« first day (2785 days earlier)      last day (2280 days later) » 
00:00 - 21:0021:00 - 23:00

9:00 PM
@TravisJ Sidebart?
@TravisJ I don't think that adds much, TBH
"Unclear" as a gold-badge privilege + reopen for any reason would go a long, long way I think
One-vote reopen? Now you're pushing it :P
@JohnDvorak already applies to Duplicates
Make it apply to anything
If you know the tag well enough to earn a gold badge, you're probably more qualified than most to say what's Unclear or Too Broad
9:03 PM
Or, heck, opinion-based or off-topic
My gut feeling is that a big chunk of frustration on SO involves the time it takes to close bad questions and the time it takes to get decent ones reopened.
Not the closing or reopening itself. The time.
What about the questions that lack proper tagging - or even any possibility of a proper tagging?
Yeah, the time. Wasting a close vote is a real discourager.
@JohnDvorak well, suddenly there's a real advantage to earning a gold badge in something like "html", eh?
I can confirm your gut feeling. Also, it's frustrating just how much must be left opened due to the lack of votes / closers' will. Also, how many answers slip by before the question is closed. Or, hell, how many answers slip by after the question is closed.
9:06 PM
But yeah... If I have 30 minutes free after eating lunch, and I see a question I want to answer that's closed... I'm not gonna see it reopened. And if I see a question going off the rails, I might be able to hang around long enough to see it closed but... It's a crap shoot and not very satisfying.
Definitely agree with that. Although the want to answer one is a little rarer.
Of course, I can because I get unlimited binding reopen and close votes. So that statement is a bit inaccurate.
Even with SOCVR, it's tedious.
But presumably I won't have that forever, and... It'd be kinda nice to not hate the site once that happens.
9:07 PM
@Shog9 sounds like you better knock out about 9k reviews tonight
@K.Dᴀᴠɪs see, that's another thing that's just really unsatisfying: review.
Yeah, the sheer volume of the close queue probably doesn't help.
@Catija You should have seen it in the past! :P
I'll occasionally find the time to review, and when I do I'll filter the queue down by as many tags (3) as I can, or just pick really specific ones.
It was like trying to put out a firenado with a squirt gun.
9:08 PM
@Shog9 Do we really need 5 cv? I think we can drop it to 4 and see a drop in the queue itself
I always have to ask to see it at all. Has the new top bar alert system helped at all?
But otherwise... So many questions I don't have an opinion on.
In my rare forays in the Close review I wasn't impressed by the level of necessity of closing most of those posts.
I simply do not care about someone's React code dump. Maybe if I used React I'd be all excited about it. Or hate it with a passion. But... I don't. So I feel nothing. No necessity.
Hell, just give 1.5 vote weight to silver badgers and the CV queue won't stay for long
9:10 PM
How many silver badgers are there on SO?
How many people use their full number of reviews on a daily basis?
And see, that's the problem: the CV queue is probably already too small.
Too small?
@K.Dᴀᴠɪs Fun fact: 4 won't work, but 3 will.
As in votes age too quickly?
9:11 PM
@Catija a handful
@JonEricson Interesting
As in, it's just a drop in the ocean of the stuff that needs to be closed, and only a fraction of that drop is effectual anyways?
That's odd.
@JohnDvorak as in, if I want to focus on questions I care about, there aren't enough of them
Like... If you're into every PHP question, there's plenty of fish
@JonEricson hmmm
@Shog9 well then it can be dynamic by tag.... If a tag is flooded with requests, lower the close threshold.
Otherwise, keep it at 5
9:16 PM
The other problem with the CVQ is that there's no obvious links to the filters. If I am super into PHP questions... Or Python... Or whatever is tops that week... I gotta know to filter, or it's gonna be hit & miss. There's some automatic filtering based on your tag prefs, but that doesn't work great for smaller queues either.
We had a system for that, but it was so slow it broke the site.
Being super into X sounds exhausting.
With no obvious benefit to the super-being person.
What? No, it's awesome I'm sure, just look how cheerful the Colonel is.
From KFC?
well, him too, but being dead'll do that to you
Apparently prior to that he was really pissed about what KFC did to his gravy recipe
9:24 PM
Apparently, their Twitter account has an awesome Easter Egg in their followers list.
But that's about all I can say about them. I don't think I've eaten at a KFC in the last decade.
so good
nothing beats a bucket of chicken
that good bucket flavor just seeps into the meat
it's just so good when it hits your lips
whoa there. Inappropriate
Clearly you haven't seen Old School.
@K.Dᴀᴠɪs For reference: youtube.com/watch?v=z5BeptSdH7M SFW
9:27 PM
#1 thing... Don't take me seriously... Ever xD
You mean, like meow?
@Shog9 needs moar Archie
When you going to run your little AU experiment on SO, @Shog9?
once I finish updating some scripts
9:34 PM
Do you expect us to have a 1 mil + user still?
Are you planning to invalidate a large chunk of JS's upvotes?
