[EM] 3 questions for Markus

[MIKE OSSIPOFF]

Markus said:

you wrote (24 Sep 2000):
 > I wouldn't want to try to guess whether there's a merit
 > difference between Plurality & Tideman(m). I'm not
 > interested in miniscule merit differences that can't be
 > meaningfully measured.

I reply:

First, the next day, I wrote again and said that I do acknowledge
that Tideman(m) is probably better than Plurality.

Markus said:

Do you see a merit difference that can be meaningfully
measured between those methods that are mentioned in your
website at http://home.pacbell.net/paielli/voting?

I reply:

Most certainly there's a measurable merit difference between the
following two sets of methods that are listed in the criteria
compliance table:

1. Approval & Condorcet
2. IRV & Plurality


Now, if you're asking about merit differences within those 2
subsets, that's another matter. Approval & Condorcet are both
excellent. Approval has exclusive criteria compliances. So does
Condorcet. It's my _personal & individual_ feeling that some of
the criteria that only Condorcet (& BeatpathWinner) complies with
are so ambitious that I personally prefer Condorcet. But I don't
claim that Condorcet is objectively better than Approval. They offer
different guarantees that are difficult to compare.

As for IRV vs Plurality, that's another tough one. The claimed
advantages of IRV over Plurality seem unimpressive under scrutiny.
There are conditions that will likely be common, maybe typical, in
which IRV reliably does worse than Plurality (Support for candidates
tapers gradually away from the voter median).

On the other hand, IRV does better than Plurality in SU simulations,
I've heard. So I don't know.

I can say pretty much for sure that Tideman(m) is better than IRV,
because Tideman(m) beats IRV by Monotonicity & Condorcet's Criterion,
but I haven't heard of a criterion by which IRV beats Tideman(m).

As for Tideman(wv) vs SSD, that's another difficult one. It seems
to me that, from what I've heard so far, Tideman(wv) can boast more
advantages over SSD than vice-versa. Maybe simulations could show
advantages the other way, maybe not. So Tideman(wv) looks a little
better right now. I emphasize that there's no important difference
between their merit. I recommend choosing between those 2 based only
on which is easier to define & demonstrate with examples. I don't
know if one is easier to define than the other, but Tideman(wv) seems
easier to demonstrate with an example. So Tideman(wv) is what I suggest
for rank-balloting proposal. For public elections.

Mike Ossipoff

_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.

Share information about yourself, create your own public profile at 
http://profiles.msn.com.