Jobst,
> ... If voter i is honest and puts R(X,i)=U(X,i) ... It worries me that U(X,i) may be much larger than C(i) most of the time, i.e. that the virtual money may be infinitesimal in value compared with typical utilities. And if I understand the method, it is doesn't work to normalize the utilities. ... > The constant term E/S makes the whole thing sum up to zero so > that no > voting money is produced or destroyed, only redistributed. Q.E.D. > > The thing most astonishing to me is that although the actual > value of > deltaC(i) is independent of i's ratings (as long as the winner > is not > changed), the expected value of this adjustment does more or > less > depend *only* on these ratings. > This is indeed astonishing from a slightly different point of view: if I rate the winner well below the benchmark lottery, my action tends to cause other voters' accounts to lower, but does not cause mine to increase, yet the total is constant. Magic! Forest ---- Election-Methods mailing list - see http://electorama.com/em for list info
