I wrote this before I wrote my most recent posting, but it didn't post. So I'm posting it again:
As I'd said in an earlier post, using the names A , Middle and B, instead of Green, Dem and Repub: If it looks as if Repub is bigger than Green, so that Repub voters don't approve Dem, then Green voters will approve Dem. And if that isn't enough to make Dem win, then Repub has a majority of the voters, and it doesn't make any difference what anyone else does. On the following day, when I discussed a similar situation, looking at better-than-expectation strategy, I said that it won't work to say that the Repub is very far away, and that will make the Dem approve the Green--because Repub is virtually identical to Dem. Saying that was an implied acknowledgement that, if Repub were sufficiently distant, that could give Dem voters reason to vote for the Green, when using expectation-maximizing strategy. Then I said that Repub is _not_ distant from Dem. So, Lomax's example doesn't show anything that I had denied or contradicted. ...except that, as I said above: If it looks as if Repub is bigger than Green, so that Repub voters don't approve Dem, then Green voters will approve Dem. And if that isn't enough to make Dem win, then Repub has a majority of the voters, and it doesn't make any difference what anyone else does. That argument suggests that, with 3 candidates, Middle is favored, and hir voters needn't approve an extreme. If that seems to contradict Lomax's example, then I call your attention to the fact that Lomax was talking only about a Dem-preferrer who is barely a Dem preferrer, and who is near the boundary of being a Green. Anyway, the following day, when discussing the situation via better-than-expectation strategy, I acknowledged that if the Repub were sufficiently distant, that could make Dems approve Green--except that Repub is considerably closer to Dem than Green is. Lomax doesn't agree about the sameness of the Republocrats, and that is ok. Mike Ossipoff ---- Election-Methods mailing list - see http://electorama.com/em for list info
