COMMENTS ON TWO FORMS OF IRV
Instant Runoff
Voting, as defined in proposed legislation in Vermont in 2002, and as used by
most proponents of IRV, asks each voter to rank the candidates. Once all rankings have been made, the
election procedure checks to see if anyone has a majority of the first-place
votes. If not, the name of the
candidate with the fewest first-place votes is crossed off each list and the
election procedure again looks to see if anyone has a majority. The process is repeated until a majority is foundÑif there are only two
candidates left, one will have a majority. (Yes, IÕm ignoring the possibility of a tie.) That process is also known as Single
Transferable Vote (STV).
In 2003 the
proponents of IRV changed the system.
If no candidate has a majority, all but the two with the greatest number
of first-place votes are eliminated, and the system chooses the one of those
two with the most first-place votes.
IÕll call this system IRV 2003 and the other one IRV 2002. The stated reason for the change
was Òsimplicity.Ó
IRV 2003 shares all
the faults of IRV 2002, but has one more. Proponents of IRV note that one of the faults of
Plurality voting is the case of spoilers.
They argue, for example, that in Florida in 2000 Nader was a spoiler
because he had no chance to win, but he took enough more votes from Gore than
from Bush that Florida went for Bush instead of Gore. IRV 2002 is not directly affected by spoilers, because
first-place votes for the potential spoiler are typically so few that that
candidate gets eliminated early in the election procedure. In a four-candidate race, IRV 2003 is
susceptible to spoilers. The
following example illustrates the point (the notation is as we did it on the
board in class):
3 30 32 35
A B C D
B A B B
C C D A
D D A C
Under IRV 2002
(which is the method we used in class), A is eliminated first, and then B has
33 first-place votes, so C is eliminated next, and B win over D 65 to 35. Under IRV 2003, both A and B are
eliminated in favor of the C and D, which are the two highest vote
getters. C beats D 65 to 35 even
though. C is ranked below the median by 68 voters. If A had stayed out of the race, B would have had 33
first-place votes, and, under IRV 2003, C would have been eliminated, resulting
in BÕs winning. Thus by entering
the race with no chance to win, A is a spoiler.