Saying it on meta instead of in comments...

Awarding a bounty by criteria such as vote counts, random selection, or alphabetic ordering of answerers' user names, is better done silently without announcing it. Anything vote-based can distort the voting, refusing to evaluate the solutions on their merits distorts the signal from the bounty, and some answerers would have not necessarily participated if the selection method were stated at the beginning.

The above is inspired by several recent bounty threads.

| |
  • 10
    $\begingroup$ Answerers who answer a question for its bounty when, unbeknowst to them, such a selection method will be used, might rightfully feel cheated. This leads to a practical conclusion which is the opposite of what you suggest: when some erratic bounty awarding rules such as those you cite will be applied, the fact should be advertised as soon as possible. $\endgroup$ – Did Jan 29 '13 at 8:51
  • 2
    $\begingroup$ I would not put 'vote counts' in with 'random selection' or 'alphabetic ordering of usernames'. Since a bounty cannot be freely divvied up among more than one answerer (as far as I know...) it seems to me that, should the question poser feel roughly the same about several different answers, s/he might use the number of votes as a selection method. I wouldn't classify such an approach as 'capricious' or 'erratic'. I know that if you haven't accepted a bountied answer after 7 days on mathoverflow, then the highest voted answer (with at least 2 votes) is auto-accepted and awarded half the bounty. $\endgroup$ – Benjamin Dickman Jan 29 '13 at 10:23
  • $\begingroup$ One problem is that those decision rules do not depend on reading the answers. $\endgroup$ – zyx Jan 29 '13 at 13:09
  • $\begingroup$ @Did, of course the worst case is to exclusively use an erratic method, and announce that fact between the posting of the first answer and the time of awarding the points. However, expecting OPs to advertise what they will do with answers would inhibit freedom on Q/A web sites, and I have therefore phrased the question more from the point of view of OP and readers' interests than those of the answer authors. $\endgroup$ – zyx Jan 30 '13 at 3:52
  • $\begingroup$ What are some of the recent threads mentioned at the end of the question? $\endgroup$ – Did Jan 30 '13 at 7:12
  • $\begingroup$ The question works equally well without the sentence at the end. I do not think it loses anything by being discussed hypothetically, and there is some benefit, in my opinion, in not indicating particular cases and users. $\endgroup$ – zyx Jan 30 '13 at 7:59
  • $\begingroup$ Then why mention it? By the way, what is the question we are supposed to discuss, exactly? At present, a point of view is stated, and a comment tries to back it up appealing to the all encompassing concept of freedom on Q/A web sites, and nothing more. NARQ, or did I miss something? $\endgroup$ – Did Jan 30 '13 at 10:03
  • $\begingroup$ There is always the close vote for anyone who would like to send the dreaded NARQ signal and thinks four other people might agree, despite the eight upvotes. Faster and easier than playing rhetorical games in the comments. $\endgroup$ – zyx Jan 30 '13 at 11:57
  • 5
    $\begingroup$ What is the question we are supposed to discuss, exactly? I would like to know before deciding to vote to close or not. $\endgroup$ – Did Jan 30 '13 at 15:51
  • $\begingroup$ There are many things that one or another person would "like to know". I, for one, would "like to know" what, if anything, your purpose in this and a hundred other antagonistic and seemingly pointless comments on MSE has been, but sometimes one has to make educated guesses and go with that. Life is all about incomplete information. $\endgroup$ – zyx Jan 30 '13 at 18:46
  • $\begingroup$ I do not know all the rules, but is there a reason that it is not possible to split a bounty in the case of multiple good answers or several partial answers that make a full one? $\endgroup$ – RussH Jan 31 '13 at 5:36
  • $\begingroup$ The bounty escalation is described here: blog.stackoverflow.com/2011/09/bounty-reasons-and-post-notices $\endgroup$ – zyx Jan 31 '13 at 10:41
  • $\begingroup$ @RussH: It is not possible. There is a related feature request: meta.math.stackexchange.com/questions/4107/… $\endgroup$ – Jonas Meyer Jan 31 '13 at 16:18
  • 1
    $\begingroup$ @zyx Why do I get the strange feeling that you try to avoid my question by changing the subject? My purpose is to get an answer to the question I asked (wow, amazing...). Declaring that my question is "pointless" won't do, unless you explain its pointlessness. So, please store your knives away, reread what I wrote, and try to stick to the subject, right? Q: What is the question we are supposed to discuss, exactly? A: $____$. (And next time, please use @.) $\endgroup$ – Did Jan 31 '13 at 19:47

Browse other questions tagged .