested in will probably be the most successful filters / filter combinations
13:18
@Magisch well there's no way they can get a line of communication between the two systems without including us.
@M.A.R. Any reason you insist on replying?
@ArtOfCode Pops himself said they want to make it so we aren'T necessary for the success of it.
@JanDvorak I'm just feeding back using this chat.
I already did, no need to stack your own
Whatever man. If it annoys you, I'd stop feeding back.
@JanDvorak That's what happened with FDSC, so I got the impression it doesn't matter
13:20
Isn't the whole point of this room giving feedback?
Come off it, @Mag. SE aren't evil, you should know that much. Heck, we sent them a list of the most successful filters ourselves a couple months ago.
@JanDvorak stacking... meh
doesn't matter too much
The point of this room is SD talking to us
@ArtOfCode Thats not evil
@JanDvorak Moah feedback => Faster deletion
I don't mean this as a negative either
Its just that we need to understand that our whole in the whole integration will be very ... limited
13:21
@Magisch seems like you're saying "SE is going to steal our stuff and leave us behind". That ain't going to happen.
@DavidPostill No. Flags lead to deletion. Green ticks in Metasmoke don't contribute.
Taking our good stuff isn't really our negative though, right?
@DavidPostill not really, only flags on the posts themselves will lead to delete; the Metasmoke feedback does nothing towards deletion.
@ArtOfCode According to undo himself the best possible outcome would be us being unnecessary in this form
@JanDvorak Flags => Green ticks
13:22
yay lag
@DavidPostill but the inverse isn't true; green ticks and MS feedback == X ==> Flags
If they want to take the filters we've created, awesome. They do still have to attribute, we're licensed under MIT/Apache, and any change to that requires unanimous agreement from all 36 code contributors.
@DavidPostill not always
@JanDvorak IIRC there was a userscript that alerted you when Smokey posted a message. Considering how I'm not the only person chatting here, you might find it useful if you find it.
@M.A.R. Not userscript, but a chrome browser extension
Details . . .
Yeah, I don't use browsers that force me to give up my privacy
13:23
But if they're just taking filters, they'll have to keep coming back and collaborating with us as we create more stuff, because we move faster than they do.
and don't respect their users' wishes
Or just move that to mods-only
If they want more effective integration, that means communication between Smokey and SE systems, and that has to involve us - nobody on the SE team has code access.
let mods create filters n stuff
I don't think they'll allow us users to do anything that other users can't
Well me user you're a mod :p
And if that gets done in the way I think (metasmoke <-> SE), then your feedback will directly influence what goes through that connection.
13:25
I'm skeptical that that will happen
It would mean effectively more influence for users with feedback privs on all SE sites
By necessity
I don't think they're likely to give us all double-weight spam flags, no. They might roll out a spam-flag-weight system based on number of helpful spam flags, but that would affect all users not just us.
And they don't do any gating outside rep and tag participation at all and have been pretty rigid on that point
@Magisch yep.
But sometimes, you've gotta roll the dice and see where you end up. They're probably thinking that too.
Would be nice
dont get me wrong.
They should probably give rep for moderation
13:27
Hell, when was the last time there was a community system, willing to integrate, as effective as this one?
Never.
But I don't see it. I think they would be more likely to copy some of the good filters and build a mod-review-queue around the entire thing
This entire thing is unprecedented.
'Sides, you're all still vetted.
Not by SE
It basically is mod-controlled, given that it's mostly moderators doing the vetting.
@Magisch that wouldn't help them the least in expediting spam off the site
13:28
@JanDvorak this
As far as SE knows, all of us without diamonds are no more trustworthy then any other user with that reputation on X site
@Magisch that's not true; you've all been pretty thoroughly vetted by experienced SE moderators.
The basing of trust on reputation (and mod status) has been a fairly rigid part of the entire site from the very outset - I don't think they'll break that paradigm for us
Maybe SD detections could be hidden from the main site until feedback? Then they wouldn't need us.
But I'm always very open to being pleasantly surprised
13:30
Why not? Again, this has no precedent. They know that as well as we do - maybe this is time to break some paradigms.
I like your optimism
My outlook is this: the effectiveness of this entire system is built on the fact that anyone can participate, regardless of rep or diamonds. Anything that changes that will reduce how effective the system is. If SE are serious about getting the best out of integration, they'll realise that and work with it somehow.
If that means people here have more power than other people... yeah, it's not an easy thing to do and we'll have some PR and HR concerns to manage for a long time. But it'll get the most out of the thing we've made.
I like that outlook
@ArtOfCode I can already see meta everywhere going berserk over that
@Magisch oh, so can I.
13:33
There are a couple dozen people on Meta.SO that think what we're doing is a violation of ToS and we should all have been banned a long time ago
But a serious SE will put a CM on that case to say "hey, yeah, we know it's unprecedented, but we trust these guys and we're willing to give it a try - like it or lump it."
