[Election-Methods] IRV unconstitutional? (replies)

Warren Smith wds at math.temple.edu
Fri Dec 28 19:48:12 PST 2007

IRV defendors should aim at showing that IRV flaws are smaller
than FPTP flaws, thus FPTP should be declared anti-constitutional if IRV is.

WDS replies: IRV has some flaws which FPTP does not have.  For example,
non-monotonicity, no-show paradoxes, non-additivity.   I personally think IRV
*is* better than FPTP (at least if we can ignore issues of simplicity and fraud worries)
but not in these respects.

For real-world examples of non-monotonic IRV elections featuring no-show paradoxes,

Don&Cathy Hoffard:
>any time a new candidate X entering the race swings the winner from Y to Z, 
>that benefits somebody (namely Z, here)

This is not true in most if not all of the [IRV] the General Elections.  90-99% of the General
elections involve two major candidates and some minor candidates.
The winner will be one of the major candidates.  In IRV the minor candidates votes are drop and their votes are now cast for one of the major candidates.

You are right in some cases where you have 3 equal candidates.

WDS replies:
I gave constructed examples before of IRV elections where a candidate by entering race
swings the winner.  E.g. http://rangevoting.org/CoreSupp.html .
So yes, I am right.   For a real-world example, in the Louisiana 1991 governor race,
see   http://rangevoting.org/LizVwiz.html
Duke by entering the race caused Edwards to win, whereas otherwise Roemer would have won.
So your "if not all" is wrong - there is at least one counterexample. 
Another is Peru 2006:
http://www.rangevoting.org/Peru06.html .
So there are at least 2 counterexamples now.        

Indeed, theso-called "center squeeze" effect in IRV is where it is
Leftist vs Centrist vs Rightist.
Centrist is the Condorcet "beats all" winner, but is eliminated by IRV because
the left & rightists squeeze him into too small a regio of top-place support.
In EVERY such situation, one extremist, by enteringthe race, swung it to the other
whereas without him,Centrist would have won.  

THis is quite common: in "1 dimensional politics", this happens 1/3 (33%) of the time
to IRV.   How we know that: see

So you are quite wrong.  This is not "rare if at all." It is "common."
The error in your analysis was to only consider the "minor" guy as entering,
and to neglect the "major" guy as an entrant.  Oops.   When you only
consider some possibilities you naturally get a lesser count than if you consider them all.
The underlyign reason for your error was your USA-2007-centric thinking,
failing to even consider the possibility that a so-called "minor" candidate
might actually be a Condorcet winner.  IRV leads to 2-party domination
(a flaw it shares with FPTP) which somewhat "justifies" your error, but that
is another problem.  :)

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

More information about the Election-Methods mailing list