[Election-Methods] YN model - simple voting model in which range optimal, others not

[Abd ul-Rahman Lomax]

At 01:02 AM 3/24/2008, you wrote:
>Yeah, I'm confused too.  I haven't even figured out how second choices
>are determined for IRV and Condorcet.

"Even"? That's much harder and was not specified. The first problem 
is easy. Just read the thing, and don't make assumptions. It was 
stated accurately. That is a standard voting table: This many voters 
held this specific position and would therefore vote, in Plurality, 
for the candidate holding the same exact position as they do.

As to ranked methods, there are various ways to do it. You simply 
need a way to determine how the voters rank candidates. The most 
obvious (but by no means necessary) way would be that all 4-issue 
matches would be preferred to all 3-issue matches, all 3s to all 2s, 
etc. But there are four 3-issue-matches for each voter. So how do you 
discriminate between them, if the method does not allow equal 
ranking? I assumed that match pattern was a binary number, and that, 
if the number of matches was the same, the candidate with the higher 
value match pattern was the preferred one.

So, for example, suppose a voter's pattern is

YYYN, and a candidate is:
NYYY. The match digit is 1 if the digits match and 0 if they do not. 
So the match pattern here is
0110.

This is a 2-candidate match. Consider

YYYN (same voter as above) and
YNNN. Match pattern:
1001.

This is also a 2-candidate match, thus ranked in the 2-candidate 
block. Within that block it is preferred to the first one described, 
because the value is higher. (This is a result of the values assigned 
to binary digits, the left-most one has the highest value).