IR elects Hitler?
seppley at alumni.caltech.edu
Wed Nov 6 21:10:54 PST 1996
>I am not in favor of the "simple Instant Runoff method".
>See my Subj: [ER] Defects in various single winner methods
> Date: Sun, Oct 27, 1996 11:50 PM EDT.
>The mere possibility of Condorcet circular ties with a bunch of
>majority intolerable candidates is a formula for disaster.
I don't think it's a good idea to use the word Condorcet in this
statement, since readers might think you're referring to the
Condorcet method rather than crediting the Marquis de Condorcet
with discovery of the circular tie.
Circular ties aren't a property of the method; they're always a
possibility because they reflect an uncertainty among the voters.
Whether or not the official method elects by Condorcet, or even
performs a pairwise tally at all, doesn't change whether of not
a circular tie exists among the voters' sincere (or voted, if a
ranked ballot method is used) preferences.
So I take your point to be: "The mere possibility of electing a
majority-intolerable candidate is a formula for disaster." It
applies as much to non-pairwise methods like Instant Runoff.
Questions: would it be a disaster if an important office is never
filled? If so, which disaster is more serious? More likely?
>H beats G, 67 to 33
>G beats S, 68 to 32
>S beats H, 65 to 35
>Circular Tie--- H>G>S>H
Yes, this is a "circular tie" (a.k.a. "cycle") even if the tally
isn't Condorcet, and even if the pairwise matrix isn't officially
This isn't close to the examples I requested, which would either
show IR doing better than Condorcet or Condorcet electing a
disapproved candidate when an approved candidate is also running.
IR elects H in this example, no better than Condorcet.
>With the lesser of 2 evils [3 evils in this case] tie breaker,
>H is elected with a mere 65 percent of the voters against him in
>his worst (and only) defeat.
What makes this a plausible scenario? Why didn't any decent
candidates compete against this field of horrors?
>Opponents of any single winner reform will have a field day in
>bringing up even the remote possibility of such a result (i.e.
>scaring the sxxx out of the average U.S. voter).
Since the same thing could happen using the existing rules, I
don't think people would treat this as an argument for the status
quo. Why would it be more likely to happen if a "plain" reform is
>To those who say such example cannot happen, I note what happened
>in Germany in 1932-1933. The Communists/Socialists joined with the
>Nazis to politically destroy the moderates
Apples and oranges? Parties make deals a lot quicker than millions
of voters. Hitler wasn't elected Chancellor by the voters, and they
never used a good single-winner method to elect anyone.
This scenario sounds more like a scare tactic which will be used
against the parliamentary form of prop rep.
>With some advance polling, such types of candidates would not
>possibly even attempt to get their names on the ballot. The risk
>of having "unknown" evil executive/judicial candidates on the
>ballot is one of many unpredictable risks.
I think this question of whether advance polling would more
effectively prevent bad candidates from competing if None might win
is worth exploring, but I don't think you've yet provided an argument
why advance polling with a good "One Must Win" method wouldn't deter
them just as effectively.
I'll try to provide one: in "down ballot" races which the public
doesn't care much about, few voters will know the candidates and few
candidates will compete. Perhaps a good solution is to add the
disapproval option for "down ballot" races, and/or when the number
of candidates is below some threshold.
Another factor is campaign money. If there are only a few
candidates, big money donors can buy all of them and expect a return
on their investments if one is guaranteed to win. With the existing
system, the poor voter can only rarely find a decent candidate who
doesn't play the money game, and the LOE dilemma keeps the voters
from turning to the financially poorer candidates. Will a good
sw method cause the voters to elevate the poorer candidates in their
rankings, knowing this won't elect the greater evil? Depends; again,
this is more of a problem with less important offices and only a few
---Steve (Steve Eppley seppley at alumni.caltech.edu)
More information about the Election-Methods