[Election-Methods] IRV "treats voters unequally"?

Warren Smith wds at math.temple.edu
Wed Dec 26 21:10:43 PST 2007

Some support for that stance is the fact that, if I say in my IRV
(Instant Runoff) rank-ordr vote that
X>Y  (i.e. I prefer X over Y)  while you say Y>X, then
IRV can count your vote but not count my vote when it comes to deciding
who to eliminate: Y or X.

As a result, unfair-seeming results can happen, e.g. Y winning
the election even though  S,T,W, and X all are prefered over Y
by voter majorities. (For example elections, see
http://www.rangevoting.org/rangeVirv.html#nasty .)

I think if you were to challenge the constitutionality of IRV, this would be
the best attack on it.  Gilmour & Venzke have a point that IRV can be interpreted as
treating all voters equally in certain senses, but in practice my sense may
be considered more nasty than their sense.

In these examples IRV *ignores* some preferences while counting others.
That seems unfair and arbitrary.

Interestingly, it can be shown that, in IRV votes with a large number
of candidates, IRV *always* ignores asymptotically 100% of the information voters
put in their votes.   See http://rangevoting.org/PuzzlePage.html#p34 .

IRV also has some other bizarre properties like in the Irish 1990 presidential
election, Lenihan voters could have made Lenihan win by voting for his archrival Currie instead.
Foolishly, they chose to honestly vote for Lenihan, causing Lenihan to lose.
See  http://rangevoting.org/Ireland1990.html .

Since Kathy Dopp is interested in election security against fraud, audits, etc,
she might want to consider the facts that
 * IRV is not countable in precincts (no such thing as a "subtotal"):
 * IRV has some of the highest voter-error (spoiled ballot) rates in the world:
 * IRV offers tremendous "nightmare potential" for creating elections virtually impossible
   to audit or total correctly.   To see what I mean consider
   where there is a near-tie EVERY round and great nations's fates hang on the fate of
   no-hope flake candidates, and keep doing so every IRV round. You think chad counting
   in Florida was bad in 2000?  That was nothing, baby.

Warren D. Smith
http://RangeVoting.org  <-- add your endorsement (by clicking "endorse" as 1st step)

More information about the Election-Methods mailing list