Tobin: Margins & Standards

[Mike Ositoff]

For some reason, now, when I try to send a message that isn't
a reply, I don't have a "to" field. I wrote EM's address in the
Cc line, and I got a returned copy that had been sent to
election-methods at netcom.com  I don't know if the copy arrived,
so I'm forwarding my returned copy. Forwarding seems to work.
I must somehow these problems with the mailer or my Internet 
accont.

First I want to say that I've sent some replies intended as
_individual_ replies. I indicated to my mailer that the replies
were not to be sent to all recipients, just to the person who
wrote the (EM) messages. Then I noticed that my messages were
posted here, contrary to my instructions to my mailer.

Hugh was talking about the Margins method, also called Young's
method: The winner is the alternative that would be BeatsAll winner
by ignoring fewest individual pairwise preference votes.

Of course that rule sounds good, and could be regarded as its
own standard. Nothing wrong with that. But it conflicts with other
standards that are important to us & many others, including the
standard that led us to want better single winner methods: Getting
rid of the lesser-of-2-evils problem. As I always say, it's up
to anyone what they want from a method. I don't claim that a standard
is "right" because lots of people consider it important. I do claim
that it's important for that reason. A standard doesn't need anything
to make it "right", because it's an individual subjective feelinga
about what one wants. But a standard is important in these discussions
if it's held by us or most of us. And by electoral reformers in general,
and if failures of that standard are complained about by voters and
if their voting is affected by that failure.

So, for those reasons, I consider majority rule, getting rid of
the lesser-of-2-evils problem, and (which is the same thing) getting
rid of the need for defensive strategy to be very important
standards.

I bring this up here because the other day I said that the
Margins method, Young's method, has some gross failures. I was
referring to its failures with respect to the abovementioned
widely-=held standards.

Sure, Hugh, you certainly aren't wrong to say that choosing the
alternative that would be BeatsAll winner by ignoring the fewest
individual pairwise preferences is what you consider the important
standard. I merely say that that standard is in conflict with the
other standards I've named in this letter, which I, & many of
us consider important.

If anyone asks me to, I'll substantiate that claim. I won't
go much into it here, for brevity. I'll just say that Young's
subtractions lose information about majorities--not conducive to
carrying out the wishes of majorities. And that the other standards
I named are closely related to majority rule. And that the lesser-of
2-evils voter is someone who insists on voting _against_ someone
at all costs, even if it means abandoning his favorite. Therefore,
preserving & counting actual votes-agaist counts is a way of
counting what is so important to that voter, while still letting
him vote his favorite in 1st place. Of course that information, too,
is lost in Young's subtractions.

As for the standard of ignoring fewest individual pairwise
preferences, who said it's ok to ignore any preferences? That's
actually a rather high-handed implication of that standard. 
Condorcet(EM), for instance doesn't ignore preferences at the
outset, isn't based on doing that. Rather, it uses all of them
to choose according to the standards on which it's based.

As Markus recently told Demorep, it's the responsibility of anyone
proposing or advocating a method, to state exactly some advantage
of it, some standard or principle or criterion that it meets
and that other methods don't meet.

Mike