Tavern on the Meta

user315433

12:04 AM

The user has two other answers which are about other things...

Jason C

1:44 AM

Comedy gold mine: contentrow.com/tools/link-bait-title-generator

I'm going to start using this to generate titles for my questions.

"8 secrets big companies aren't telling you about ObjectInputStream."

"7 things Lady Gaga has in common with template classes."

Journeyman Geek

Ooo

SmokeDetector

2:42 AM

[ SmokeDetector | MS ] Shortened URL in body: Parenting with automatic weights generates empty vertex groups by Aman on blender.SE

user315433

3:03 AM

@SmokeDetector Sad that the ban on goo.gl was shortlived on all but most active sites.

user315433

Both Super User and Information Security are now https sites

user315433

Apparently, this switch involves changing site id? On the moderator list, SU now got id 428 (was 3) and Security got 429. These ids are normally issued when sites are created...

user315433

The site switcher uses old id numbers, however: 3 for SU, 162 for Security. A mess. :/

Jason C

@NormalHuman I'm sure it's going to be a week full of tweets for nick craver.

I do not envy their todo list.

Journeyman Geek

3:41 AM

lol

@NormalHuman thanks for the headsup

SmokeDetector

5:43 AM

[ SmokeDetector | MS ] URL in title, bad keyword in body, bad keyword in title, pattern-matching website in body, pattern-matching website in title, +1 more: www.hairlosspropecia.org/vlamorous/ by dgasdyhf24 on graphicdesign.SE

[ SmokeDetector | MS ] Offensive answer detected: Does Greek mythology list a calendar date for creation? by john on mythology.SE

SmokeDetector

6:37 AM

[ SmokeDetector | MS ] Blacklisted website in answer: Cannot find where Google Play Music stores downloaded files (different from similar posts) by David jackylucky on android.SE

rene

8:42 AM

delv-pls 20k Excel/word meta.stackexchange.com/questions/292229/excel-word-question /cc @bart @ShadowWizard

delv-pls 20K PA on Ohms meta.stackexchange.com/questions/292226/… /cc @Bart @ShadowWizard

This comment is true but is it rude: meta.stackexchange.com/questions/292230/… ?

SmokeDetector

[ SmokeDetector | MS ] Username similar to website in answer: Looking for a modern-rock song by djbaap on musicfans.SE

SmokeDetector

9:09 AM

[ SmokeDetector | MS ] Bad keyword in link text in answer: Hair loss due to excessive exercise by hellengeek on health.SE

Shadow Wizard

9:32 AM

@rene rude. Period. (flagged.)

rene

Reality can be rude ...

Shadow Wizard

the word is offensive, that's all... :)

@rene already did

@rene I did, why didn't you? Out of del votes?

SmokeDetector

[ SmokeDetector | MS ] Link at end of answer: Global POI database in JSON format? by user6016034 on gis.SE

rene

@ShadowWizard it wasn't at -3 ... so I cheated hoping you would down vote ... :(

SmokeDetector

[ SmokeDetector | MS ] Few unique characters in answer, offensive answer detected: Robots with blood by Ben Dover on worldbuilding.SE

Shadow Wizard

9:41 AM

@rene hmm? Well it is now. :)

Martin

@rene Thanks for the interest in the question (perhaps bug-report) I posted. I have tried to find a few examples where links to meta.SE posts are rendered as titles. (And the source of the post only contains url.)

rene

Yeah, I saw. Definitely some regression. I blame Nick upfront .. ;)

Martin

It seems that it might even be site dependent - after your comment I searched for other examples, so I found out that it works in tag-wikis on math.SE but it does not on MathOverflow. (At least in those examples I was able to find.)

Of course it's nothing critical - now that I am aware of this, I will check results of edits to tag-wikis more carefully.

rene

It has something to do with the domain name I guess. So any site on *.stackexchange.com will do the look-up. That would explain the difference between math.se and MathOverflow.

