3

I noticed that:

  • answers on this site usually get sorted descendingly according to their score,
  • in general, answers are either upvoted or left untouched rather than downvoted, and
  • I (and, presumably, most of us) do not have the time to read ALL the answers thoroughly if there are many of them, but do have time to scroll through the topmost.

This leads to the situation that in the presence of several answers, the topmost answers are likely to increase their score and stay on the top, regardless of whether there are better answers below.

Is there a technical means to avoid this bias? Surely, you can sort by active, oldest, votes, but those sorting modes would simply remove one bias and introduce a different bias instead. Moreover, the default sorting mode is "votes". My suggestions would be to introduce some randomness into the orders. I'm not sure how much randomization is necessary (fully random or taking the votes orders and moving a random answer to the top): feel free to comment.

EDIT: Following the comment of Massimo Ortolano, if someone could migrate to the general meta discussion, I'd not be against.

4
  • Answers on all sites are ordered in the same way. This has long been discussed on the main meta, e.g., meta.stackexchange.com/search?q=answer+order
    – Massimo Ortolano
    Jul 6, 2017 at 21:36
  • There already are options for active, oldest and votes . Are you saying you do not want people to be able default to votes?
    – StrongBad
    Jul 6, 2017 at 21:36
  • One simple way around this problem is that the answer at the very top is actually the one selected by the question-asker as the accepted answer. Hopefully if no one else reads all the answers, the person who had the initial question does.
    – Bryan Krause Mod
    Jul 7, 2017 at 6:27
  • On the main meta your question would be probably closed as duplicate of the Fastest Gun in the West Problem The highest voted answer proposes the implementation of the sorting criterion described in Federico's answer. Thus, SE is well aware of the issue, but solving it is probably not in their priorities or they think that a new sorting criterion might cause other issues.
    – Massimo Ortolano
    Jul 10, 2017 at 18:58

1 Answer 1

Reset to default
5

It has been suggested that a better sorting criterion is the lower bound of the Wilson score confidence interval for a Bernoulli parameter. Apart from the sexy-sounding statistical name, it's the system used to sort comments in Reddit and other sites. It is intended to fix exactly this kind of bias.

You must log in to answer this question.

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