Show us the ballots, mikeo

[MIKE OSSIPOFF]

Don said:

Greetings list members,

Mikeo wrote:

From: "MIKE OSSIPOFF" <nkklrp at h...>
To: election-methods-list at e...
Subject: Re: [EM] Best Method In Use
Date: Tue, 17 Dec 2002 06:15:45 +0000

I only know of 3 single-winner methods that are used in public
political elections, and all 3 are used in national elections:

1. Plurality (aka 1-vote Plurality or 1st Past The Post (FPTP) ).
2. Top-2 Runoff (aka Runoff)
3. IRVf (aka the Alternative Vote, Preferential Voting,
Majority Preferential Voting, Hare)

Donald here: Actually, there is a number four Mikeo, it's a variant of
Approval Voting, it's called `Plurality-at-Large for single-seat'
(Correction: Approval Voting is a variant of `Plurality-at-Large for
single-seat'). It is used to elect the president of a council in many
American cities, big and small. This variant of Approval Voting does not
elect the choice of the people, it elects the candidate with the most pity
votes.

I reply:

Don has forgotten to define the method that he's naming. Though it's
anyone's guess what he means, it sounds as if he's talking about
Plurality, aka First Past The Post (FPTP). Plurality was in my
list of methods that are in use.

Mikeo: "Of those, I claim that Runoff is the best."

Donald: Runoff is not the best, anyone who is not a fool should be able to
see that. Runoff requires a follow up election and it is only valid for
three candidates in the original election.

I reply:

Yes, with 3 candidates, Runoff works like IRV. When there are more
candidates, that's when IRV is likely to go wrong in a way that
Runoff won't.

Don continued:

The only thing that can be said
in its favor is that the voters know who the final two candidates are to be
in the follow up election, but that piece of information could be
considered as a possibility when ranking choices in an Irving election,
without a second trip to the polls. Besides, Irving is valid for more than
three candidates.

I reply:

When there are more candidates, that's where IRV fails in a spectacular
way, a way in which Runoff won't fail. As I said, that's when
IRV can fail to elect a CW who is the voted favorite of more people
than any other candidate.

Mikeo: "One advantage that Runoff has over IRV is that, with Runoff, at
least a CW can't lose if s/he comes in 1st or 2nd in the 1st balloting,
whereas in IRV a CW can lose even if s/he's the favorite of by far the most
people. The scenario in which that happens isn't at all implausible. It
requires only that candidates' favoriteness support taper gradually for
candidates farther away from that Plurality winner CW. Then, the
eliminations at the extremes, and the centerward transfers, will reach
candidates adjacent to that CW before they reach the CW, and the CW will
likely be eliminated.
IRV's poor social utility scores in simulations resulted from the
frequent elimination of median candidates, probably happening largely as
described above."

Donald: You should be ashamed of yourself. This is merely more of the
same gobbledegook that you have been handing out for the last five years.

I reply:

This list has been here for about 8 years. Wasn't I handing out
gobbledegook during the list's 1st 3 years? :-)

Don continued:

It is clear you haven't learned much in five years. People are not going
to think you are super smart just because they don't understand what the
hell it is you are talking about.

I reply:

Is that your way of saying that you want an example? Because my
description of the problem scenario was rather complete. All that's
lacking is an example.

I've posted these examples before on EM, and so I guess I should
take time to post them again for you.

I've simplified this example by leaving not listing B voters' lower
choices. The relevant fact is B's elimination.

A set of IRV ballots:

60: ABCDE
70: BACDE
100: B
83: DECBA
75: EDCBA

A & E get eliminated, and their transfers leave C with the fewest votes.
C gets eliminated though C is CW and though C is the voted favorite
of more people than is any other candidate.

Want a more extreme example?:

50: ABCDE
51: BACDE
100: C
52: DECBA
49: EDCBA

This time, though C is CW, and though C is the voted favorite of
about twice as many voters as any other candidate is, C gets eliminated
again by IRV. As I said, this kind of thing is the reason why IRV
did so poorly in social utility simulations.

