There has been a fair amount of controversy over the recent change to the default sorting of answers, going from a vote- and time-based system, to a purely vote-based system that randomizes the ordering of questions that have the same number of votes. There have been at least a couple of discussions of potential problems that this introduces elsewhere (here and here, for example). I propose an alternative that is a compromise of the old and new.

I suggest that we borrow a principle from scoring in cycling, in which riders who arrive in the same pack get the same time. The way this would work is that, for purposes of sorting, answers are rounded to the nearest minute (or two minute interval -- seconds/120 * 120 using integer math). We follow the previous scheme whereby questions are first sorted by votes, then by time, but answers with the same sort time and vote values are ordered randomly.

This will retain the same basic historical view of the data. It will be easy to see which answers differ significantly in time, discouraging cutting and pasting answers as an obvious vote grabbing ploy. Users who answer quickly (and correctly) are still rewarded. There will no longer be an incentive to downvote a higher rated answer simply to get into the same "vote bracket" -- at least if you're outside the "pack" window.

At the same time, this removes some of the FGITW problem in that you'll have a minute or so (on average) to compose an answer, even after the "more answers are available" message appears, to complete your answer and still be "first." The pack window could be adjusted over time based on historical data to optimize it so that it really captures the same sort of principle as used in cycling.

FWIW, I have no problem removing what others may perceive as a bias towards early answers per se, since I think that in all but a handful of cases, good eventually triumphs early. However, the current system allows someone to simply cut/paste my answer and have their ansewr be presented in a way that it looks like I copied them, not the reverse -- at least to an inattentive reader who simply considers the order of presentation. Since it is natural to assume that the ordering is chronological, this isn't an obscure risk. My suggestion at least mitigates this risk since if they arrive outside the "pack", it will be clear to even a casual user what has happened.

  • Interesting suggestion. But it kicks out the KISS principle. There are still randomly sorted answers. Honestly, pressing F5 and everything messes up (again and again) is hurting my head. Aug 27 '09 at 12:25
  • 4
    +1 For a cool-headed suggestion. Aug 27 '09 at 12:33
  • 3
    To be clear, my preference is to just go back to the previous method. However, given Jeff Atwood's response, I don't think this is likely to happen. I'm just trying to find some common ground.
    – tvanfosson
    Aug 27 '09 at 12:34

While this is a good compromise, we're compromising with Reddit, which should have little to no bearing on how this site is run. The order should be very, very simple:

order by Score desc, LastActivityDate asc.

Anyway, two minutes is just too long. Questions within one minute of each other should be grouped together and randomized, if we're going to randomize at all.

Frankly, I'm confused as to what has happened since June:

We agree with Paul Graham: in general, there tends to be a correlation between the length of a response and its quality. We've also observed this pattern on Stack Overflow. It isn't always true (TL;DR), but it's a verifiable pattern. Either be the first and quickest, or be the best and most comprehensive!

As a side effect, randomizing the answers within a minute takes some of the competitiveness and fun out of answering these questions. Obviously, the current ordering destroys it completely, so grouping them within a minute is a good compromise.

  • 1
    +1 for reddit comment - I read the reddit post in question and it looked to me like a bunch of whiny elitist douchebags who were upset that their mediocre answers weren't accepted and/or the quickest. In summary, it was "all voters are retarded" or "the system sucks" with basically no justification whatsoever about why "it sucks". My favorite comment was "That's exactly what StackOverflow is. An RPG for junior devs pretending to have clue." Technically I'm a junior dev but I'd like to think I've offered some helpful solutions. Also, tell that to Eric Lippert, Jon Skeet, Scott Hanselman, et al.
    – John Rasch
    Aug 27 '09 at 15:27
  • 1
    @Eric: The second desc should be omitted.
    – mmx
    Aug 27 '09 at 22:01
  • being the first and quickest still confers benefits, even in random sort order. If you're the first, you have the most time to get upvotes. Oct 2 '09 at 10:32

I'm wondering how that fits into a popular question where many people answer with similar answers at the same time?

Wondering if a setup where the owner of the question can only see the answers for the first few minutes and the rest of the users only see theirs (if they answered) would be viable? Then after a few minutes the gates open and all the answers are visible.

  • 2
    With hundreds of duplicates! Aug 27 '09 at 12:42
  • If a question has answers I tend to comment, not answer -- unless there aren't any good answers. If I couldn't see the answers on a question, I would probably answer more questions than I already do. I don't really see this as an improvement. In some ways it even detracts from the character of the site as being collaboratively edited.
    – tvanfosson
    Aug 27 '09 at 12:50
  • @tvanfosson good point. Aug 27 '09 at 12:52

You must log in to answer this question.

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