[EM] Eric Gorr and David Gamble's Running Debate:

Donald Davison donald at mich.com
Thu Aug 14 01:57:29 PDT 2003


Eric Gorr and David Gamble's Running Debate:

Eric Gorr wrote:  "For example, why is it the right thing for A to win in
this case:   40 A     35 C > B     30 B   when clearly > 60% of the voters
preferred B over A?"

Donald here:  Dear Eric, you are in error.  What your example shows is that
40 voters prefer A over B or C and that 30 Voters prefer B over A or C and
that 35 Voters prefer C over A or B.

The source of your error is that you are assuming Approval Voting as the
standard to evaluate these ballots.  You have eye-balled the ballots and
you falsely assumed that you had the right to convert the second rankings
of 35 > B into full votes.  That's a No No, those are not B votes, they are
second choices for B, there is a difference.

When you compare two or more methods you cannot use one as a standard.  For
example, if we were to compare apples vs oranges vs bananas, we cannot use
the banana as a standard and say the apples and oranges fail the test
because they are less like bananas than bananas.  Likewise it is not proper
for you to say that Irving fails because its results are less like Approval
than Approval (Approval would be the banana).

If I were to say that Approval and Condorcet fail because they will not
always elect the `Irving Winner', you would not accept that as a standard,
but you are doing the same thing, you are saying that Irving fails because
it does not always elect the Approval winner, and likewise the Condorcet
people are always saying that Irving fails because it does not always elect
the `Condorcet winner' as if this so called Condorcet winner was some sort
of Gold Standard.  The Condorcet Winner concept is merely hype.

The only single-seat method that could be used as a standard is Exhaustive
Ballot because it will be the voters who will decide the next balloting -
not you.

Suppose a club that conducts an election at the end of each year to elect a
president for the next year.  One vote per member (unranked).  Unranked
means no lower choices, which means that you will not be able to claim that
the voters really wanted to elect your turkey candidate, the third or lower
running candidate.  The voters will decide which candidate they want to
elect.

Whenever there is a candidate with a majority, he is the new president, the
election is over.  If no majority candidate, a new balloting takes place in
thirty minutes.  No candidate is eliminated, but any candidate is allowed
to withdraw.  Voters are free to vote the same or vote a different
candidate on the next ballot.

Now, how do you think the balloting will go?  I think it is possible that a
candidate might drop out, but which candidate?  Not one of the two leading
candidates.  If any candidate is going drop out, it will be the last
candidate, more so if he is far behind.

If any voters change their votes it will not be the voters of the two
leading candidates, those voters can smell victory, they are not going to
give it up.  Those voters will be looking to the other voters to change
their votes.  One of the two leading candidates is going to win, if the
last place voters want any voice in the final decision they will need to
change their vote to one of the leading candidates.  Some of the voters of
the last place candidate will crack first and change their votes - you will
not be able to stop them, with them using the secret ballot.


D Gamble wrote:  "The example given by Eric is not that realistic ( it is
not likely you would see such a vote set in a real election)."

Eric Gorr:  "Is it really not realistic?"    [8/6/03]

Donald:  Yes, it is really not realistic, `Let me count the ways.'

First - It is unrealistic for you to assume that all the C voters will
march lockstep and vote all together for candidate B as their second
choice.

Second - It is unrealistic for you to assume that none of the B voters will
make a lower choice in light of the fact that all the C voters ranked their
candidate B as second choice.

Third - It is unrealistic for you to assume that none of the A voters will
make a lower choice in light of the fact that the election is very close.
Both the high and low candidates are within five percent of the center
candidate.  This election is too close for any voter not to make a second
choice if he wants to be sure he has a choice in the final decision between
the last two candidates.

A more realistic example would be the following:

    30 AC        5 CA       5 BA
    10 AB       30 CB      25 BC
   ------      ------     ------
    40 A        35 C       30 B

Using Irving, candidate B is still eliminated, but 25 votes are transferred
to candidate C which makes him the winner.

One of your two favorite candidates won, so what's your problem.


On the other hand, making unrealistic examples is par for the course on
this list, so you are among friends (friends in deception), your example is
no worst than any made by others who like to bad mouth Irving.

Speaking of the worst example, when I first joined this list (about six
years ago) I went to its web site to see what was there.  One of the things
I seen was the worst of the worst election example ever concocted.  It was
by Rob L (not Rob Richie) as follows:

   Rob L: "Suppose the following election results:

               45 A    45 B   10 C

           There is no majority, therefore candidate C wins."

           (end of example)

Donald:  Simple as that.  There were no lower choices included for any of
the vote sums.  This example told me much about the mindset of most of the
people on this list.  They feel that if there is no majority then that is a
good time for them to use some phony argument in order to have their
`Turkey Candidate' win the election.  They will argue that with no majority
the compromise candidate must be declared the winner (the compromise
candidate is their turkey candidate by definition).

We could call this single-seat election method the Compromise Method.  On a
scale of one to ten, with one being the best, I would rate it a twelve, two
below Plurality.


It is not the compromise candidate that must win, it is the last place
voters that must compromise and change their votes.  There is no reason why
the tail should be able to wag the dog.


Donald,





More information about the Election-Methods mailing list