If they want to disband us... good luck with the spam.
why are the Meta's so upset?
@JanDvorak Given that Pops called us awesome, I'm gonna say they don't fancy doing that.
13:34
robots taking over there jobs?
@CaffeineAddiction because power
Some people are really finnicky with allowing anyone to do anything
with great power comes great responsibility
@Magisch and I'm glad none of them are part of this project.
Making omelette. Breaking eggs.
If they want to forbid everyone from everything, they should start with asking, not with deleting questions.
13:35
As am I
I mean the entire premises of SE is power to the people ... get rep and then self govern
When I first came here I had about 4k rep on the entire site
If any of those people had been in here I would have never gotten a chance to do anything
Is !!/blacklist still broken btw?
not a clue
lemme ssh in, then you can try it
I refrained from using it for fear of crashing smokey
!!/blacklist CaffeineAddiction
13:38
@CaffeineAddiction You are not a privileged user. Please see the privileges wiki page for information on what privileges are and what is expected of privileged users.
okay go
!!/blacklist thisisatest\.com
@Magisch Got a link to these posts?
@Magisch HEAD isn't at tip of origin's master branch
13:39
Well then
ah, its for domains
@Andy Not on hand
have we got unpulled commits?
Do you want me to dig some out?
that would do it
13:39
Could be
From this morning
!!/pull
Have a pull :p (I'm not using !!/pull)
Restart: API quota is 8117.
@Magisch, yes. I haven't seen claims of ToS violations. I'd like to see the arguments.
13:40
try a blacklist again
!!/blacklist thisisatestthesecond\.com
@Magisch Blacklisted thisisatestthesecond\.com - the entry will be applied via autopull if CI succeeds.
I may have missed it while reading through the transcript, but what is the main concern regarding integration?
Looks promising
@Andy You can probably guess who. No one really clamed ToS stuff. It's just people don't like SOCVR stuff.
13:40
do you guys have a sql backend or are all the domains kept in a text file?
@hichris123 I can guess one main user. But that user is usually more bluster than anything.
Which in some instances is true, some instances is just exaggerated.
CI on daaf6bc succeeded. Message contains 'autopull', pulling...
@Magisch You know I was involved in that conversation right?
13:42
Restart: API quota is 8100.
so no, blacklist is not broken
Yay \o/
@Andy that SE aren't going to want to do anything that'll give privileged users more power than other users.
so they'll end up just taking the patterns, instead of properly integrating the systems
It's possible, but I don't think that's likely.
@Magisch who is pops
!!/start
what is the !!/ for
13:45
@CaffeineAddiction Pops is a community manager (CM) for Stack Exchange.
I see it in the history
@ArtOfCode CM > Diamond?
@MarkYisri that's the syntax that denotes a SmokeDetector command
Aug 16 at 19:14, by Pops
This is absolutely not about you; the work y'all have done on Smokey is incredible--nearly literally so--but over years, people's life situations change, and it would be... awkward, at best... to find ourselves in a spot where you current people have all moved on (next level of school, harder job, more kids) and no new reviewers have stepped in to take your place.
And
Aug 16 at 19:14, by Pops
This is an odd situation that I don't have an answer for yet. On the one hand I agree with what someone said last week that there should be human eyes on this process; we're not advanced enough yet to let spam detection be 100% automated. On the other hand, I'm kinda hesitant to build any system that relies on any small group of humans to vet all posts.
Specifically
@CaffeineAddiction CM's are employees of SE, so yes. Moderators are just high-powered volunteers.
13:46
Tells me they're very reluctant on relying on (unreliable, maybe even non moderator) users for this
@Magisch I see nothing about moderator status in there
I took that as implicit
that's just them saying we want to be cautious about making sure this system will last
for moderators you can ensure they are at least somewhat active
@Magisch I didn't
13:47
normal users can disappear a day later
@Magisch so can moderators
I could leave tomorrow.
yah, us normal users are disposable ;p
You wouldn't be a moderator much longer after that, though
@Magisch six months, to be precise, until SE emails me to find out what happened.
So, my question to both sides: Does it matter if SE does this whole thing in phases? What if they did start by taking the most accurate filters?
Does it matter?
13:49
@Andy are there any systems in place to take the heuristics of which filters are the most used / successful
@Andy I don't think it does. I'm just here to allay concerns that SE is going to take our work and dump us.
