[EM] Markus, 2 March, '05, 1317 GMT

[Markus Schulze]

Dear Mike,

I wrote (27 Feb 2005):
> Election methods are defined on _cast_ preferences
> and not on _sincere_ preferences. Therefore, whether
> a given election method passes a given criterion
> must depend only on how this method handles _cast_
> preferences. Therefore, a criterion can be well defined
> only when it can be defined without mentioning _sincere_
> preferences.

You wrote (2 March 2005):
> Wrong. Now some are defined on preferences, as opposed
> to only votes.
>
> Of course you can define anything any way that you want
> to. So you can define a criterion as a voting system
> Yes/No test that is doesn't mention preference. But that
> would be an unnecessarily narrow definition of a criterion.
> A voting system criterion is a Yes/No test for evaluating
> voting systems. The ones that you like don't mention
> preference. But, contrary to your claim, quoted above,
> those ones that you don't like, which mention preference,
> are criteria, by any reasonable definition of that term.
>
> I'm not asking you to like my criteria. But you need to
> understand that "the criteria that Markus likes" doesn't
> mean "the set of all criteria".
>
> Certainly, if we accept your definition of a criterion,
> which says that not mentioning preference is a requiremen
> for criteria. But your definition is arbitrary and
> unjustifiable.
>
> So  that's what it comes down to: You try to jusify your
> claim that my criteria are undefined because, by your
> definition, criteria are defined only if they don't
> mention preference. I suppose one can justify any 
> statement, by one´s chosen definitions.
>
> Ok, now you've answered my question. You've told me why
> you think that my criteria are undefined. Now this
> discussion has concluded, because you've answered my
> question.
>
> There's no need for more discussion about why you believe
> that my criteria are undefined, unless you have another
> reason for believing that, or unless you can post an
> example of a situation (configuration of candidates,
> voters, and voters' preferences) with which it isn't
> possible to say definitely that a some particular method
> passes the criterion or fails it.

You are the only one who defines criteria in terms of sincere
preferences and not in terms of cast preferences. Why should
anybody define criteria in terms of sincere preferences only
because you do that?

*********

You wrote (2 March 2005):
> Yep. BeatpathWinner and CSSD are completely different
> implementations. But of course there's also a sense
> in which they're one method, which is why I often say
> BeatpathWinner/CSSD. But I don't want to debate that
> issue.

Well, what you claimed in your 27 Feb 2005 mail was that
the Schwartz heuristic ("SSD") was not a heuristic for
the Schulze method. You wrote: "When I said that SSD is
equivalent to Schulze's method, that statement wasn't
entirely correct."

Now you write, that the Schwartz heuristic ("CSSD") and
the path heuristic ("BeatpathWinner") are two different
heuristics for the same method, the Schulze method. You
write: "BeatpathWinner and CSSD are completely different
implementations. But of course there's also a sense in
which they're one method."

I agree with your 2 March 2005 statement, but not with
your 27 Feb 2005 statement.

Markus Schulze