This question is motivated by this example. In it, several users (including myself) posted answers, none of which were complete and satisfying. However, they did receive varying amounts of upvotes.
To stimulate a better answer, a bounty was posted. This did stimulate a nice new answer. That answer got some upvotes, but not as many as some of the previously posted answers. Whoever posted the bounty did not award it to anyone, and the time approached where it would be auto-awarded.
Specifically in order to award the bounty to the new answer, one or more users gave my answer downvotes. I'm not worried about the points (neither the bounty nor the downvotes). However this is a distortion of what votes are supposed to mean. The comments to my answer suggest that this distortion presented a moral dilemma to the downvoter(s).
This leads to two related questions:
- Is there some way to measure how often this happens, i.e. downvotes within the final hours of an expiring bounty that affects which answer receives the bounty?
- Is there some change to the bounty award mechanism that could solve or mitigate this problem? Perhaps award bounties proportional to total votes, rather than to just the highest-rated answer.