[EM] majorities and ordinal-only pairwise methods

[Jobst Heitzig]

Dear James!

You wrote:
> 	I agree with those who consider "approval" (score of 1 or 0) to be a kind
> of cardinal information. 

That's a strange understanding of "cardinality", don't you agree? There
is nothing to be counted in a yes-no-question. It's only the question
whether I find that candidate OK or not.

> I'd like to know what makes you think that
> approval information can be sincere (i.e. what is the definition of
> sincere approval?)

I did not say that I think that approval information can be sincere or
that I knew a definition of sincere approval. I certainly know that most
of the time I would easily be able to answer the question whether I find
the candidate OK or not. I also know that I could almost never give any
cardinal rating to a candidate since I don't know what that number is
supposed to be meaning. The formal definition of approval is an open
problem.

You also wrote:
> 	I'll repeat what I wrote in my last message, in the same thread: 
> 	My main purpose in using cardinal information is to incorporate
> preference strength into defeat strength, and thus to protect defeats that
> consist of many strong preferences. 
> 	Using other information in addition to ordinal information also has the
> potential to allow voters to engage in effective counterstrategy without
> altering their rankings, which generally leads to greater stability in
> strategy/counterstrategy scenarios. (This second reason is shared by
> methods like AERLO/ATLO, S/WPO, etc.)

Then I'm sorry to have misrepresented what your main point is.

> 	Oh, wonderful! Are you fully aware of my "approval-weighted pairwise"
> method (AWP)? AWP is basically the same thing as CWP, except for the fact
> that only two cardinal scores can be given: 1 and 0. I suggest that it has
> anti-strategic properties that are comparable to CWP, and that it is
> significantly more strategy-resistant than DMC/RAV, AM, margins, and WV.

I know, and that is a great pro for AWP. That's also why I essentially
proposed something like AWP as "grand compromise" half a year ago. But
DFC and DMC are definitely much more simple without having considerably
more strategy problems, and that's why I currently prefer DFC to AWP.

Yours, Jobst