Some correction. I wrote: > In the sincere situation > x A>>D>B>C > y B>>D>C>A > z C>>D>A>B > the winner should be one of A,B,C, with probability x/n, y/n, z/n, > respectively, since D is not approved by anyone. > > DFC (Democratic Fair Choice) gives this result!
The last is wrong, of course. DFC is not so extreme but still gives D some probability, depending on the relative sizes of x,y,z: Without loss of generality, assume that x is larger than y,z. Then either y>z and DFC will elect A with prob. x/n and D with prob. (y+z)/n. Or y<z and DFC will elect A with prob. x/n, C with prob. z/n, and D with prob. y/n. Sorry for messing up the example, Jobst ---- Election-methods mailing list - see http://electorama.com/em for list info
