Dave Ketchum wrote: > > Bart is saying: > Starting with: 49 ABC, 2 BAC, and 47 CBA (estimating a near tie). > A and C backers could agree to lock out B by truncating to A and C. > Then each worries about whether they can expect the other to truncate. > > In this case I like better acting on regard for SELF: > Starting with: 49 ABC, 2 BAC, and 47 CBA > Translation: A and C backers each PREFER B over their major enemy. > A and C backers hear something UGLY about B > Selfish response is: 49 ACB, 2 BAC, and 47 CAB > No value in a fragile agreement. Or, if B remains attractive to > some As and/or Cs, why would they join the proposed agreement - or honor > it even if they agreed ahead of time?
Since you have changed the sincere preferences so that B is no longer the sincere Condorcet winner, your example has nothing to do with my point. Although it may well be that in my original example, if some of the A and C voters have a solid enough agreement, they might be even better off using the more extreme strategy of order-reversal rather than simply truncating. This would likely require fewer cooperating voters on either side. > Seems to me the words about expected utility are misleading. A and C > backers obviously rank their preference first. They properly go next for > whichever remaining candidate is their preference among such. > > Also seems to me that "low-utility candidate" gets misused: > ABC says that this voter assigns most utility to A and something > less - perhaps negative - to C. Voter has placed B between - PERHAPS > almost as positive as A; PERHAPS almost as negative as C. No PERHAPS about it, my example had the voters placing B almost as negative as C. Unfortunately the Yahoo site may have destroyed the formatting. Here it is again, with underscores in place of the spaces: _______Liked <-----------------------> Disliked_ 49%___A____________________________________B___C _2%___B______A_________________________________C 49%___C____________________________________B___A > Looking at the initial estimates above, A and C could perhaps be > rated as low-utility with some voters rating A or C as high-utility and > others giving each the opposite rating. B could possibly be rated as > moderate-utility, for noone has assigned B last choice. Well, I could have made B the voters' last choice, but then my example wouldn't ---- Election-methods mailing list - see http://electorama.com/em for list info