These examples aren't implausible. It's reasonable that the CW
might be favorite of the most, and that favoriteness support would
taper gradually away from that CW.

These diverse examples show how rubust a scenario it is.

Don continued:

Donald: This is not the first time you have bragged about Approval Voting
being used in the UN and some societies. By now, I should think that you
would be able to show us some real ballot results from one of these
elections, which will tell us more about Approval Voting than all your
gobbledegook.

I reply:

You're asking me for U.N. ballots, and ballots from all those
professional societies that use Approval. Is there some reason why
you believe that they're more accessible to me than to you? You want
them? Go get them. It's been pointed out that the long use of Approval
in those applications indicates satisfaction with it. When you obtain
the ballots, exactly what will you look for in them?

Don continues:

Approval Voting is just fine for any meaningless election, but as the
election becomes more political we will find more voters bullet voting

I reply:

But you said that voters will be giving out all those pity votes. And
now you're saying that more and more voters will bullet vote. Which is
it? :-)

I've answered that objection many times on this list, most of the
times when you repeated it as you are now. That's why I've sometimes
lost patience and expressed some characterizations. Ok, here we go
for the Nth time:

If a Nader-preferrer now strategically votes for the Democrat instead
of Nader, because he believes that he needs the Democrat to beat the
Republican, then why should he stop voting for the Democrat just because he 
can vote also for Nader in Approval? That doesn't make any
sense.

If anything, some people who now vote for Nader, so that Nader's support 
will be counted, will, under Approval, vote also add a
strategic vote for the Dem. If their only reason for not voting for the
Dem in Plurality is so that they can show support for Nader.

I'm not
saying that I recommend voting Dem in Approval, though. The Dem won't
deserve a vote in Approval anymore than he does now. Yes, Don, I'll
bullet vote in Approval when it's Nader, Dem, & Repub. But all those
Nader preferrers who now vote Dem will continue to give the Dem a vote
in Approval. And some unknown number of those who now vote for Nader
will also vote for Dem in Approval.

But then I'd bullet vote, with that lineup, with Condorcet too, for the same 
reason.

With IRV? I wouldn't vote.

Don continues:

and
therein lies the trump card that Irving holds over Approval Voting. While
there is no danger for the voter to make choices in an Irving election

I reply:

...unless you count the danger that your last choice will win because
your traveling vote hasn't reached him yet when he needs it, resulting
in the election of your last choice.

IRV lets you express all your preferences...and may or may not count
them.

The difference is that Approval reliably counts every pairwise preference 
that you vote, each one that you consider important enough
to be one of those that you vote, though it isn't possible to vote them
all.

In Approval you decide which of your preferences will be counted.
In IRV, IRV decides that for you, idiosyncratically.

I said all this in a recent message.

Don continued:

, but
if most of the voters do bullet vote, Irving can survive and still function
as Irving, but Approval cannot function as Approval when faced with bullet
voting by most of the public

I reply:

Now quite clear what you mean there. If everyone bullet votes, then
IRV, Approval, & Condorcet all choose the same candidate.

Don continued:

, it will turn into a variant of Irving (poetic
justice).

I reply:

If everyone bullet votes, the result will be like Plurality with all
those methods. But that doesn't mean that IRV & Approval have
been made into Plurality. In Plurality the voter is only allowed to
vote for 1. In Approval, the only voter who votes only for 1 is the
voter who only wants to vote for 1. That's very different from Plurality.

As I've said before, with Approval, for the first time, everyone
would feel free to vote fully for his favorite. That can't always
be said for IRV, with which many will sometimes strategically need
to bury their favorite.

Mike Ossipoff



_________________________________________________________________
Add photos to your e-mail with MSN 8. Get 3 months FREE*. 
http://join.msn.com/?page=features/featuredemail&xAPID=42&PS=47575&PI=7324&DI=7474&SU= 
http://www.hotmail.msn.com/cgi-bin/getmsg&HL=1216hotmailtaglines_addphotos_3mf

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em