We all know that 6-8 units is a very rough estimate for development with SE. They like to roll things out in waves too. What if the first wave is just getting the low hanging fruit?
naa- by Magisch
[ SmokeDetector | MS ] Bad keyword with email in answer, blacklisted user: What was the language of Adam? by elliest on islam.stackexchange.com
tpu- by ArtOfCode
@ArtOfCode Personally, that's a nice victory if they can do that :)
But, like Pops said, in the entire system, there will be some human eyes on it somewhere
@SmokeDetector f
13:51
[ SmokeDetector | MS ] Bad keyword in link text in body, pattern-matching website in body: How to Watch UFC 205 Live Stream? by sports 2478 on superuser.com
tpu- by Magisch
But, their goal isn't complete elimination of spam because it's not realistic. It's reducing the perception of spam overwhelming the sites. We - this room - know there is a lot more being blocked before it's posted. Most other users don't know that. They see spam posts or worse the spam waves, and think SE can't stop it.
If they can reduce that perception, they have a quick win. Then the next phase is when they do the real development work - blocking the spam that's not easily detectable.
Restart: API quota is 7993.
PR to remove the dummy value I just stuck in blacklisted_websites.txt while testing the !!/blacklist command.
why are all the blacklisted domains kept in a flat text file rather than a sql db?
like sqlite3
@CaffeineAddiction shrug
13:58
flat text file is more easily movable and recoverable
And it can be on git
CI on 3bb57da failed.
@Magisch but sqlite3 can be built from a flat text file, dumped to a flat text file and live updated
Ask undo then
He explained it to me like this.
@ArtOfCode We have a CI problem
@Magisch I noticed
@Undo why flat text file instead of sql db? Curious
14:00
But what advantage would a database give us?
live update no need to rebuild, also easier to do things like track statistics of filter usage
I mean you would still rebuild but not just because a domain or user was added
Database wouldn't have a purpose
fetching an entire table is expensive so it'll be done only on startup
we cache the blacklist in memory, so we'd still have to update the memory cache
In which case a textfile serves the exact same purpose but without the overhead of running a database engine
true enough I guess
14:02
sure, we wouldn't have to do a full CI cycle, but it's not like there'd be no updating to do
CI on 0dca9d0 succeeded.
sqlite3 isnt exactly heavy
Hey guys want a little robot fun? (chatbot)
And the CI prevents people from writing in stuff that may break the regex parser
I think
ah, point taken
14:03
@Magisch Um. Not sure on that one.
@ArtOfCode I am guessing unit tests would fail and build would be reverted wouldnt it?
@CaffeineAddiction This entire thing is running on a raspi
@CaffeineAddiction I don't think we have any tests that would catch invalid blacklists.
@Magisch well, not at the moment, but normally yeah
@Magisch I know, my company runs sqlite3 on a Beagle Bone ... it runs like a champ
not that my EC2 instance is much bigger
14:04
@ArtOfCode Yeah, it would break immediately on CI.
@hichris123 would it? Just an entry in the blacklist file? Huh, what tests do we have for that?
running the entire thing through a regex parser I imagine
[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title, pattern-matching website in body: IMPACT Alpha Fuel Reviews RESULTS - FREE scam or Risk? by zamriusmania0 on drupal.stackexchange.com
tpu- by ArtOfCode
@SmokeDetector tp
@MarkYisri If you're about to advertise your chat room again, I suggest you reconsider.
14:18
@ArtOfCode The tests run findspam/etc. If there's an invalid regex, Python will throw an exception. But if the regex doesn't match anything, it won't catch it.
So there is some excellent protection against people breaking the entire thing accidentally
the worst that can get through is someone using a failed regex that wont match anything but wont do anything either
Backoff received of 10 seconds on request to questions/2469;2492?site=monero at 14:20:16 UTC
[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title, pattern-matching website in body: Alpha Fuel Reviews - Product Info - Worth It? by Pariz1958 on workplace.stackexchange.com
tpu- by ArtOfCode
14:39
[ SmokeDetector | MS ] Non-English link in answer: Unable to run git project in android studio by Lê Thường Quân on stackoverflow.com
false positive
15:22
@Undo what is the reasoning behind special_cases > time_sensitive > else in bodyfetcher.py
link?
Not every site gets as much spam as others, most spammers aren't spamming on the math site, while spam on drupal is more frequent, the special cases lists waits until there are at least X amount of posts for that site, before resolving them all in a single API call, some sites are placed in the time sensitive as spam is more common between certain times on the day
Some sites have Spammy Hour
... that ^
@Ferrybig ah, ok so you can pull multiple posts via one API call so special_cases is to make efficient use of this ... got it
15:31
Hi undo
the blacklist command finally works
@SmokeDetector f
 
1 hour later…
16:56
@Ferrybig @ArtOfCode from what I can tell SmokeDetector is making use of ChatExchange ... which does a webscrape of SE instead of using SE API 1.0 or 2.0 ... is this correct or am I missing something?
@CaffeineAddiction correct. We can't use the API, because there is no API for chat.
@ArtOfCode and the powers that be are cool w/ this?
@CaffeineAddiction yeah, there's not enough demand for a chat API.
