20

I would like to better understand how attempts to circumvent Stack Overflow question blocks impact other sites. I believe this can be estimated by finding out how many migrations to Stack Overflow were rejected because asker is question blocked.

Can we get some statistics on that please? I am particularly interested in numbers for last month, 2, 3 and 4 months, preferably broken down by "source" sites.


Not directly related, but while we're at it, I would appreciate to additionally see similar statistics of migrations to Stack Overflow that were rejected because of cross-posting.

  • more recent stats for Programmers (April 2016 and April 2015) are available at Programmers meta: How many questions do we get from users recently blocked at SO, how many of these are closed / deleted? – gnat Aug 3 '16 at 6:15
  • 1
    That methodology would severely underestimate the impact. The kind of user who would be question-banned on Stack Overflow would also be likely to post their low-quality questions elsewhere. Those low-quality questions should not be eligible for migration to Stack Overflow, based on the "Don't migrate crap" rule. – 200_success Oct 29 '16 at 8:12
  • @200_success agree - back then I didn't know that there is a better way to estimate the impact based on an (unknown to me then) fact that system keeps records of when user bumped into block. More accurate stats based on this better way were obtained for two next years (2015 and 2016) for one of the sites listed here per request referred to in my prior comment – gnat Oct 29 '16 at 8:38
17

Ok, here are the numbers for the past 120 days of migrations. Extracting the intended destination site from the logs is enough of a pain that I didn't bother, but this is broken down by would-be source site:

Migrations blocked by q-bans Source Site Name 
---------------------------- --------- 
17                           Super User
4                            Server Fault
4                            Game Developers
10                           Statistical Analysis
1                            Web Apps  
11                           Webmasters
2                            Apple     
1                            Ubuntu    
10                           Unix and Linux
3                            WordPress 
63                           Programmers
4                            IT Security
3                            Graphic Design
7                            Database Administrators
2                            Code Review
1                            Code Golf 
1                            Signal Processing
1                            Academia  
1                            Computer Science
1                            Windows Phone
1                            ExpressionEngine
1                            Magento   
1                            Network Engineering Stack Exchange
1                            MathOverflow
Migrations blocked as dups Source Site Name 
-------------------------- --------- 
16                         Stack Overflow
7                          Super User
20                         Server Fault
2                          Mathematics
5                          Unix and Linux
10                         Programmers
1                          IT Security
6                          Database Administrators
3                          Mathematica
9                          MathOverflow

And finally, here's a fairly useless graph of q-ban-rejected-migrations-by-month for Progse, over the past 12 months:

Migrations from Programmers.SE to SO blocked because author was q-banned on SO


All that said, this is a really imprecise way of gauging how many folks are asking questions on Progse because they got q-banned on SO. Here are some better numbers:

  • 3884 users with accounts on Stack Overflow asked questions on Programmers during the past 120 days
  • 442 of those users have been q-banned on SO at some point in time
  • 320 of those users have been q-banned on SO during the past 120 days

Those 442 users asked 613 questions on Progse during this time period, and out of those questions:

  • 424 were closed
  • 429 were deleted
  • 414 scored less than 0
  • 76 scored more than 0

For comparison, 5038 questions were asked on Progse during the past 120 days by folks who didn't have q-banned accounts on SO, and out of those:

  • 2724 were closed
  • 2429 were deleted
  • 2035 scored less than 0
  • 1448 scored more than 0

Here's a slightly less useless graph to go with all that:

questions asked on progse by q-banned SO users

Tim's been working on a design for enhancing our our recidivism system to discourage this and other sorts of unproductive behavior.

  • thanks! Looks like Programmers is a place to go for those willing to trick the question blocks :) – gnat Aug 5 '14 at 7:17
  • regarding graph, to make it less "useless", you may additionally feed into it data on questions closed by Oded, with comments containing word "ban" – gnat Aug 5 '14 at 14:35
  • If we're interested in making this at all accurate, might as well cross-reference the accounts and logs & use that then, @gnat. Keep in mind, q-bans are not a fixed state - they're based on the current state of your past questions when you go to ask a new one, so it's quite possible for folks to quality for a ban in the past but not currently, and just as possible to quality for a ban now but not have ever actually encountered it. – Shog9 Aug 5 '14 at 23:19
  • Out of curiosity, would it be possible to look for "Of those 613 questions, n were titled exactly the same as a question they asked on SO" – Travis J Aug 5 '14 at 23:31
  • 4
    Sure, but that's a lot of extra work for little gain; 70% of these questions are crap, does it really matter what sort of crap? Either they're asking on one site because they got blocked on another, or they're just really bad at asking questions on any topic. And yet in a certain sense, these are the cream of the crop - over this same time period, 25,087 users encountered the q-ban on SO, so we're really talking about a tiny, extra-motivated fraction of a much larger group. – Shog9 Aug 5 '14 at 23:36
  • @Shog9 - Haha, okay that makes sense :) – Travis J Aug 5 '14 at 23:39
  • 5
    This does not tell the whole story. On Server Fault we routinely see people post as guest in order to attempt to evade a question ban. Multiples per day appear to fit this profile. At this point I have stopped even attempting to migrate questions from guest users. – Michael Hampton Aug 6 '14 at 0:30
  • 2
    See the note at the end about the recidivism project, @Michael; that particular form of abuse is a bit tricky to stop, but Tim has a Plan... – Shog9 Aug 6 '14 at 3:40
  • I hope it involves getting rid of cookie based accounts entirely. – Michael Hampton Aug 6 '14 at 3:41
  • 1
    Right now, cookie accounts actually have more protection against this sort of abuse than registered accounts do, @Michael. Not enough, but... Shoring that up & making it work for all new accounts is really the only thing that's gonna make a difference. – Shog9 Aug 6 '14 at 3:46
  • i.stack.imgur.com/aXfX2.png – gnat Nov 14 '14 at 6:39

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .