For context, I mostly use sites of "medium" size and traffic, like unix.SE - it's not clear whether that makes a difference to this question.

From experience, it seems that the effect of good answers "bubbling up" by collecting upvotes over time does not always work as good as I would hope.

If there are some quick initial answers in short time, and they are voted up, it seems very hard for a later answer to catch up with the votes of the initial answers.
The problem seems to be that the "bubble up" mechanism basically works, but is too slow - it seems that even after multiple years, there is only a little progress in many cases.

It would be very interesting to have some kind of numbers, supporting or rejecting this - or giving any partial hint.

Examples are not of much use, I think;

Are there any existing statistics in this area, or attempts to analyze how effective the mechanism of "bubbling up" of answers by votes works?

  • 3
    It would be interesting to know but I don't see how any sort of automated statistics could be gathered. Seeing late answers clearly worse than earlier answers (or straight out wrong) isn't exactly unusual.
    – PeterJ
    Commented Nov 9, 2014 at 5:52
  • I am tempted to post two answers here in this question to demonstrate that the good one will bubble up in a few days. On a serious note, it does work, good posts always bubble up. Commented Nov 9, 2014 at 7:09
  • 1
    @InfiniteRecursion We'd need some "quick initial" or "fastest gun" type answers first, to bubble up through... Commented Nov 9, 2014 at 7:30
  • Even late answers bubble up, let's wait for someone to provide some hard data here. Commented Nov 9, 2014 at 8:10
  • @InfiniteRecursion I mainly use medium size/medium traffic sites like unix.SE - I wonder if it may depend on the traffic on the site. It could even make a big difference; I would think the visit count per time on old questions makes a big difference. Commented Nov 9, 2014 at 8:17
  • Bubbling up can be accelerated when accompanied by downvotes on incorrect or less detailed answers. I think the value of some downvoting can often be overlooked.
    – PolyGeo
    Commented Nov 9, 2014 at 10:07
  • @PolyGeo I think the reputation cost of downvoting answers makes it a very unpopular option.
    – Louis
    Commented Nov 9, 2014 at 10:28
  • @PolyGeo Downvoting incorrect answers helps, but I do not think a less detailed answer should be downvoted - that would be a big "policy change", regarding what a downvote means. Commented Nov 9, 2014 at 10:43
  • 1
    I have an answer that gives an up-to-date solution instead of a 3 years old outdated answer. Still the outdated one has 4 times more votes. Moreover it is the accepted answer. I hope that people doesn't follow that one, just because my bubble is slow...
    – totymedli
    Commented Nov 9, 2014 at 14:17
  • One problem here: stats can't really be produced, only anecdotal evidence. There's no way to objectively rate a good answer, beyond votes. Commented Nov 9, 2014 at 17:54
  • @AndrewBarber It may not be possible to get clean statistical data, but I can imagine that some approximation could be possible, that needs some interpretation. For example, it would be an interesting difference between "The percentage of answers that bubbled up, of all late answers at least two years old, is:" 3% versus 20%. Actually, 15% may indicate it works well. (A rough random sampling of how many old late answers are "good" would help for context). Commented Nov 9, 2014 at 18:46
  • @PeterJ Good thought....that's why there is a 'Late Answers' review queue.
    – MTL
    Commented Nov 9, 2014 at 19:15
  • What about looking on partially bubbled up answers - like when it got 10 upvotes over one year, and the top answer has 40 upvotes? Finding these could give a hint whether it may work "too slow" in some sense. Commented Nov 9, 2014 at 19:47
  • I think this somewhat depends on how many answers are there in the question. If it attracts lots of meh answers, this may make it harder for those who use "active" tab to get to answers worthy of voting... which probably makes another stats request, how many site regulars view questions using "active" tab
    – gnat
    Commented Nov 9, 2014 at 21:22


You must log in to answer this question.

Browse other questions tagged .