[EM] what's wrong with random ballot (Hay voting)

[Warren Smith]

I agree with Ossipoff.

That is, Hay voting is really a parameterized class of methods (parameter P).
When P=1/2 you get the original method.  When P-->0+ you get a log-based method.
When P-->1-  you get random ballot.

There are reasons to believe P=1- is actually the best member of the family.
If so, then Hay voting is actually worse as a voting method, than random ballot.
Which is well known to be a horribly bad voting method.

So Hay voting is not a good voting method.  However, from the point of view of a theoretician,
it has some neat properties.  Namely, strategic vote = honest vote.  Actually
ranodm ballot already had that property, but Hay goes further in that it actually
causes the voter to reveal her honest utility values.   Also, Hay with P=1/2 can
guarantee that the candidate with the greatest summed utility for all of society
is the one with the largest chance of being elected.  I cannot think of any other
voting method that can say that (aside from, say, Clarke-Tideman-Tullock but it in some
sense isn't really a "voting method" since you pay money as your vote).

Warren D Smith
http://rangevoting.org