There is a problem with the first of the two evaluation methods I
described, in that it doesn't work when you have two or more overlapping
sets of equally ranked votes.  Two weak candidates could defeat each
other.

Example:

12 A
10 (A=B)
13 B
10 (B=C)
14 C

If you convert (A=B) to (B>A), and (B=C) to (C>B), before dropping
anyone, you get:
24C > 23B > 12A; C wins.

Under the (correct) second method, you drop A first.  You now have:

23 B
10 (B=C)
14 C

Drop C; B wins.

Bart

Reply via email to