Participation & SARC

[MIKE OSSIPOFF]

>But, under the reasonable conditions where the best strategy in
>Approval never includes skipping, voters in Approval can never
>worsen their outcome compared to what it would be if they
>hadn't voted.
>----
>D--
>
>3 AB
>2 B
>1 C

But the A voters wouldn't have gotten a better result by
not voting in that example, and so they didn't worsen their
outcome by voting.

Those voters misjudged. The ones who needed B didn't vote for
him, and the ones who didn't need him did vote for him.

The only reason for the C voters to have not voted for B is if
they thought they were a majority, when actually they were 16%.
When actually C is beaten by A.

If the A voters feel sure that A will get more votes than C,
so they don't help B, then it's reasonable that the C voters will
know that too, and will know to help B.

Maybe your example could be explained if everyone believed that
C would get more votes than A.

Usually the difference between having a majority and having
fewer voters than the opposite side is a pretty big misjudgement.

Of course misjudgement is possible, and Condorcet avoids
those strategy decisions. But if you're comparing Approval to
rank methods other than Condorcet, they'll sometimes fail in
worse ways that I've described. As I've said, better a solid
reliable handtool than an automatic power tool that will go
haywire and act in unintended ways.

>
>Using Approval-- B wins.  Using common sense- A wins with a first choice
>majority.

Condorcet returns a tie between A & B. If Plurality were the
tiebreaker then A would win. If ties are solved "by lots", then
A & B each have half a chance of winning. A looks like the
right winner.

>
>Making additional choices in Approval can obviously hurt earlier choices.

I don't consider it as voting against your favorite, with respect
to the 2nd choice. You merely aren't voting a preference between
those two, because you've chosen to instead express your preferences
for A & B over C. Maybe you felt that C's undesirableness
justifies that. Maybe you felt that A won't be a frontrunner, and
so voting B vs C is the way that you can influence the election
in your favor.

With IRV, transfering a vote from your favorite to your 2nd
choice would certainly be likely to help your 2nd choice beat
your favorite. The only thing that saves your favorite from that
is the fact that IRV eliminates your favorite before it lets you
help your 2nd choice. And though your 2nd choice rank rating
won't hurt your (eliminated) favorite, it often won't help your
2nd choice either, often to your regret, if someone publishes
the pairwise results after the election, revealing that you
could have made the 2nd choice win instead of your last choice
by voting him in 1st place.

Mike Ossipoff




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