How many questions do we get from users recently blocked at SO, how many of these are closed / deleted?

Question

Recent efforts in SO close vote queue make me think that quite a lot of users started hitting the question ban there.

There is long known "tradition" of attempts to circumvent such bans by posting at inappropriate sites:

The core issue here seems to be the abusive "I can't ask any more, so I'll ask my question on a similar site, in the hope it is migrated"...

I wonder how much impact could this have on Programmers. To understand this, I would like to find out how many questions per day are posted by users that were banned at SO less than a month ago, and how many of these questions end up closed / deleted?

It would be also helpful to learn similar statistics for questions from users who:

• aren't yet banned, but are already warned about approaching a ban at SO
• are rate-limited to ask questions at SO by anti-recidivism system

---

Basically I am trying to figure whether it's something that can be dealt with in our regular ways, or it is worth looking for a dedicated approach to address (like, say, system triggering automated flag for mod attention when post from such a user gets close vote / flag).

Related:

• (MSO) Let's burn down the close queue!

  • (MSE) Block migration if user is suspended/question-blocked at destination

  • (MSE) Warn new users when they ask a question after a previous question is closed, downvoted, or deleted

  • (MSE) What is the reasoning behind limiting "recidivists" to post one question per week?

  • (MSO) Why won't the system allow me to ask questions for several days?

• (MSE) Stats on how attempts to circumvent SO question blocks impact other sites

• (Meta P.SE) Provide 10k users more close votes

Answer 1

While running a test on Stack Overflow, there were some questions about whether we were driving traffic to Programmers, so I decided to rerun some of the numbers in Shog’s original answer and compare the 30 days prior to the test to the testing period.

Prior to the Stack Overflow Test

First, let's look at the 30 days of activity here on Programmers prior to the start of the test, the time period was February 15, 2016 to March 16, 2016. During this time there were:

• 1452 questions asked here, from
• 1283 distinct account, of which
• 1143 have/has account on Stack Overflow
• 298 of these users hit some sort of block on Stack Overflow in the 30 days.

From those 298 users:

• 354 questions were posted
• 200 were closed
• 244 were deleted
• 233 scored < 0
• 124 had at least one answer
• 74 scored > 0

For comparison:

• 701 questions were closed in total
• 843 were deleted
• 801 scored < 0
• 596 had at least one answer
• 434 scored > 0

Looking into the details, I also checked the IP addresses associated with these posts to see how many came from blocked address on Stack Overflow. Of the IP Addresses, 288 hit some sort of block on Stack Overflow. From those addresses:

• 339 questions were posted here
• 185 were closed
• 216 were deleted
• 208 scored < 0
• 128 had at least one answer
• 81 scored > 0

During the Test Period (March 16, 2016 - April 13, 2016)

The numbers over the test were:

• 1388 questions asked, from
• 1202 distinct account, of which
• 1087 have/has account on Stack Overflow
• 248 of these users hit some sort of block on Stack Overflow

From those 248 users:

• 307 questions were posted
• 151 were closed
• 158 were deleted
• 193 scored < 0
• 119 had at least one answer
• 63 scored > 0

By comparison for the rest of the questions asked during this time:

• 543 questions were closed in total
• 578 were deleted
• 729 scored < 0
• 607 had at least one answer
• 418 scored > 0

Of the total IP Addresses, 298 hit some sort of block on Stack Overflow. From those addresses:

• 355 questions were posted here
• 164 were closed
• 172 were deleted
• 211 scored < 0
• 135 had at least one answer
• 81 scored > 0

Conclusions

In the 30 days prior to the test, approximately 23% of the users asking question on Programmers had hit a block on SO, but during the test this dropped to 20%. Blocked users asked approximately 24% of questions prior to the test, and 22% of questions during it. While there was a slight increase in the number of questions from blocked IP Addresses, it went from 23% to 25%, I'm not seeing a huge indication that the test on Stack Overflow drove traffic to Programmers.

Answer 2

I'm working on something related to this right now, so it's worthwhile to pull the data. More on that in a bit, but first...

• 1380 questions asked here in the last 30 days, from
• 1219 distinct accounts, of which
• 1100 have/had accounts on Stack Overflow
• 184 of whom hit some kind of block on Stack Overflow during the past 60 days (this includes not just the quality-ban, but also spam blocks and 4 types of rate-limits).

From those 184 users...

• 227 questions were posted during the past 30 days.
• 152 were closed
• 135 were deleted
• 161 score < 0
• 62 received at least one answer
• 25 score > 0

For comparison,

• 771 questions were closed in total during the past 30 days,
• 636 were deleted,
• 728 score < 0,
• 523 were answered, and
• 323 score > 0

But... that's not the whole story. See, posting here in response to a block on Stack Overflow is a real desperation move - and there are less extreme tactics to try first. Like sockpuppets...

I grabbed the IP addresses associated with those 1380 posts and looked to see how many had been associated with blocked questions on Stack Overflow in the past 60 days. From those 255 IPs,

• 267 questions were posted here during the past 30 days,
• 168 of which were closed,
• 154 of which were deleted,
• 176 of which score < 0,
• 87 of which were answered, and
• 35 of which score > 0

Now, folks share IPs, and hop between networks. IP-based banning isn't a great long-term solution for anything... But if your goal is to slow down folks searching desperately for any text box into which they can type, then it ain't bad. Which is why we've been working on a system for temporarily classifying networks according to the number of bad posts coming from them. I took the currently-live data for this and compared it to the questions posted here during the past 2 days:

• 9 questions from potentially "bad" networks
• 4 of these closed
• 5 of these deleted
• 5 scoring < 0
• 2 answered
• 3 upvoted

Total stats for the same time period: 166,47,37,58,30,58.

Conclusions

Thus far, blocks on SO are a pretty weak predictor for quality. While there is a massive overlap between users on both sites, blocked users don't account for a very large chunk of the total here, and the false-positive rate is high.

Network-based tracking is even worse... But there's a caveat: we're only tracking networks with problematic posts on Stack Overflow right now, so all of those posts (and users) having issues on Programmers itself don't count. Right now, the network quality signal is used to feed Triage review on Stack Overflow... Potentially, turning on tracking for Programmers would allow it to do the same thing here, and save a considerable amount of time identifying these questions on both sites.