Talk:Random-Approval Voting
Horrible Clone problems?
The voter writes the name of a candidate she consents on a standardized sheet of paper. She does this for each candidate she consents; one name per standardized sheet such that if there are n number of candidates she consents, then there are n sheets of paper she writes on.
She place all of the sheets, which she has written candidate names on, into a hopper. All other voters place their sheets into the same hopper. Then, one of the sheets is randomly chosen from the hopper. Whichever candidate occurs on this randomly chosen sheet is the winner of the election.
Suppose there are 3 options for movies to watch with two of your friends: Alice wants to see They Live. Bob wants to see Mars Attacks. You like Police Academy. (Bear with me. Extreme examples are a way to show things clearly). You also like the other Police Academy movies: II, III, IV, V, VI, VII. By mentioning those you increase the likeliness that you get an option you want.
The more candidates a party runs the higher the likeliness that one from the party wins. This problem exists with Borda, Random Candidate and this method (but not with Random Ballot with only one lottery ticket per voter). I consider that problem to be harmless if the voters and the candidates are the same group.
So try this: Instead of directly voting on the issue, vote for people in the voter group you find agreeable with one lottery ticket for each different candidate you like. The winner of that lottery decides.
This could be used for Borda, too. The voters vote on ALL voters via Borda. The winner of that decides on the issue. (Yeah, I know Borda has not only horrible clone probs, but also horrible burying probs). Counting with big groups will be a bit sucky considering that the number of voters is directly linked to the number of candidates. Given that I usually don't trust representatives but want to vote on issues this modification could be modified a bit: Every voter gives one (no more, no less) suggestion on the issue. "What Alice/Bob/somebody else already suggested" is an okay suggestion, too. Then the vote on the suggestions for the issue is taken via the Borda count. For ergonomic reasons the "what Alice/Bob/somebody else suggested" stuff is left out of the ballots. But the ballots get counted as if it was still there. This modification could be used for all methods that have horrible teaming/crowding problems if for some insane reason you can't use something totally different. --R.H. 09:11, 13 September 2006 (PDT)
- Instant version: Ranked ballots. Replace all options with clone sets. The number of virtual clones an option is replaced with is equal to the number of first preferences received (which also means an option with no first place rankings gets deleted which means failing Condorcet). After that you count the info from the modified ballots by the Borda Count or whatever other method with horrible teaming/crowding problems. --R.H. 11:32, 27 October 2006 (PDT)