Condorcet language: largest loss?

[Mike Ossipoff]

Subjects in this letter:

1. Name of that method
2. Bad Example (you wrote it)
3. Wordings for Condorcet

1. Name of that method:

I've seen the name "Simpson-Kramer method" applied to something
that sounded like the method you describe, where votes-against 
in victories as well as defeats are looked at. Bruce has seen that
definition, and can tell me if I've misinterpreted what Simpson &
Kramer meant, but in the meantime, I'll call it "Simpson-Kramer".


2. Bad Example:

That _is_ the bad-example, the one you gave in your posting.
When Dole is the biggest of the 3, and the Dole & Clinton voters
truncate, Clinton & Nader will tie, because both will have only the
Dole voters against them in their biggest defeat.

Dole can't win by truncation in Simpson-Kramer, so that abovementioned
tie is the only problem in truncation.

But I claim that tie is undesirable. For one thing, if the Clinton
voters want Clinton to win instead of Nader (when, as you said, the
Plurality tie-breaker would elect Nader), they mustn't truncate. But
if it's known that they won't truncate, and that many of them like Dole
2nd, then the Dole voters have it very nicely set up for them to win by
order-reversal.

Besides, Simpson-Kramer is indecisive, having to use a tie-breaker
in a big public election. It doesn't look good to have to depend on a
tie-breaker, and there's something funky about that dependable tie.

Plus, when the tie-breaker elects Nader, that's unfair to the Dole voters
who didn't truncate, if there are some who didn't, and it's unfair to the
Clinton voters. Some say "It serves the Dole voters right--they didn't
support the compromise so they lost it". No, some of the Dole voters
probably would support the compromise, though maybe not enough. 
And the Clinton voters too are part of that majority who prefer
Clinton to Nader, and they didn't fail to vote Clinton over Nader.
So Simpson-Kramer is acting somewhat unfairly when it electes Nader
in that double-truncation example.


3. Wording for Condorcet's method:

Yes, that's a problem. I've used brief but not really exact language,
and exact but wordy language, and combinations of them. For something
brief, how about:

The winner is the alternative with fewest votes against it in a
defeat.

But I always like to add:

In other words, determine, for each alternative, which alternative that
beats it ils ranked over it by the most voters. The number of voters ranking
that other alternative over it is the measure of how beaten it is. The
winner is the alternative least beaten by that measure.

For the brief wording, another one I like is:

The winner is the alternative whose worse defeat is the least, as
measured by how many voters rank the defeating alternative over it.


Mike


  



--