@TravisJ you know, Australia. The land down under. Where women glow and men plunder.
I just want to have popcorn ready for meta
Yeah, I been there, just thought maybe there was a different reference being used.
9:35 PM
my theory is that giving every SO user a Vegemite sandwich will fix what ails 'em
Two Australian candidates in this election, likely at least one wins.
@Shog9 Only if by "fix them" you mean "kill them" :P
Plundering will ensue shortly.
Ahh Shog's waiting for after the election. So he doesn't invalidate a couple potential mods xD
So, here's my current working plan...
Back when I did the AU invalidations, I didn't have a very reliable way of automatically weeding out false positives
9:39 PM
@Shog9 .................................................... ouch
So, I was really conservative
Since then, I've learned a bit more about SQL
And the terrible, terrible things you can do with recursive queries
This sounds... like I will need to double up on my popcorn
take a vacation from work
watch the fireworks
Also, I found out there's such a thing as community search, which means I can be lazy to some degree and rely on the work of others without having to understand their magic maths
...that's how this works, right? Copypasta == perfection w/ 0 effort?
9:42 PM
sounds legit
Anyway... So we have this built-in system for identifying suspicious votes. Which is pretty solid at picking out nastiness but bad at identifying the size of it.
Chances are, if it caches 3 suspicious votes there's like 4, or 10
If it catches 2, every day... There's probably dozens.
But cranking up the sensitivity gets you more false positives
... and more pissed off people
except on AU, where they're just happy you left your Vegemite at home
so instead, I'm gonna start with just looking for this signal and tracing out from it. If I can take that pretty reliable 3-vote sequence and invalidate 4, that saves a lot of annoyance right away.
@ShadowWizard :)
If someone's constantly triggering low-level alerts... Then I start the big slow analysis that traces out all their connections to everyone else on the site or network. Identifies unlikely levels of overlap. Identifies their entire history of cross-voting. And invalidates it.
9:46 PM
What about people with like 20 rep and 700 upvotes
There are just people like that
There are people who do nothing but suggest edits and use the rep to downvote stuff they don't like
lots of benign outliers
there are also people who try to hide who they're targeting by burying it in a huge noisy wave of votes to others
it's not usually as hidden as they suspect
the problem with, say, upvoting the top 100 users on the site and Bob who is totally not a sock is that Bob sticks out like a sore thumb
well, especially when their IP's match.... Ooooops
poor bob. He was always a nice guy imo
there was someone on SO recently who created two accounts and scripted their voting to coincide with automatic deletion. That was fun.
I'm somewhat curious about how the small tags will play out. Like when there's 8 people watching the tag, and your votes, good or bad, come from those 8 people.
@K.Dᴀᴠɪs amusingly, the patterns there are usually less suspicious than elsewhere.
Except when they're all co-workers and have a morning routine of upvoting everything that was posted for the past day.
9:53 PM
that's surprising
Those... kinda suck.
of course
But anyway. if this pans out, I'll be able to stick with hitting maybe a thousand votes a day, correcting as I go, instead of millions.
So instead of one big hit, you're talking about months of depression
If I do this right, that should fall off pretty quickly
Folks learn to not do things
The current scripts hit maybe 100/day
of course, a good chunk of that is because folks work around them
so, less predictable is good
10:00 PM
this room certainly got quiet
@K.Dᴀᴠɪs But that implies that the 8 people watching the tag are the only 8 people answering questions about it. Plus, I don't think that there's enough voting on SO to get too worried about that. My impression is that the average question/answer doesn't get much over 2-3 votes, if that...
OK, median.
I'm guessing the average gets skewed by the posts with 1000+ votes.
It depends on what you include.
prepares to lose half of rep due to SFTP/asmeurer cross-voting in sympy tag
10:05 PM
Many posts which do not make it to 35 views or 1 upvote end up wasting away, and will drag down the average.
I'm sure that SEDE can make a pretty table of posts by score.
Kinda looks like an L
log scale pls
that's posts that got at least 1 vote in the last year; if you include those that got none...
...it still looks like an L
L is a pretty letter.
10:07 PM
I try to ignore 0 voted posts now.
If someone else didn't explicitly feel the need to see an answer then why bother :P
> score:1 views:35 closed:no is:question created:2d.. intags:mine
> 7-14 days
someone got the estimate wrong here...
... So... what does 6.58 mean?
It's between 1000000 and 10000000
Log 10 is the number of zeros after 1.
10:17 PM
0 means there is 1 such post. 1 that there are 10. And so on.
So, if I want to calculate that... what do I have to input to convert 6.58 into 3,837,680
Also, will SEDE label both the x and y values?
not readily
@Catija 10^6.58 in Chrome location bar works. Or in Google search, or a calculator, or WolframAlpha.
I prefer the location bar, because it shows the result without navigating away.
@SFTP That's fine... I just didn't know the equation. :)
00:00 - 21:0021:00 - 23:00

« first day (2785 days earlier)      last day (2280 days later) »