SmokeDetector

9:55 AM

[ SmokeDetector | MS ] Pattern-matching website in body: Best bike review? by Andrew Dupre on bicycles.SE

[ SmokeDetector | MS ] Pattern-matching website in answer: Best Bike Rack for Honda Pilot? by Andrew Dupre on bicycles.SE

angussidney

@SmokeDetector that's spam, please flag (user has posted 3 other answers with the same website). Also flag the user's other answers while you're at it

SmokeDetector

[ SmokeDetector | MS ] Manually reported answer: POI Datasets for North America by user6016034 on gis.SE

[ SmokeDetector | MS ] Manually reported answer: Where to find POI data for places of worship (e.g. churches)? by user6016034 on gis.SE

rene

10:17 AM

@angussidney now all gone

SmokeDetector

10:58 AM

[ SmokeDetector | MS ] Bad keyword with email in answer: How recent should the English language certificate be to apply for Master thesis? by Lopez Gomez on academia.SE

Ramanujan

1:09 PM

How to get like this in my profile too?

rene

@Ramanujan <img src="http://stackexchange.com/users/flair/9212094.png" />

or using default markup: ![](https://stackexchange.com/users/flair/9212094.png)

Ramanujan

1:29 PM

@rene this show reputation only in chemistry,I want link for total reputation

rene

you need to have > 200 rep on those sites first.

Ramanujan

I will comeback after 8 days if there is some problem

Bye

rene

1:47 PM

@Oded Can I somehow know from looking at your dev/prettify-full.en.js which commit you used from the origin at the prettify git tree? Because I was wondering why lang-rust still uses the build-in lexer but you do have typescript support which was in a later commit. It almost looks like you cherry-picked commits, which seems ... risky?

Jason C

2:19 PM

@Ramanujan Go to your profile. Click the "Edit Profile & Settings" tab. Then, on the left click "Flair". Instructions and options are there.

It will only show chemistry for you because it filters out sites with under 200 rep.

So get more rep on the other sites and they'll end up in the total.

Ramanujan

2:36 PM

@JasonC I have more than 1k reputation on math.se,but due to suspension it is not showing up

Jason C

@Ramanujan Ah yeah. That's because during suspension your rep is temporarily reduced to 1. After the suspension is over it'll show. You may have to wait a day or two though, flair rep display is a couple days behind.

SmokeDetector

3:39 PM

[ SmokeDetector | MS ] Link at end of answer: Looking for specific examples of University spin-offs where the researchers can earn part of the benefits by Mark on academia.SE

Shadow Wizard

4:29 PM

@JasonC 24-36 hours

20

A: Why is my flair not updated with my actual reputation score?

Flair is updated every 24-36 hours. Just wait.

Jason C

@ShadowWizard I.e. "A day or two". :P

Shadow Wizard

@JasonC yup

eh, one of my first questions

Jason C

Thanks. Your unsolicited confirmation has given me confidence to face the rest of the day.

Shadow Wizard

:D

Jason C

Lol

Oh

That reminds me, let's see

Reticulating splines, and.............. drum roll

Wtf

@ShadowWizard Oh lol. My transcript analysis tool picked up your username as "Sha Wiz Dow Ard".

Anyways your most commonly spoken (excluding arbitrary stop words) word in here is "@rene", lol.

And I have to re-run the two-word phrase hold on.

rene

4:35 PM

I must be popular

At least with one member in this room ;)

Jason C

Your most common two word phrase in here excluding phrases that contain said stopwords is: "MOST LIKELY". Followed by "@TGM @TGM", then "BUG REPORT".

I can pull pings, hold on, lol.

ProgramFOX

@JasonC You have been running analysis on all Tavern messages?

Jason C

Yeah

ProgramFOX

haha nice

Jason C

I have all the transcripts if you want them. I can grab them for other rooms too it takes a few hours though from throttling.

rene

4:38 PM

it must be Sunday...

ProgramFOX

I don't need the transcript now but if I ever do I'll ask you

Jason C

@rene Your most common two-word phrase is "CLOSE VOTE", lol.

ProgramFOX

haha

rene

How surprising :/

Jason C

Tavern top 30 ping leaderboard as of yesterday:

  12335: SMOKEDETECTOR
  10006: HICHRIS123
  9883: BJB568
  9847: TIMSTONE
  9403: SHADOWWIZARD
  8069: RENE
  7956: BART
  6936: DRACHENSTERN
  6908: SHOG9
  6850: POPULARDEMAND
  6070: JANDVORAK
  5827: MOSHE
  5595: REBECCACHERNOFF
  5441: YIJIANG
  5392: UNDO
  5220: UNIHEDRON
  4557: INFINITERECURSION
  4491: JASONC
  4487: BALPHA
  4316: BRAIAM
  3670: ANIMUSON
  3621: JADARNEL27
  3436: MOOTINATOR
  3434: RCHERN
  3388: MICHAELMROZEK
  2979: DOORKNOB
  2951: FRANK
  2895: PROGRAMFOX
  2623: THEUNHANDLEDEXCEPTION

ProgramFOX

4:43 PM

I have been pinged 2895 times, wow

Andrew T.