[ SmokeDetector | MS ] Bad keyword in title, link at end of body, pattern-matching website in body: Muscle Boost X is described as a dietary supplement by kilorgram on security.stackexchange.com
tpu- by MAR
[ SmokeDetector | MS ] Link at end of answer: Bloquear código fonte Django by Alan Ponte on pt.stackoverflow.com
[ SmokeDetector | MS ] Few unique characters in body: HELP please ! Probability question by Andover on stats.stackexchange.com
Backoff received of 10 seconds on request to questions/40301;40303?site=japanese at 17:15:18 UTC
Backoff received of 10 seconds on request to questions/22579;22580?site=french at 17:17:37 UTC
17:33
[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title, phone number detected in body, phone number detected in title: viagra for men 1-800-935-7817 canadian pharmacy,, v.i.a.g.r.a. .f.o.r. .w.o.m.e.n by Shekhar Bachchan on askubuntu.com (@ThomasWard)
@SmokeDetector k
@SmokeDetector Please be consistent here. Is it for men, or women?
Or .w.o.m.e.n?
Backoff received of 10 seconds on request to questions/129619;129617?site=mathematica at 17:41:47 UTC
@M.A.R. lol?
17:53
@SmokeDetector vandalism
Backoff received of 10 seconds on request to questions/129618;129618?site=mathematica at 17:55:04 UTC
How's SmokeAPI going, @Andy?
@ArtOfCode Can you tell I'm working on it? :)
That's where my feature requests are coming from
However, to answer your question - well.
@Andy The requests might have given me a small hint
Awesome
what kind of thing are you making here? Application-type thing, or basically just a Python wrapper to the API?
Python wrapper
18:08
So intended to be re-consumed by other developers? You might want to make the API key a configurable thing in the finished product, then, if you haven't already thought of that, so that not every app that uses the wrapper is locked to your one key.
@ArtOfCode Done :)
Cool :)
SD F
19:18
@SmokeDetector Sounds like a signature thingy
N or F.
@SmokeDetector f because the question asks for recommendations
So F.
Someone should edit the signature out though.
[ SmokeDetector | MS ] Link at end of answer: DSLForge editor to support ANTLR v4 by magdi on stackoverflow.com
naa- by Jan Dvorak
@SmokeDetector email address removed
Dom
Dom
This seems to have missed the SmokeDetector: stackoverflow.com/questions/2349991/…
19:26
@ArtOfCode Thanks for the support :)
no worries
@Dom nah, that's NAA not spam
Dom
Dom
Doesn't it catch NAA also?
Nope, just spam
Dom
Dom
Ok
NAA is way harder to classify
@SmokeDetector f
@SmokeDetector f
19:35
[ SmokeDetector | MS ] Link at end of answer: Is there any Facebook friends network/graph downloader program that still works? by Arlind on softwarerecs.stackexchange.com
@SmokeDetector f
[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title, phone number detected in title: Cheap Viagra Pills <1800-935-7817>++ Buy Viagra Online by Ankit Saxena on askubuntu.com (@ThomasWard)
tpu- by Ferrybig
tpu- by DavidPostill
Restart: API quota is 5530.
20:02
!!/errorlogs 30
20:20
[ SmokeDetector | MS ] Blacklisted website in body: Control Dual Sim Mobile Data by Burks.bro on android.stackexchange.com
tpu- by Kyll
^ 100% scam
burn. with. fire.
20:40
yep
I flagged it as spam but retracted cause figured it was possible that it was a legit number
@CaffeineAddiction absolutely. Having squished this a couple hundred times, I know it's scam.
@ArtOfCode Hmph. Since I'm not not-Iranian, I dunno if there's ever a legit number
@CaffeineAddiction it's a scam
Well, if it's a legit number, what's it doing in an SE answer?
20:42
Hotmail/Microsoft/Canonical/Ubuntu/Yahoo/Google/Facebook etc support staff just don't hang around Stack Exchange to post their support phone numbers. 99.9% chance that's a scam.
@CaffeineAddiction rule of thumb, in my opinion: "If it says ABC support number" and a random toll free number or international number, don't trust it. Burn with fire
I keep typo-hitting enter more and more these days geez
@ArtOfCode replace that with 100%
there is no "Hotmail" support number
true enough
also, google the phone number
you see it ties to multiple "support numbers"
20:42
@ThomasWard yah, i didnt get as far as googling it
ASUS, Acer, Dell, Microsoft, Google, etc.
that's not a "real" number then, so scam.
20:54
@SmokeDetector f
Can/Should we migrate Programmer detected posts to Software Engineering instead?
Can? Probably. Should? Don't know. What does it get us?
Stupid answer: A working favicon.
Non-stupid answer: Everything on the same site remains together.
Can? Definitely. Post.where(:site_id => 50).update_all(:site_id => 330).

« first day (1073 days earlier)      last day (3044 days later) »