[EM] Catchpoles RC fails your IIAC

MIKE OSSIPOFF nkklrp at hotmail.com
Wed Jan 23 21:51:38 PST 2002


I'd said:

>Did I say that a voting system must meet the Pareto
>Criterion? If you define voting systems in that way, Sequential
>Pairwise wouldn't be a voting system. Not many would agree with
>that definition. Excuse me, but did I say anything about Pareto?

Markus replied:

It seems that you don't know what "e.g." means.
"E.g." means "exempli gratia" ("for example").

I reply:

Shouldn't that be "exempli gratiae", if we want the "for"
to be expressed?

Ok, so you were giving an example of something that I didn't say.

But what you'd said was irrelevant, because I didn't say that you
were wrong about RC passing your IIAC because RC isn't a voting
system. I merely said that RC isn't a voting system.

That's something that I said last time, and which you seem to have
missed.

******

Markus continued:

You wrote (20 Jan 2002):
>It's reasonable to say that a voting system is a choice method
>that bases its choice on voting. But I also admit that Random Candidate
>could be called a degenerate voting system in which people can
>vote, but their votes are ignored. Rather like the 2000 Presidential
>election.

The problem is that when you have problems understanding a mail or
when you don't agree with someone or when you are dissatisfied with
a mail, then you spam the EM archieves with tons of insulting mails
instead of trying to describe your problems/disagreement/dissatisfaction.

If you had described earlier how you define "voting systems", then I
would have mentioned that David Catchpole has proposed a modification of
Random Candidate where --when a Condorcet winner exists-- this Condorcet
winner is elected with a probability of 2/(N+1) and every other candidate
is elected with a probability of 1/(N+1) and where --when no Condorcet
winner exists-- each candidate is elected with the same probability of
1/N. This modification of Random Candidate is a "voting system" due to
your definition of "voting systems" and meets IIAC.

I reply:

Say there are 2 voters who, being used to Plurality, believe that they
need to move a pre-existing lesser-evil candidate up to 1st place
when the new candidate is added. Before the new candidate is added,
there were 2 candidates, and those lesser-of-2-evils voters had voted
for X. When the new candidate, Z, is addedd, Z is their last choice.
They think Y has a better chance of beating Z, and so,
they move that lesser-evil Y up to 1st place.

Y had previously lost to X by one vote. Y pairbeats Z. Due to how
those 2 voters change their votes when Z is added, Y becomes
BeatsAll winner. Other voters don't change how they vote between
X & Y.

Before Z is added, X pair-beats Y and is BeatsAll winner. At that
time, Y's probability of winning is 1/3. But when we add Z, now
Y's probability of winning is 2/4, because those lesser-of-2-evils
voters have made Y into BeatsAll winner.

Adding the new candidate has increased Y's probability of winning,
given the voting strategy of those lesser-of-2-evils voters.

David Catchpole's RC fails your IIAC.

Shall we send that to the journals too? :-)

In any case, if you can actually find a voting system that passes
your IIAC that's ok too. But if you can find one that could
realistically be proposed that meets your IIAC, that would be
more interesting.

Mike Ossipoff




_________________________________________________________________
Chat with friends online, try MSN Messenger: http://messenger.msn.com



More information about the Election-Methods mailing list