Everybody loves our cat idol more than pair of glasses

Jason C

I should calculate that but also include reply messages that don't contain explicit pings.

Maybe later today.

Maybe this week I'll make a fun web app with full chat search and amusing stats.

It'll probably take all week to get up-to-date on all the rooms' transcripts.

But once I'm caught up it's pretty quick to stay updated.

I just need to not have this happen, or stop using built-in serialization and use a proper database or redis or something.

That's partially also why I can only run two-word phrase checks on one user at a time.

user315433

One duplicate on the list, REBECCACHERNOFF = RCHERN

Jason C

Yeah I pull it by text real quick just now. I do have the necessary information to associate different names of the same user together, though, I'll work on that.

All I have right now is a crapload of raw data, and some hastily thrown together analysis. The stop words are another issue. Instead of a stop list maybe it'd be interesting to just eliminate e.g. the top 100 most common words globally in a given room.

user315433

The number of pings depends on the habit of the person you're talking to often - if they tend to hit reply or just type a message. bjb prefers the former, and hichris talks to him a lot (or used to, anyway), hence the high ping count.

Jason C

4:52 PM

Yeah I want to add unpinged replies into the stats, too.

user315433

I don't see how that is possible

Jason C

The final stop word list is pastebin.com/2sF2ryCB if anybody is curious. That's all the words that were ignored due to both the initial static stop list and pattern matching.

@NormalHuman It is. Ish. I have the reply-to id's for reply messages. The part that isn't possible is replies where the person didn't hit the reply button. But it's as close as you could get.

Now if you've got some crazy nlp ai you want to throw at it to try to figure out who is talking to who about what, be my guest, heh.

Also it counts @NormalHuman @NormalHuman @NormalHuman @NormalHuman as 4 not 1. I will deal with that, too.

ProgramFOX

@JasonC That would actually be an interesting project... take text (representing a chat message) as input and find out which Tavern user said it.

Jason C

I accidentally ignored pings to Won't with the stop list, lol.

@ProgramFOX Oh you mean discarding the known info about who said it and just guess? That would be really interesting. It helps that you have the correct results to train against.

ProgramFOX

yeah

Jason C

4:58 PM

And a pretty massive data set.

You could have similar fun with post comments.

Shadow Wizard

@JasonC yeah, think it was my name when I first used chat

Jason C

That'd make a fun bot, too, lol. If the guess was wrong: "That sounds like something X would say!"

ProgramFOX

If SEDE allows you to download that much data...

Shadow Wizard

@JasonC oh wow... we should get a room, @rene and me..... :D

Jason C

You can get it from the dumps or the API.

ProgramFOX

4:59 PM

it probably does, isn't there something like data dump?

yeah :P

Shadow Wizard

@JasonC huh? Who's @TGM?

ProgramFOX

@ShadowWizard TGMCians

Jason C

Probably would take forever with the API. What I'd do is use the dumps, then the API to make up for the out-of-date ness.

And I'd wait until the next quarterly-ish dump update.

Shadow Wizard

@JasonC WHAT?!?!?! I'm not in top 3?! BLASPHEMY! :P

Well, things are better in Den.... :D

Jason C

Actually I don't think I could realistically manage keeping my internal comment list up to date. SE has like... data centers for that. My laptop and internet connection, maybe not so much.

Shadow Wizard

5:01 PM

@ProgramFOX but why I pinged him so much?

ProgramFOX

@ShadowWizard You know that better than I do.

Jason C

@ShadowWizard It's your most common two-word phrase. It's misleading.

ProgramFOX

@JasonC I'm currently messing with a mail server on my cloud, but if that's done I'd like to get your transcript to see if neural networks are somewhat useful for classifying which user said what.

Jason C

You probably had one message where you typed @ TGM like a zillion times.

He's low on your actual ping list.

