On Thu, Jul 16, 2015 at 2:25 PM, Sean Hunt <scsh...@csclub.uwaterloo.ca> wrote: > On Thu, Jul 16, 2015 at 2:22 PM, omd <c.ome...@gmail.com> wrote: >> (d) If the valid options are ordered lists of preferences, the >> outcome is decided using instant-runoff voting. In case >> multiple valid preferences tie for the lowest number of >> votes at any stage, the vote collector CAN and must, in the >> announcement of the decision's resolution, select one such >> preference to eliminate; if, for N > 1, all eir possible >> choices in the next N stages would result in the same set of >> preferences being eliminated, e need not specify the order >> of elimination. > > Is that "for some N > 1" or "for all N > 1"?
I think it's pretty obviously "for some" in context. > Also, in IRV, the tied > preferences are all eliminated, rather than breaking the tie. The > exception is if all remaining options are tied. This isn't universal. According to http://wiki.electorama.com/wiki/Instant-runoff_voting: -- ALL: Eliminate all tied candidates at once. Good for weak candidates (with less than 5% of votes), but can lead to strategic nominations, which cause IRV implementations using this method to not be spoiler proof -- I think vote collector abuse is vaguely more interesting than strategic voting. Because I'm dumb, I had to write a script to come up with a concrete outcome where the difference matters: Candidates: A, B, C Votes: C, C, A, B>A In the first stage, A and B are tied for last, at one first choice each (but C has more). If we eliminate them both at once, C obviously wins. But if we eliminate B, we're left with C, C, A, A, which is a tie that could be broken in A's favor. In other words, B is a spoiler candidate.
