« first day (499 days earlier)      last day (3655 days later) » 
00:00 - 23:0023:00 - 00:00

11:02 PM
0
Q: Duplicates MKII, Canonicals, more aggressive duplicate prevention

Second RikudoThis question is another take on the duplicates problem. It steals is inspired by ideas from the verious posts and feature-reqeusts posted here on meta. Disclaimer: In this post, I am referring to the "Users" or "The People" or "The Gang" or "Everyone". What I actually mean is the bunch of folks...

@Mysticial ^
What do you think overall though @Mysticial?
 
@MadaraUchiha I like the idea. But I'm afraid that we don't have the technical capability of selecting the right duplicate.
 
@Mysticial Why's that?
 
We're not Google. xD
 
Because you have to get the suggested dupe correct. Right now, the "correct" dupe that is auto-selected is only in the top 5 if at all.
 
@Mysticial That's where the suggestion for the "canonical weight" comes in
The higher the canonical weight a question has, the stronger it appears as a duplicate source.
 
11:14 PM
@MadaraUchiha Which I get the feeling will require a lot of tuning. And the false positive rate would be very high.
I'm not against the idea of presenting a list of questions. And if the user says "no" to all of them in a short amount of time and it gets closed as a dupe of one of them - insta-question-ban the user.
If the user brushes it all aside, that user clearly doesn't give a shit.
In which case, we (the users) probably shouldn't give a shit either.
Another brutal idea would be to "quietly" partition the user-base.
Going back to my 4 camps answer. There's really only 2 factions at odds with each other. Caretakers vs. vampires/whores.
If you could separate the communities, everyone would be happy.
But of course we can't really do that on SO. Since that goes back to gazzilion dupes of having a "newbie" version of SO.
But what you could do is algorithmically bias the questions that users will see.
If you algorithmically detect a user is camp 3 (repwhore), then bias the questions that they see towards those from all the camp 2 (help vampires).
If you algorithmically detect a user is camp 1 (caretaker), bias their front page in favor of questions from other camp 1 caretakers.
I don't think it will be too difficult to find the right algorithm to categorize each user.
It should be pretty clear cut in most cases.
 
I think dividing the question votes into several categories would significantly help the noobs determine what the hell is wrong with their questions
for example add a special vote button for high-rep users (camp 1) like "no research" and that would automatically make a template message appear explaining to the poster what they're lacking
 
@user1306322 The problem is that it would be abused by trigger-happy users.
High-rep doesn't mean trusted.
And you could say that in a comment instead.
 
then the punishment for that should be severe
this is a very important function, not just upvote/downvote
which by the way still is regulated in some ways (mass vote reversal) if abused
flags too are regulated somewhat (reduced number of flags for declined flag count)
 
11:30 PM
It's a big can of worms.
That might cause more problems than it would solve.
 
and honestly would you rather post a comment by typing the message manually each time you see a bad question or just press "no research" or "no code" instead?
it will convey exactly the same message to the OP
and take less time
 
@user1306322 There was a movement to that a while back. They had a meta post of canned comments.
But Shog nuked the entire question because all the canned comments were too snarky.
 
this might be dangerous, yeah, but it shouldn't be a very widely accessible function to avoid mess
actually if most users can't be trusted with the quality control of the site's content, I really have very little ideas how to save the situation :\
if only there was a core base of users who know exactly how the things should be done, everybody else could try to follow their example
otherwise I just don't know
democracy doesn't do a very good job for countries either
 
@user1306322 that's because none of them are an actual democracy
 
@Krazer neither is SO
and it shouldn't be
 
11:37 PM
SO cannot be a democracy. Because the help vampires out-number the good users 100:1.
 
right now with how reputation works out for SO I don't think it's a valid approach to that specific site
 
I've been thinking about it, and it seems to me that the reason why the category 1 people are so annoyed with everything is because they constantly see bad questions when they browse the site, and bad questions are bad. To make them happy, all you have to do is balance out the bad stuff with better good stuff.
So just add some moe anime girls to the header and sidebar and it will fix all the hate.
6
 
other, lesser sites seem okay right now
 
@LoganM Or just hide the bad stuff from them.
 
@LoganM I support this resolution
 
11:38 PM
That idea has been thrown around before. But it got repeatedly killed off on the old meta.SO.
 
do that and it won't be long before you start hating them too
people will go mad
 
@user1306322 Hating who? The moe anime girls? Who would do that?
 
I was actually thinking that we might want some canned comments here, particularly for use on rec requests. TeX.SE uses them pretty extensively (meta.tex.stackexchange.com/questions/430/text-building-blocks).
 
"Y'all need moe!" will be heard on every corner
it could be a very harsh solution to the problem, but we could make up some "entry exam" of sorts to figure out whether a user is at least competent enough to read the help, and at most will try to find an existing solution by searching similar questions before posting his own
 
@senshin There are some situations that seem to come up repeatedly, like new users asking for recommendations. It wouldn't be a terrible idea to have prepared comments for those situations.
 
11:44 PM
I concur
 
well yeah, copypasta doesn't just appear on its own, it's very useful
but there is a ton of different cases as can be seen on that TeX post, there should be some auto-detection for the most relevant to be pushed to the top
like would put the corresponding auto-comment on top of the copypasta list
you guys know that without me
 
This is interesting: our second highest referer for this month is german.se
 
@Krazer how do you know that and what does it mean?
 
@user1306322 Mods are omniscient.
 
Mods can see things that we can't even smell.
 
11:51 PM
@user1306322 because my name is blue, and no, respectively
 
@Mysticial maybe that's a good thing for us, considering the situation on SO
 
@Krazer How is that even possible? Where would they click in from besides the hot questions list?
 
@senshin I don't know but SO is still #1
 
@senshin Hot questions, sidebar ads, user profiles, or the topbar.
 
I odn't recal we having any hot questions since 3/25
 
11:53 PM
the second most referred language in anime seems to be German…
 
@Krazer I think we might have had one or two that made it on the list since then, but none that lasted very long
 
Facebook and what I assume is G+ traffic is starting to pick up very gradually
fairy tail 2014 and sasuke rinnegan are hot topic this month it seems
 
speaking of democracies, what do you guys think could be the most efficient way to rule countries?
 
They do have this meta post, but it's only got about 50 views:
3
A: Are question on the meaning of spoken dialogues on topic?

WrzlprmftI address the several arguments against such questions separately: Not of general interest: I think individual movies are of sufficiently general interest for a Stack Exchange as they are consumed by a wide audience. With Science Fiction & Fantasy, Movies & TV and Anime & Manga, there are at le...

 
@MadaraUchiha Actually, now that I think about the whole "separating the factions" idea, it might just be as simple as using the tag system.
 
11:58 PM
AFAICT that's the only place that anyone's linked to us on their site.
 
Jeff Atwood had an idea of computing a user's "citizen score".
Help vampires having the lowest score and vice versa.
You can then place these into buckets and assign them a tag.
And we already know that SO does a great job of biasing the questions that we see towards the tags that we are interested in.
 
00:00 - 23:0023:00 - 00:00

« first day (499 days earlier)      last day (3655 days later) »