@ShadowWizard your top 5 actual pings are:

WORD: 1440: @ RENE
WORD: 1151: @ BART
WORD: 534: @ DERPY
WORD: 493: @ JANDVORAK
WORD: 446: @ TELKITTY

bjb568

@JasonC wow, I'm just a few thousand below our beloved bot

Jason C

5:04 PM

Lol

Shadow Wizard

@ProgramFOX maybe knew... Honestly need to search to find who this is. Lol

Andrew T.

prepare for ping-fest

ProgramFOX

@Sha There you go chat.meta.stackexchange.com/…

Jason C

@ProgramFOX Yeah just lmk I can zip it all up it's not too unwieldy.

@ShadowWizard chat.meta.stackexchange.com/transcript/89?m=2596365#2596365 is the two-word phrase culprit.

Shadow Wizard

Lol

Shadow Wizard

5:23 PM

Heh, wonder what is @Bart's most common word? @Jason

Probably stick... Or smiting... ;)

Jason C

@ShadowWizard "QUESTION", "META", "REALLY".

I'll improve analysis later though. Right now the stop words are making the results less than interesting.

It's just very quick text stuff, I put all the effort into obtaining and parsing the data, all the analysis is currently in the "oooooh now that I have it let's see what I can hack in real quick" phase.

Shadow Wizard

Kenny is dead... :-(

Jason C

Also on the todo list is normalizing word forms, and less aggressive word boundaries.

Shadow Wizard

@JasonC lol, unexpected!

Jason C

And not splitting hyperlinks at word boundaries. I had to filter out "COM" and "STACKOVERFLOW", for example, as a quick way around not treating links as their own words.

Shadow Wizard

5:31 PM

@Jason maybe have a lol leaderboard too?

Jason C

Haha I can do that now hold on.

What do we want: ha, haha, lol, lmao, rofl?

hah

Shadow Wizard

Heh

Think only lol will do ;)

Jason C

5:49 PM

@ShadowWizard docs.google.com/spreadsheets/d/… In tabular form for your sorting pleasure.

Note that, sadly, nobody has ever said l0l or l0lz (until now).

Our top 20 lol'ers:

Josh	1411
Sha Wiz Dow Ard	1153
George Marian	996
Jason C	807
UniKitty	778
drachenstern	683
rchern	406
hichris123	388
Neal	335
user159773	283
J. Musser	280
mootinator	253
Doorknob	251
Moshe	246
Sam	240
Aiden Bell	235
ircmaxell	232
Unihedron	219
Vishwas Sharma	218
Fosco	185

That's by user ID not username so no splits there.

Ha, our top (and only) 10 lawlers on the other hand:

Jason C	19
mootinator	3
bjb568	2
Anna Lear	2
George Marian	1
Moshe	1
badp	1
GnomeSlice	1
nhahtdh	1
ArceusMaster0493	1

I really need to stop saying that.

Oh I should do lols:message ratios.

Maybe later

bjb568

You're doing good work.

Shadow Wizard

@JasonC lol only second

Who is Mr. Sharma?

OK, next... NSFW words leaderboard? @Jason is this feasible? ;)

Jason C

6:19 PM

@ShadowWizard Gimme a list.

user315433

Seven dirty words

Jason C

This'll be split a bit since I don't normalize word forms.

Lol wow

I cobbled together a list of 32 off the top of my head and amusingly the only one that was never said in here was "c******ker" from the 7 dirty words list.

We are a dirty room.

ProgramFOX

Yay... I spent an hour wondering why my mail server didn't work when I accessed it through port 25 and it didn't even log anything.

Turns out, my ISP blocks port 25 :P

It finally works now.

bjb568

yay

ProgramFOX

Except, mails sent from it end up in your spam folder, probably because Amazon still needs to process my request to give my server a PTR record.

@JasonC Are you still here, and able to send me the transcript files?

Jason C

6:28 PM

Haha

@ProgramFOX Yeah one second

@ShadowWizard @NormalHuman NSFW word list in tavern (list itself highly NSFW) ordered by users with highest total counts from my list.

SmokeDetector having the foulest mouth by far.

The inclusion of "damn" is debatable.

@ProgramFOX Gimme a few to compress and upload everything, will take about... 20 min-ish.

ProgramFOX

:o

Jason C

Don't blame me, 7zip lzma is slow as hell.

