[EM] simple methods' merit limit

[MIKE OSSIPOFF]

I'm still on the list, because I haven't had time to do the replies that
I said I was going to do before quitting the list.

By "simple method", I mean a method whose only input is one Plurality
vote, approval set, ranking, or ratings-set from each voter, in
one balloting. Maybe there could be a better wording, along the
lines of saying that the balloting consists only of one opportunity for the 
voter
to indicate how favored each candidate  shall be on  the ballot (as by 
voting for him
or not voting for him in Plurality or Approval, or by rank position, or by a 
points score).
So I'm not saying that I've defined "simple method" in the best way 
possible, but
you know what I mean by the term.

By that defininition, SSD is a simple method, but ordinary Runoff
isn't.

Blake earlier mentioned that, with the winning-votes methods, in
a zero-info election, there can sometimes be strategic incentive to
insincerely rank some candidates equally. I'd replied that that
strategy only pays off in certain special natural circular ties.
It's a temptation to improve one's expectation, but it isn't
a serious strategy _need_, in which strategy is needed to avoid a
majority rule violation or to protect a sincere CW.

That's why I don't consider that problem to be nearly as serious
as the extreme majority rule violations of methods like IRV &
Ranked-Pairs(margins), and the resulting strategy need of those
methods.

Still, I don't deny that it's a problem and an embarrassment.
In some political elections, and in many other polls, where there
acceptable & unacceptable alternatives, I'd rank all the acceptable
altetrnatives in 1st place in wv. Even though I needn't do that
to prevent them from being defeated in a majority rule violation,
one still wants to do one's best to avoid the election of an
unacceptable alternative. If the only big candidate-merit difference is
between  the acceptables   & the unacceptables.

So, with the wv methods, there can be some genuine incentive to vote
in a way that's something like Approval.

What that shows is that, even with the best simple methods,
it's very difficult to improve on Approval--and they can't improve much on 
Approval.

Some good non-simple methods have been proposed here, and I don't
doubt that many or most of them are better than any simple method.
Though some of those non-simple methods are too complicated to propose
publicly, it can't be said that no non-simple method can be publicly
acceptable, since ordinary Runoff is such a method. For instance, the
method in which rankings (and maybe other ballotings too) are collected
in a 1st balloting, the results published, and a 2nd balloting
held by Approval, is uncomplicated and is more obvious-sounding than
the good simple rank-methods.

By the way, that's an improvement over Voter's Choice, if it's
acceptable to hold more than one balloting. But in our polls that
I helped conduct, I felt that it would be better to not ask people
to vote more than once. Of course if that method is used instead of
Voter's Choice, one would want to allow all 3 kinds of ballotings,
and count the results of all the proposed methods using those ballotings,
and post those results before the 2nd balloting by Approval is held.

So maybe the simpler non-simple methods should be considered as
public proposals.

If the voters insist on only one simple balloting, then I still
prefer Approval, CR, SSD, CSSD, or maybe Ranked-Pairs(wv). Or even PC, since
I'm willing to settle for more modest reforms than some others are.

Mike Ossipoff




_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp