[Election-Methods] YN model - simple voting model in which range optimal, others not

[Abd ul-Rahman Lomax]

At 12:17 PM 3/27/2008, Dave Ketchum wrote:
>Ok, I give up on poking at this one.
>
>While the stated votes may be possible, I do not accept them as being of
>enough expectability to be useful in comparison among the election systems.

Dave, you've been reading the Election Methods list for quite some 
time. Election methods are compared on a theoretical basis, all the 
time, based on constructed scenarios that show *possible* behavior. 
For an example, it's commonly asserted that Approval Voting fails the 
Majority Criterion. Why? Well, it is possible that one candidate gets 
a majority and does not win, because another candidate gets a 
*larger* majority.

Now, how likely is this? Consider a two-party system. It is, by 
definition, rare that the winner isn't from one of the two major 
parties. Approval Voting would be introduced into this context 
because it can allow voters who support a third party candidate to 
vote, as well, for a major candidate and thus not waste their vote, 
reducing the "spoiler effect," and possibly vote-splitting effects.

In this context, how likely is it that two candidates gain a 
majority. Imagine that the electorate is, say, 45% D, 45% R, and 10% 
G. That's a large third party.  For two candidates to gain a 
majority, almost certainly what we'd have to see is a significant 
number voters voting for both the R and the D. Sure, it will happen, 
but because there will also be a significant number of G voters who 
don't vote for either an A or a B, it's quite unlikely that two 
candidates will gain a majority, and, indeed, it remains quite 
possible that no candidate will. No voting method can guarantee a 
majority winner except by forcing voters to make a choice, say on a 
ranked ballot, as in Australia, where full ranking is required.

It was not claimed that this would be a common scenario, only that it 
was possible, just as it is possible to have more than one candidate 
gain a majority in Approval, and it is possible for 2/3 of voters to 
vote *against* a candidate and that candidate wins under IRV. These 
studies show something about the possible behavior of election 
systems, and, in fact, Plurality is known to be ridiculously bad; but 
that is only under relatively rare conditions, in *most* elections 
plurality works well enough, producing the same result as, say, IRV 
or other more advanced methods would. It's the exceptions, though, 
that are worrisome.