[EM] Re: completing Condorcet using ratings information

[James Green-Armytage]

Chris Benham wrote:
>With a very "high intensity" ratings ballot, it should be possible to do
>without the 
>plain rankings ballot. With a handful of candidates, why would a sincere
>voter want to give two candidates the same 
>ratings score out of 100, and yet rank one above the other?  

	Yes, it's tempting to do it without the ratings ballot. That would make
things a bit simpler of course. The only reason I like to allow people to
give equal ratings for unequally ranked candidates is so that they can
maximize the strength of the preferences that they want to maximize
without changing their rankings. In my example, I like for the Dean >
Kerry > Bush voters to be able to give both Dean and Kerry 100 while still
specifying a preference for Dean. Would it be so bad for them to give
Kerry 99? No, it wouldn't be a big deal, I admit. Still, if people want to
give equal ratings and unequal rankings, I'd like to give them the option
of doing so. I think that it would be a bit annoying to have rankings like
100 for Dean and 99.9999999999 for Kerry, but I guess that's another
option, which would enable you to do without the ratings ballot.

>Also, if cycles are rare, or if  too many voters think they
>are, then maybe too many voters will not bother filling out the ratings
>ballot.

	Hmm. If people don't fill out the ranking ballot, then you can create a
default one for them based on the ratings, interpreting equal ratings as
equal rankings of course. Also, if people don't fill out the ratings
ballot but do fill out the rankings ballot, you can create a default
ratings ballot by putting the top choice at 100, the lowest at 0, and
equally spacing the rest of the preferences.

>the ratings should be scaled and maximised among the members of  the
>Schwartz set,
>between steps  2  and 3.

	Sorry, Chris, what exactly do you mean by "scaled and maximized"? Is that
like raising the highest candidate to 100 and the lowest to 0?


my best,
James