[EM] Reply to DEMOREP on Lord Jenkin's Proposals

Mike Ositoff ntk at netcom.com
Mon Sep 28 17:09:51 PDT 1998

> Greetings,
> Jenkin's Proposal: AV Plus is a hybrid of two separate voting methods. "AV"
> (Alternative Vote) is a system of electing MPs for constituencies, the
> "Plus" is a system of topping-up from a list to achieve something nearer
> proportionality.
> >DEMOREP: This [Alternative Vote] is the infamous so-called instant
> >run-off(IRO) method with its well known defect of being able to NOT
> >choose a compromise candidate.  Example:  45 AC  40 BC  11 CB  4 CA
> >C loses using IRO. B wins.
> Donald: First let me say that this run-off routine was going by the name of
> Alternative Vote long before the Center for Voting and Democracy(CV&D)
> coined the term Instant Run-Off. By the way, allow me to bring you up to
> date, CV&D now calls it Instant Run-Off Voting, with the letters IRV
> instead of IRO. (I know - minor detail) I prefer to call it Choice Run-Off.
>      It is true that with Choice Run-Off as the method, Candidate C will
> lose, but it can be argued that candidate C does not deserve to win.
> Consider that 85 percent of the voters voted for someone else as their
> first preferred choice. The voters of candidate C do have the power to
> decide which of the two front runners will be the winner. The C voters are
> in the position of being the "King Maker".

But it can't be said that the expressed wishes of 85%, on a particular
issue, have been violated, since no such majority prefers any
particular candidate to the CW.

Oh  thanks, the middle voters can be "kingmaker" and make anyone
king except for their candidate, the one who beats each of the
others in pairwise comparisons, the one who'd beat each one of
the others in separate 2-way elections.

> >DEMOREP: However, head to head C beats A 55 to 45 and C beats B 60 to 40.
> >The proper method for a single winner is the Condorcet method that compares
> >each pair of candidates head to head. If one candidate wins all of his/her
> >pairings he/she is a Condorcet winner.   If there is no such candidate, then
> >a tiebreaker can be used.
> Donald: Wow, two different winners, with two different methods, using the
> same example. They both cannot be the real winner. Which one is correct?
> Will the real winner please stand up.

You've said that before; the choice needs to be made based on
what properties are desired. You need to get past the idea
that one method is the only right one.

>      What factor or data was used in one method and not in the other? I see
> now, the reason we have different winners is because the second choices of
> the voters of candidate A and B were used in the Condorcet method and not
> used in the Run-Off method. In other words the second choices of 85 percent
> of the voters were used to help defeat their first and most preferred
> choices. Condorcet was able to fool the voters - it is not nice to fool the

Wrong. In Votes-Against, truncation won't help, and full ranking
won't hurt, unless order-reversal is being used against your
candidate, who is a middle CW.

I suggest that what isn't nice is violating the expressed wishes
of a majority of the voters. It isn't at all clear how you're
claiming that Condorcet fools the voters. One reason why what
you say is garbage is that you don't support your statements,
which allows you to make some really bizarre ones.

> voters - they will react by not making any lower choices in the future.
> When they do stop, it is good bye Condorcet, because Condorcet needs lower
> choices in order to steal the election for the third place candidate. It is
> not to the advantage of the voters to make lower choices in a Condorcet
> method election - while it is to the advantage of the voters to make lower
> choices in a Run-Off type of an election.
>      Because most of the voters may not make lower choices means that
> Condorcet has a flaw, but that is not the only flaw of Condorcet - the
> Circular Tie is another major flaw of the Condorcet method. By your own

I recently answered that circular tie objection of yours, and
you just continue to repeat it.

> research, one out of three Condorcet elections will end up in a circular
> tie. After we have a tie there is nothing we can do to show which candidate
> is the real winner and to show that that candidate has a majority of the
> votes in use. A circular tie breaker is no better than drawing straws.

Nonsense. Votes-Against protects majority rule even when there's
a circular tie, and avoids the lesser-of-2-evils problem. The
LO2E problem, as I said, is the reason why most of us want a
better single-winner method.

> DEMOREP: >I will estimate in advance that the current circa 20 percent
> >minority rule to elect a bare majority of the members of the
> >UK Parliament will go up to about 35 percent minority rule with AV plus.
> Donald: No, 35 is low. I say 50+, if they do not use an artificial
> threshold. About 45+ if they use the same threshold as New Zealand - which
> causes about ten percent of the votes to be wasted. This leaves 90 percent,
> and it follows that 50+ control of 90 is 45+.
> Donald: In closing I would like to question your example. I have seen these
> kind of examples posted many times by Pairwise Guys.
>      Example:  45 AC  40 BC  11 CB  4 CA   Is this one of Mike O's examples?
>      You should be aware that this is a concocted example. Voters would
> never vote like this in a real election in the real world. Notice that the
> 85 voters of candidate A and B have marched lockstep and selected candidate
> C as every one of their second choices. This is unbeliveable - this would
> never happen. Notice that this example gives candidate C a total of 100
> first and second choices. No candidate would ever recieve a choice from
> every single voter. Notice that while candidate C got 100 choices,
> candidate A only received 49 and candidate B only received 51. A very
> slanted example. Notice that while the votes of candidate C is divided into
> two parts the votes of candidates A and B are not divided at all. More
> deception. People do not vote like this. I will show you how people vote:

 Look, Don, do you want me to give you an example with 100 million
voters, voting in various possible ways between 20 candidates?
Simplification is necessary. The fact that your pitifully inadequate
method can be shown to fail so easily with just 3 candidates says
something about the method. It should be obvious that by adding
more variety in how people vote, one could more easily make
anything happen. The results in my examples could be easily
created with more complicated examples. Why, in my examples
to the C voters all vote B over A? Because I'm showing what can
happen even when there's a majority for B over A, and because
under the simplifying assumption of a 1-dimensional policy space,
the C voters will prefer B to A. And because simplicity is desirable,
and in no way makes the example less valid. I could make the same
thing happen with more kinds of voters & more candidates.

Besides, to demonstrate a violation of a criterion, it's only
necessary to show one instance of that violation. That's all
it takes to show noncompliance. Same thing whenever one wants
to show that a problem can happen. Voting in my examples isn't
implausible, though it's simplified.

>      Do not let the Pairwise Guys influence you. They have an agenda. Their
> game plan is to get some single seat method installed that will allow them,
> by hook or by crook, to win seats for their third party candidates in
> single seat districts electing lawmakers. Their method of choice is
> Condorcet - a good hook.

I'm not interested in any 3rd party. I don't belong to any
party, 3rd or otherwise. I don't have an agenda to help a 3rd
party. My agenda is merely to make it possible for every voter
to not have to vote anyone else over his favorite.

After hearing all the blather from Don & the like, you must
understand why I claim that discussion of rank-counts is hopeless.
This is what you'll encounter when you try to explain a good
count rule to any large group. That's why I suggest that you
instead propose Approval, to avoid the endless & tangled
debate between rank-count rules. Approval lets everyone vote
for their favorite, never makes anyone vote a less-liked alternative
over their favorite. (unlike IRO & Margins & FPP).


> Regards,
> Donald Davison
>      \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
>      ///                 N E W    D E M O C R A C Y                ///
>      \\\ Home of Citizen's Democracy   http://www.mich.com/~donald \\\
>      /////////////////////////////////////////////////////////////////

More information about the Election-Methods mailing list