I'm gonna keep that one off the star wall, sorry, lol.

user315433

bjb is tied with Shog at 109 dirty words total... that fancy college-prep education is showing.

Jason C

@ProgramFOX Ok eta stabilized at 26 minutes from now. I'm gonna grab lunch. I'll ping you later.

ProgramFOX

Okay

bjb568

6:56 PM

@NormalHuman I am Shog'a equal.

Me an BIG KITTEH now! :3

rene

7:07 PM

Would it be OK to flag this as abusive?

Jason C

@ProgramFOX dropbox.com/s/dvuysx49qsok097/tavern_transcripts_02.7z?dl=0. So, there's a couple big CSVs, and also all the raw html. There's "clean text" in the csvs but, it's sloppy, and also breaks up e.g. hyperlinks and stuff without caring. So you may want to use the regular "text" or the "html" columns, or if you want to just do something better, go through the raw html files.

Also I quickly cobbled together the CSV writer code, I can't guarantee that they aren't broken, all I did was quote everything and replace " with "".

But the html files are there so, whatever.

ProgramFOX

7:44 PM

@JasonC cool, thanks

SmokeDetector

8:22 PM

[ SmokeDetector | MS ] Link at end of body: Touchpad/Trackpad Drivers by Alpaca Your bags on askubuntu.com

user315433

Really, a.cocaine.ninja/grvhpa.png ?

ProgramFOX

With a very basic Python program, I counted that there have been 202741 different words in the Tavern (including stop words).

That immediately kills my initial idea of the program to try to identify a message's poster from the content :/

Because it would eat memory too quickly.

user315433

Alternative idea: a web app "guess who said that".

ProgramFOX

that's an idea!

Jason C

@ProgramFOX I got 245180 from the "clean" text. But yeah same ballpark. And yeah memory constraints have been a bitch.

I guess I could turn my virtual memory back on. I turned it off to spare my SSD from abuse. But if I'm going to do some disk-based storage I might as well just try using a crapload of "ram" first.

M.A.R.

8:36 PM

Hey, what's my most-said word?

Jason C

@M.A.R. "@SMOKEDETECTOR", lol.

M.A.R.

O.o

ProgramFOX

xD

Andrew T.

Yeah... he it is popular

M.A.R.

@JasonC How many times have I been talking to a bot?

Jason C

8:38 PM

@M.A.R.

WORD: 546: @ SMOKEDETECTOR
WORD: 433: @ SHADOWWIZARD
WORD: 339: META
WORD: 325: @ RENE
WORD: 264: PEOPLE
WORD: 262: OH
WORD: 233: SOMETHING

Andrew T.

Oh

M.A.R.

Ow

@Sha you're not trying enough to be popular

Jason C

"meta" may come from pasted links, so take that one with a grain of salt.

M.A.R.

@JasonC Might be because I never say MSO or MSE to avoid confusion, but always say meta.SE and meta.SO

Jason C

You do talk about people a lot, lol: chat.meta.stackexchange.com/…

(My totals don't count onebox text)

ProgramFOX

8:40 PM

what's my list of words, @JasonC?

Jason C

WORD: 535: @ SMOKEDETECTOR
WORD: 336: SPAM
WORD: 297: QUESTIONS
WORD: 273: STACKEXCHANGE
WORD: 236: OH
WORD: 234: BECAUSE
WORD: 213: @ SHADOWWIZARD
WORD: 205: @ BJB568
WORD: 191: POST
WORD: 191: TPU
WORD: 183: META
WORD: 174: @ UNIHEDRON
WORD: 172: MESSAGE
WORD: 166: YES
WORD: 161: QUESTION
WORD: 159: @ HICHRIS123
WORD: 147: SMOKEY
WORD: 140: @ JANDVORAK
WORD: 139: FP
WORD: 135: ALIVE
WORD: 133: CHAT
WORD: 128: @ RENE
WORD: 127: GONE
WORD: 126: YEAH
WORD: 123: ANSWER
WORD: 122: BOT
WORD: 121: USE

M.A.R.

Message of the year

Mar 1 at 7:58, by M.A.R.

Why? Why do people ask stupid questions? Why do people not think? Why do people think asking multiple questions instead of thinking would work? Why? Can stupid questions have any meaning?

Jason C

You and mar are very similar lol.

ProgramFOX

"Alive" haha

M.A.R.

