IRV inconsistency

[Buddha Buck]

At 03:35 PM 05-17-2001 -0400, you wrote:
>Mr. Schulze wrote in part-
>
>On the other side, Condorcet methods are criticized very frequently
>because the winner depends "only" on the pairwise matrix while
>other information is ignored.
>--
>D- Which is why I suggest an *absolute* YES/NO vote on each choice with only
>the YES majority choices going head to head (which only shows *relative*
>votes).

In your opinion, would it be reasonable to assume that if a voter favors 
candidate A to candidate B, and is willing to give an *absolute* YES vote 
to Candidate B, they would also give an *absolute* YES vote to candidate A 
(and if they would give a NO vote to Candidate A, they would also give a no 
vote to B)?

If so, instead of asking the voter to both rank the candidates and to vote 
yes/no in each candidate, I can see three options:

(Sample ballots assume that the voter has a preference of A>B>C>D>E in a 
five-candidate election, and would vote yes on C and higher, no on D and lower)

1) Ask voters to only rank the candidates they would vote an absolute *YES* 
to, with a special "none-of-the-above" mark (to indicate a NO vote to all 
candidates).  Ballot:  A>B>C

2) Ask the voters to use dyadic rankings,  Ballot: A>B>C>>D>E

3) Ask the voters to rank a mythical "None-Of-The-Below" (NOTB) candidate 
as well as the real candidates.  Ballot: A>B>C>NOTB>D>E

The difference between 2) and 3) is the in which a voter could indicate 
unanimous disapproval of all of the candidates.  If there is no >> in a 
dyadic ballot, is it universal approval, or universal disapproval.  I think 
3) is easier for a voter to understand, as well.

I don't like 1) because I see the information below NOTB to be 
useful.  While the voter casting the ballot above obviously feels that D 
and E are both "evil", he would prefer that the lesser of the two evils 
gets elected over the greater (if none of his preferred candidates voted, 
that is).

>How many U.S.A. President candidates could get a YES majority (assuming no
>incumbent is running again) ???

It's hard to say.  With plurality elections and less than 50% of the 
eligible population voting, the data from elections isn't there to answer 
this question.

>More detailed *absolute* info would be nice --- but requiring *too much* info
>input from the voters may be asking too much (noting Mr. Simmons' Oregon
>FAVOR education efforts).
>
>That is, what is the upper info input limit for the *average* voter about
>choices (which would thus be used to determine choices) ---
>
>YES/NO plus ranking (1, 2, etc.)  ???
>
>Scale voting (100 to 0) ???
>
>Something else ???

"Grading" (for each candidate, an A, B, C, D, F grade)

Ranking + NOTB

More?