[EM] Why Margins isn't as democratic or ethical as

Blake Cretney bcretney at postmark.net
Mon Feb 7 15:18:41 PST 2000

Rob Lanphier wrote:

> Another message from Mike...
> ---------- Forwarded message ----------
> Date: Mon, 07 Feb 2000 07:59:46 GMT
> From: MIKE OSSIPOFF <nkklrp at hotmail.com>
> To: robla at eskimo.com
> Cc: nkklrp at hotmail.com
> Subject: Why Margins isn't as democratic or ethical as votes-against
> Rob--
> This is my letter about why votes-against makes more sense
> than margins, completely aside from the matter of
> criterion compliance. It never occurred to me to put it this
> way when I was on the list. Would you forward this to EM for me?

I think I remember Mike presenting a very similar argument

> Here's why I claim that Blake Cretney's Margins method
> isn't as good as Schulze, or any other genuine Condorcet
> version. (The translations of Condorcet's words indicate that
> Condorcet judged the strength of a defeat by the number of people
> voting in favor of that defeat).

Could someone please send me that translation, or any translation
believed decisive on the issue of Condorcet's beliefs.  Everything I
have seen by Condorcet argued based on probability, and therefore
suggests margins.  I would be surprised to see him presenting the kind
of argument Mike gives here.  I would be interested in seeing any
published paper arguing for winning-votes over margins.
> We'd like it if there is one candidate who beats each one of
> the others. We of course only drop a defeat because we have
> to, because they're in conflict for choosing a winner.
> Dropping a defeat means overruling a pairwise count result
> voted by the people, and so it isn't done lightly. We want
> to minimize the number of voters whom we overrule.
> Since Condorcet's own wording of his proposal on which most
> of our Condorcet interpretations are based is an iterative process
> in which we iteratively drop the weakest defeat, I'll talk in
> terms of that procedure. It clarifies & dramatizes the issue
> of which way to judge defeats, when we consider which defeat to
> drop.
> If A beats B, and we drop the defeat A>B, then we're overruling
> the  voters who won that pairwise defeat result. If we don't
> drop A>B, we _aren't_ overruling the voters who opposed that
> result by voting B over A. We aren't overruling them because they
> were already overruled when the votes were counted. They lost.

That could be used to justify any method.

> So, if we want to minimize the number of voters whom we overrule
> by dropping a defeat, then we must minimize the number of people
> who voted in favor of the defeat that we drop. Drop the defeat
> that has fewest people voting for it. That's obviously what it
> means to overrule as few voters as possible.
> ***
> To put it differently, Blake would treat those 2 sets of votes
> the same, count them with the same weight: The voters who
> voted A over B, and the voters who voted B over A. Even though
> A beat B, because more voted A over B than vice-versa, and the
> A>B voters are the ones who won, and the B>A voters are the
> losers.
> ***
> Ethically & democratically, it makes much more sense to
> drop the defeat for whom the fewest people voted.

Speaking to Rob Lanphier, do you agree with this argument, or are you
merely forwarding it?  It all seems to hinge on belief in majority
rule with a very specific and personal definition.    BTW, have you
read any of my long-running debate with Mike on the margins vs.
winning-votes subject?

I'm not interested in getting into another debate with Mike about
this subject on this list.  If anyone wants to know what we would say,
they can check the archive because we already said it.

I am, however, planning to post an essay on the web that will review
my position on this subject.  This will be easier to read than the
archive, but of course will only present Mike's arguments in order to
be refuted.  I plan to ignore the particular argument presented here
unless other people seem to find it convincing.

Blake Cretney

More information about the Election-Methods mailing list