Fwd: [EM] Center for Range Voting Formed

[Adam Tarr]

The following is a message Warren sent to me, and asked me to forward
to the list.

-AT

---------- Forwarded message ----------
From: Warren Smith <wds at euclid.math.temple.edu>
Date: Aug 11, 2005 3:00 PM
Subject: Re: [EM] Center for Range Voting Formed
To: wds at euclid.math.temple.edu, atarr at purdue.edu



>Adam Tarr,  atarr at purdue.edu:
Warren,

Your "strategic Condorcet" strategy is a terrible strategy in most
situations.  Why do you think burying your least favorite frontrunner
would help you?  Even more puzzling, why put your favorite frontrunner
in front of your true favorite?

There are specific, very narrow situations where these sorts of
strategies can be useful.  But  They are absolutely the exception and
not the rule (at least in winning vote Condorcet).  A simple sincere
ordering is more strategically adept as a default strategy, for most
good Condorcet voting methods.  I know if I were voting in an MMV
election or a Beatpath election, I'd much prefer voting that way to
burying the second.

--WDS REPLY:
I did not say it was the best strategy.  I merely claim it is
an obvious strategy, which *sometimes* is best, that many members of the public
will adopt.
If that be so, then Condorcet will prevent third party candidates from winning.

>On the other hand, your Range voting strategies seems pretty spot-on.
So to me, the most accurate comparison I can make using your
simulation is sincere Condorcet against strategic Range.  Condorcet
wins, by your simulation.

WDS REPLY:
Comparing honest-voter condorcet-LR versus strategic-range=stategic-approval,
yes, the former does win, at least in the 2 example election scenarios
with 5 candidates on the CRV site.   However honest-condorcet-LR does
better in one
and worse in the other case shown, versus honest-voter-approval, and
worse in both cases versus honest-voter-range.

Consulting the larger data compilation at
http://math.temple.edu/~wds/homepage/voFdata
we find that honest-voter condorcet-LR is usually (always?) better than
strategic-range=strategic-approval.

However, I do not see why this is "the most accurate comparison."
Seems to me the most valid comparison is honest v honest or
strategic v strategic.   In the hon v hon comparison,
range beats condorcet.  In the str v str comparison range also
beats condorcet, however with the CAVEAT that the condorcet voters in my sim
were not using their optimum voting strategy (which is, as far as I know,
unknown) but instead were using the vote which would have
been strategically optimum for them if it were a Borda vote
(if I recall right) which is known.

(This is stated in my study, no attempt to delude on the issue...)

--I am note sure whether your letter to me was CCd to the EM list.  If
it was, then
please also post this reply on that list.  Thank you.

wds