I've been doing my bit on the close queue but also thinking about the larger problem and had a thought.

How many Questions get to the point where they have 4 close votes but then at that point the final result is to leave them open?

If that number is small enough as a proportion of the total close queue then that would suggest to me that the 5th vote is not actually adding a lot of value to the accuracy of the process.

Reducing the number of close votes needed to 4 would make the close process 20% more efficient, which should then cause the numbers in the queue to start trending towards 0 as at the moment the growth in entries in the queue is less than 20%.

The Burn down experiment has certainly made some progress but that progress seems to be slowing and I can't see figures for the total close queue so we don't even know whether we are making any overall progress or just illusionary progress.

Considering the effect of that 5th close vote can actually be directly measured through analysis of the database we can actually make an informed decision as to whether it adds value to the process.

  • 7
    Interesting. How many questions with three close votes end up not being closed? Because then you could make do with three votes. Or maybe two. Or maybe just one. – Frédéric Hamidi Mar 26 '14 at 11:46
  • 1
    Fuzzying the number massively increased reviews for like 2 weeks, but the review rate is back to what it was, I'd be incredibly surprised if there is less than 120k (i.e the size pre-fuzzying) in the queue after we burn through this last 20k. I think reducing the number of votes required to 4 is a pretty sensible solution. – OGHaza Mar 26 '14 at 11:59
  • @FrédéricHamidi that's a good question even though I think you probably meant it ironically. Potentially you could reduce it below four but the only way to know if to see the figures. Hard figures would allow a proper objective analysis of how much accuracy is gained by each extra Close Vote required. – Tim B Mar 26 '14 at 12:57
  • 2
    @Tim, I did mean it ironically. I believe reducing the number of close votes will mechanically lead to more questions ending up closed although they should remain open. IMHO a "bad" question remaining open is better than a "good" question being closed. Primum non nocere, as they say. – Frédéric Hamidi Mar 26 '14 at 13:12
  • @FrédéricHamidi you believe. And your belief may be right. Belief is not a replacement for solid facts though and is a poor driver of policy. If someone does know a way to extract these figures we can get those solid facts and an evidence-based decision can be made. If the figures show large numbers being opened with 4 close votes then clearly your belief has been confirmed. – Tim B Mar 26 '14 at 13:16
  • 1
    If on the other hand (and I think this is certainly possible, maybe even likely) in fact the number opened is tiny then reducing the threshold might make sense. After all following your argument I could say why not require 6, 7 or 8 close votes. After all there might be questions getting closed at the moment that should be left open. No number is perfect, just some are better than others. – Tim B Mar 26 '14 at 13:17
  • @Tim, absolutely. My understanding is that the number 5 has been chosen as a good middle-ground, quite possibly after data analysis, but a senior member would have to chime in to confirm that. However, IMHO five votes should be retained even if there is a small number of questions with 4 votes being deemed worthy enough to remain open. – Frédéric Hamidi Mar 26 '14 at 13:20
  • @FrédéricHamidi Especially since when the breakdown by off-topic close reason was posted, the vast majority were "need to demonstrate a minimal understanding" which most people were using as "didn't show enough effort". If people downvoted low-effort questions instead of trying to close them, there would be significantly fewer questions in the queue, allowing the stuff that needs to be closed to actually get closed. – Troyen Mar 26 '14 at 20:16

You can see a snapshot of these results from shog9's initial assessment of the close-vote burndown:

Of tasks with 4 close votes

 Completed: Close Completed: DoNotClose Completed: Edit 
 ---------------- --------------------- --------------- 
 4900             135                   14              

So roughly 3% of questions that make it to 4 close votes are ultimately left open. Unfortunately, the data was not posted for 3+ or 2+ questions.

| improve this answer | |
  • Thanks, that's useful for the first data point. As you say it's a shame he didn't post the data for 3+ or 2+ :( – Tim B Mar 27 '14 at 9:41

Not sure how everybody will react on that, but what about weighing the number of required close votes depending on user status? For example, users with <10K rep - need 5 votes, mods >10K rep need 3 votes (5/3 of a vote each). A combination thereof can be calculated and rounded down, so 2 mods + 2 regular users works out to 4 close votes required.

If you think giving more weight to mods is unfair, consider that those rep are hard earned and granted as a measure of trust by all other users, regardless of their reputation level. Full disclosure - yes, I am one of those >10K users. :)

| improve this answer | |
  • It's not a bad idea, it doesn't really answer the question though so I can't upvote it :( – Tim B Mar 27 '14 at 9:39

You must log in to answer this question.

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