tea.mathoverflow.net - Discussion Feed (Automatic MO Bumps)

Jonas Meyer comments on "Automatic MO Bumps" (8906)

Jonas Meyer — Thu, 09 Sep 2010 17:19:53 -0700

Cam, yes, that's the one. I didn't notice that, but your conjecture seems highly plausible after taking a little time to browse the Unanswered list. That list doesn't tell me how many times questions have been bumped, but of the questions whose last modifier is MO, a disproportionate number have at least one answer (based on my unscientific and limited scan).

Joseph O'Rourke comments on "Automatic MO Bumps" (8901)

Joseph O'Rourke — Thu, 09 Sep 2010 13:02:31 -0700

As Willie Wong says, if a question has no answers with positive votes, then it will be recycled automatically by the bot. Often the original poser of the question is no longer participating in MO. Here is one solution, but it only works if you made an answer yourself. On this question

http://mathoverflow.net/questions/6888/computing-correlation-between-time-series-with-missing-data/26556#26556

I changed my 0-vote answer to community wiki so no one would think I was trying to gain rep, and asked for someone -- anyone -- to vote it up. Someone did, and so now that question will not be recycled. A very limited solution! But if you don't mind voting for a random answer, you can do the same: vote up +1 any answer, and it will no longer be recycled endlessly.

This is not well-known, apparently. Should probably be explained in the FAQ somewhere...

Cam McLeman comments on "Automatic MO Bumps" (8890)

Cam McLeman — Thu, 09 Sep 2010 09:56:15 -0700

Interesting. Jonas, is it this question that you're referring to:

http://mathoverflow.net/questions/15775/does-equality-of-the-operator-norm-and-the-cb-norm-for-every-bimodule-map-over-a

If so, I might conjecture that the posts getting bumped disproportionately often are those with at least one answer, but has no answer with a positive number of votes. Both yours and the one I mention have answers with 0 votes, whereas many questions with no answers are not being bumped.

My sample size for this experiment/conjecture is about 3 or 4, though.

Jonas Meyer comments on "Automatic MO Bumps" (8885)

Jonas Meyer — Thu, 09 Sep 2010 07:05:14 -0700

I agree with Cam that some questions get bumped disproportionately, and in fact this led me to ask a similar question that got no response months ago:

http://tea.mathoverflow.net/discussion/472/how-does-the-random-bumping-of-unanswered-questions-work/#Item_1

Incidentally, the question I referred to there has been bumped 2 or 3 more times since that July update.

WillieWong comments on "Automatic MO Bumps" (8882)

WillieWong — Thu, 09 Sep 2010 06:13:43 -0700

Someone mentioned to me that questions can only get bumped if it has no accepted answers, and has no answers whatsoever with positive number of votes.

Cam McLeman comments on "Automatic MO Bumps" (8881)

Cam McLeman — Thu, 09 Sep 2010 06:03:28 -0700

This is probably documented somewhere that I don't know about, but I was wondering what the algorithm was for deciding when the MathOverflow user auto-bumps certain questions to the top.

Not that it's at all a bad question, but I feel like we've seen http://mathoverflow.net/questions/33533/name-this-pro-p-group a disproportionately large number of times, even compared to questions with similar number of votes, favorites, and answers.