[EM] FWD - Mike Ossipoff comments on the Tannenwald letter

[Donald E Davison]

  ------------ Forwarded Letter -------------
From: MIKE OSSIPOFF <nkklrp at hotmail.com>
To: donald at mich.com
Cc: seppley at alumni.caltech.edu, nkklrp at hotmail.com
Subject: RWE & Bucklin strategy. 3 candidates.
Date: Tue, 08 Jun 1999 18:59:11 PDT

Don--

Two points regarding your reply to Tannenwald:

1. RWE (Runoffs Without Eliminations) is incomparably better
than IRV, with any number of candidates, including 3.

RWE _doesn't_ give the same result as IRV in 3-candidate
elections. Of course, with any 2 methods, there could be some
particular examples where the 2 methods give the same result.

In 3-candidate elections where the middle candidate's voters
don't vote a 2nd choice, and where Middle gets eliminated first,
and Middle is CW (Condorcet winner, the candidate who'd pairwise
beat each of the others under sincere voting), RWE elects the
CW, if at least one of the other candidates' supporters ranks
the CW 2nd. An example:

40  25  35
  A   B   C
  B       B


IRV starts by eliminating B, and, since the C voters' 2nd choice
has been taken away, A wins.

RWE declares B the lowest candidate, but his voters don't
list a next choice to transfer to. The next lowest candidate
is C. He transfers to B, who then has 60 votes, and is declared
winner by RWE's stopping rule, since B now has votes
greater than half the number of voters--B now has a majority.

***

2. I agree that RWE is considerably better than Bucklin, but
I disagree with Tannenwald's statement that, if Bucklin is
used, people should be somehow discouraged from doing
strategic truncation. Truncation is the defensive strategy in
Bucklin. One should avoid extending one's ranking farther than
necessary. Avoid extending your ranking farther than what
you estimate to be the best candidate that, with your help,
can get a majority. Hoag & Hallett called truncation a failure
of Bucklin; actually it's just the correct defensive strategy
for that method. If people truncated too much, that would
cause them to lose that lesser-evil compromise with which they
could beat something worse. Maybe people over-truncated because
they overestimated the winnability of their favorite. Maybe
people unused to a method other than Plurality naturally tended
to vote as in Plurality (and were inclined to vote sincerely
in that method, as is often the case in municipal elections).

The same truncation strategy is also the defensive strategy of
RWE. Don't let CVD convince you that that's a bad thing.
Though, in my opinion, there are better methods, the ones that
have been called the VA methods, RWE improves hugely over IRV.
It's worth examining the difference to show why that is:

When I suggested RWE to Rob Richie, of CVD, as a very acceptable
replacement for IRV, Richie, if I remember correctly, didn't
like it because inclusion of a lower choice in one's ranking
could defeat a higher choice. Sure it can, hence the defensive
strategy I described. But does that make RWE worse than IRV?

Note, please, that that lower choice in your ballot would never
get a vote from you unless everyone whom you've ranked higher
has transferred due to being lowest candidate. In IRV, everyone
you've ranked above that lower choice would be eliminated.

So the way IRV protects your upper choices from defeat by
votes you give to your lower choices is by eliminating your
upper choices when they transfer. IRV protects your upper choices
by eliminating them. Thanks a lot, CVD.

"Sire, the peasants have no shelter from the rain. They're
attempting to shelter themselves with discarded newspapers and
cardboard boxes. O won't you do something to end this deplorable
situation?!"

"Yes, slay the peasants".

***
When I pointed out this obvious fallacy in CVD's objection to
RWE, there was no reply.

***

Feel free to forward this to Tannenwald &/or to EM, & any
other list.

***

Mike Ossipoff
_______________________________________________________________
Get Free Email and Do More On The Web. Visit http://www.msn.com