@JasonC Oh, I have a feeling about Rubik cubes now

Jason C

8:43 PM

Lol

Smokey's is kinda funny.

It's the expected message words then transitions into spam keywords like "skin", "muscle".

"Drupal" is near the top of smokeys list, lol.

M.A.R.

Hehe

Jason C

I wish there was an easy way to figure out who was actually pinged from partial ping names.

M.A.R.

@JasonC And the fact that I have gone by a million names in the meanwhile

Jason C

@M.A.R. 280934: [DEAD, I??, M.A.R., MARamezani, PhMgBr, Rubisco, TIPS, in?z?m??.A.M, ?.?.?., ???] -- non-ascii be damned.

M.A.R.

Sounds about right.

Jason C

8:55 PM

Whoa wait a minute

@NormalHuman You're zaq!

Lol

Oh man:

811: [9 Shogs a-Shogging, Joshua H. Heyer, Nine Shogs Shogging, Shog9, ShogDYIN', Shogging through the snow]

I vote "9 Shogs a-Shogging"

1: [Community, Jeff Atwood]

:O

ProgramFOX

If I only count users who posted 3000 messages during the past 365 days, then we have 13 users posting 40703 different words. That would be less heavy to process in memory.

Jason C

That seems reasonable.

ProgramFOX

It's probably reasonable as well to exclude Smokey from that list.

Jason C

I mean if it's just single words and you're trying to do something with a map keyed on word, I got good memory efficiency with a compact trie.

ProgramFOX

no that's not what I'll be trying to do

Jason C

9:01 PM

But

Butts

We've all got one.

ProgramFOX

I was thinking to see how far I could get with neural networks, but that means employing a weight matrix of size (different_words x users) and that's big.

bjb568

9:44 PM

Kool kids do linear algebra without matrices.

You could try to exploit some kind of sparse structure or find some algebra whitchcraft that will let you get away with a smaller matrix and probabilistic sampling.

Shadow Wizard

@JasonC hehe, better not run this in Den.... :D

Thanks!

@JasonC nope, two.

(I think?)

2

@M.A.R. I am popular for you... :P

Jason C

10:01 PM

@ShadowWizard Lol what?

We've all got two butts (you think)?

I dunno how you do it on your side of the planet but over here the butt:people ratio is roughly 1:1.

Shadow Wizard

hehe, for some reason thought butt only refers to one "cheek". Blonde moment. :P

Jason C

@bjb568 @ProgramFOX You could definitely make the matrix smaller if you filter out words with word:user covariance below a certain threshold although you still have the problem of calculating covariances.

But I know this is possible. Actually, a long time ago I had a similar question, this lady I worked with solved it and I never figured out how. I'll have to see if she's still around / alive. Somehow she was doing PCA on 50000 300000-dimensional points, successfully.

And I had meant to ask her about it because I wanted to do something similar for an audio analysis project, but then I never finished the project and forgot.

Shadow Wizard

:/

Jason C

And it has kinda bugged me ever since.

Shadow Wizard

10:38 PM

NAA one more delete vote needed @Bart @rene

Jason C

@ShadowWizard Word to the wise: Watch out with the hot question dupes, the HNQ vs. the "hot" tab can catch you by surprise. The hot-questions tag seems to be pretty ambiguously used.

Shadow Wizard

@JasonC yeah, I know. Big mess. But I just re-closed an old question.

Jason C

Then there's all those questions from when the hnq was on the dropdown... I gave up trying to make sense of it, lol.

Shadow Wizard

I'm 99.999% sure closed question won't appear in HNQ list, but can't find official confirmation.

@JasonC yup, I think the algorithm is the same, but again can't be sure. Though @Shog can confirm once and for all... :)

Jason C

@ShadowWizard It's weird, isn't it? I was sure too and couldn't find anything on it except vague discussions about maybe possibly proposed possibilities for possible considerations of potential algorithms.

Shadow Wizard

10:46 PM

Yup.

Jason C

It's super weird that meta.stackexchange.com/questions/292214/… has so many downvotes.

Shadow Wizard

@JasonC no idea either, yeah

Jason C

Ah I guess rene's edit improved it a lot.

I kinda wish more people would come into this chat room and complain about bans and stuff. It's really entertaining. It doesn't happen enough.

It's fun to fend them off, and even more fun to convert them from whiners to cat-gif posters.

Shadow Wizard