Kevin--
No matter how unconventional a balloting format you suggest, or how bizarre and
unproposable a method of determining who
wins, this definition will apply and mean something. It also works fine for all
methods that have been used or proposed:
Definition of voting x over y:
You're voting x over y if switching x and y on your ballot is more likely to
change the winner from x to y than from y to x.
[end of definition of voting x over y]
In all actually proposed or used methods, it's always obvious that one is a lot
more likely than the other. But you might propose
a method in which that isn't so. You might propose a method in which it's
unknowable whether, by my definition, you're voting
x over y or voting y over x. But such a method could hardly be called a voting
system. The whole point of voting is that, at least probably,
you're helping someone against someone else.
If you like, "more likely" could be defined by, or replaced by, "consistent
with more configurations of other voters' ballots", but such
a change isn't necessary.
If you propose a voting system in which, by my definition, it's unknowable
whether you're voting x over y, or voting y over x, that's ok,
as long as Rob Richie doesn't spend enough to get it enacted somewhere.
Mike Ossipoff
----
Election-Methods mailing list - see http://electorama.com/em for list info
