[EM] Calling something by a name that means something else. Justification for PC.

[MIKE OSSIPOFF]

I'd said:

>I thought you were referring to PC, because some people on this list call
>PC "MinMax".
>
>You see, this is why I've tried to discourage "MinMax" as a name for PC: 
>"MinMax" is used with too many different meanings for it to be of any use
>as the name of a count method.


James replied:

	I was under the impression that it is not entirely clear what completion
method Condorcet was actually proposing.

I reply:

Yes, Condorcet's proposals weren't worded as specifically as we'd like. 
Still, his drop-weakest proposal sounds a lot like PC. PC is surely the best 
suggestion for the literal interpretation of Condorcet's drop-weakest 
proposal.

There's the issue of how Condorcet meant for defeats to be measured. But 
Norm posted a quote that settled that question. That quote proposed wv as 
the defeat-strength measure.

Though PC sounds like what Condorcet was proposing, it could be suggested 
that if Condorcet were actually conducting a vote, he might actually improve 
on it, using Smith-PC, or SSD. That's what, I claim, justifies calling SSD & 
Smith-PC Condorcet. RP is Condorcet because it's the best interpretation of 
Condorcet's keep-strongest proposal.

But please don't start an issue about what Condorcet meant, because that 
will surely bring Markus out of the woodwork.


You continued:

I have heard some people argue
that it was minimax

I reply:

MinMax or minmax isn't a method name. It's been used with too many different 
meanings for it to mean anything specific.

You continued:

, and some people argue that it was Kemeny.

I reply:

I've heard that Kemeny was a complicated, elaborate procedure. I'd be 
curiious how anyone could read that into one of Condorcet's proposals.

You continued:

In light of
this ambiguity, I think that the term "plain Condorcet" for minimax...

I reply:

There is no one method called minmax. Just a few days ago Chris was using it 
to mean something very different from PC. The method that you're referring 
to has been called PC. It's also been called Basic Condorcet. Those are the 
two names that have been used for that method and not for other methods.

You continued:

... might
be slightly misleading

I reply:

No, because when you look at Condorcet's drop-weakest proposal you'll 
recognize it as PC.

But aside from that, the mere fact that PC is _one_ of the Condorcet 
interpretations, and is also the simplest and plainest, obviously qualifies 
that method for the name "Plain Condorcet", regardless of any issue about 
what Condorcet meant.

You continued:

, or controversial at best.

I reply:

Sure, anything can be controversial here. As you might have noticed,  the 
matter of
whether "I hadn't heard of that" (at a specified past time), and "I haven't 
heard of that"
mean the same thing is controversial to Markus, as is the issue of whether 
"I don't claim that" means "I have never claimed that".

You continued:

If I'm wrong, and it is
totally clear that this was the basic method that Condorcet proposed, then
I suppose that "plain Condorcet" is a reasonable name.

I reply:

It's a reasonable name anyway, if it's a better candidate for the literal 
interpretation of one of Condorcet's proposals. It's also a reasonable name 
merely by virtue of the fact that it's at least _one_ of the interpretations 
of a Condorcet proposal and is especially simple and plain.

You continued:



	As for the term 'minimax,' I'm not saying that it is the best term in the
world, but I don't think it's too bad.

I reply:

Yes, "minmax" is a great name if you like terms that have several different 
meanings. Chris, a few days ago, used that name for a method that counts a 
candidate's votes-against him in his victories as well as his defeats. But 
when you say "minmax", you might want to add clarification about which 
minmax you're referring to.

You continued:

Basically, the winner is the
candidate whose worst loss is least bad. Hence minimax.

I reply:

No. Chris uses that name to refer to a method that elects the candidate who 
has fewest votes against him in _any_ pairwise comparison, whether that's a 
defeat or a victory. That's not PC.

You continued:

I think that the
word comes out of game theory, where it is applied in a quite different
context, but with a slightly similar meaning.
	I have also heard the terms "successive reversal," "sequential dropping,"
and "Simpson" used to describe this simplest of Condorcet methods.

I reply:

Sequential Dropping could indeed be a name for PC, but when I coined the 
name  SD, I applied it to a method that successively drops the weakest 
defeat _that is in a cycle_.

The _Journal of Economic Perspective_, Winter '95, defines Simpson-Kramer as 
a method that looks at votes-for in victories and defeats. Is there 
somewhere else where "Simpson" is defined? Exactly what does that definition 
say, exact wording?

You continued:

You continued:

I
usually use "minimax." I don't know why. Maybe because it's short.


I reply:

PC is shorter, with the further advantage of not also being used to denote 
entirely different methods.

You continued:

	Anyway, it might not be an issue of world shattering importance, but that
is my two cents on it.

I reply:

No, it isn't an issue of world-shattering importance, but it's never a good 
thing to call one method by a name that's used to denote a different method.

Mike Ossipoff

_________________________________________________________________
Take advantage of our limited-time introductory offer for dial-up Internet 
access. http://join.msn.com/?page=dept/dialup