seppley at alumni.caltech.edu
Sat Feb 8 18:27:11 PST 1997
>Mr. Eppley wrote:
>>MOAV isn't intended to apply to electing a Legislator in a single
>D- Correct. I suggest a proxy p.r. system for legislative bodies as
>I have mentioned for at least a year.
Demorep apparently suggests we continue to use the horrible Plurality
or Top2Runoff when electing legislators from single-winner-districts,
until some unknown time in the presumably distant future when we can
achieve proxy PR. It's silly to wait until then when we can make a
big improvement now in the single- winner-district voting method, at
no extra cost while we're working on general reform of the
single-winner voting method.
If MOAV is intended to accompany a proxy PR system, and won't work
well without proxy PR, then I propose we postpone making MOAV part
of single-winner reform until it's time for the proxy PR reform.
Hopefully by then we'll have a good single-winner method in use, so
the voters can rank the various flavors of PR systems.
I've made this point before, and Demorep hasn't replied to it.
This is a repetition of disorderly discussion.
>>MOAV is not a single-winner criterion. It's a "one or none"-winner
>>criterion. There's no way to guarantee a majority of the voters
>>will approve of the person elected (or selected) for the office.
>D- As I have mentioned around 5 times, the legislative body can
>fill any executive or judicial vacancy if no candidate gets a
>majority of all the voter's votes.
Demorep has indeed said that many times. And I have answered it
often, but he has ignored my reply: There's no reason to expect that
a choice of the legislature would be guaranteed the approval of a
majority of the voters. And since we can expect legislatures to be
non-PR for the foreseeable future, say hello to President Gingrich.
Demorep has often recited the example of Weimar Germany, in which
the *legislature* chose Herr Hitler; the "legislature solution"
Demorep's continued repetition with a lack of response to these
points is another instance of disorderly discussion.
>History and common sense indicated that there is ultra-danger in
>having any election method that permits nonmajority winners for
>executive and judicial offices (which is why plurality nominations
>and elections are ultra-dangerous).
History hasn't had a chance to with Condorcet's method for public
offices. This will be a majority rule system, not a plurality
system. So Demorep's point is irrelevant.
>How many times does Hitler and his murderous plurality/minority
>politics have to be mentioned?
It's non sequitor to bring up plurality rule in this discussion.
We're all agreed, I think--except maybe Matthew Shugart, if he's
still here--that plurality rule voting needs to be replaced by
majority rule. Condorcet and Smith//Condorcet are majority rule
systems. They're just not "majority approval" systems, something
Demorep believes is vital--but he hasn't provided a compelling
reason why some decent candidate won't run and beat Hitler if
Condorcet or Smith//Condorcet is the sw method, or why he thinks
the legislature would pick a majority-approved executive if the
voters turn thumbs down to all of the candidates. More
>There is only a probability that even majority winners will not be
That argument isn't relevant to the choice of single-winner voting
method. That argument is about the importance of checks and balances
and constitutional rights.
>No method can guarantee against strategic voting - see Mr. Arrow's
More disorderly discussion. Demorep continues to ignore the point
that some methods are far more prone to manipulation than others.
>>>D (earlier)- >If each vote is an approval vote in the 35-32-33
>>>example, then only B gets majority approval.
>>That's a big "if". How can it be justified? How would a voter be
>>able to both vote "No" on a candidate and rank that candidate ahead
>>Since IRO elects A, not B in this example, Demorep has explicitly
>>acknowledged that IRO violates MOAV. (This contradicts what he
>>wrote about IRO earlier.) MOAV//IRO is not the same as IRO.
>My suggested ballot for executive and judicial offices would be:
>Vote YES for each candidate that you find acceptable to be elected.
>Vote NO for each candidate that you do not find acceptable to be
>elected. Use numbers (1, 2, 3, etc.) to rank your first, second,
>third, etc. choices
Yes, but this proposed method is NOT plain Instant Runoff. Demorep
needs to stop claiming the properties of plain IRO are the same as
this more complex method.
>If shorthand notations must be used, then the above is
>MOAV//Condorcet//IRO or MCI.
Well, it's certainly not plain IRO. But it looks like
MOAV//BeatsAll//IRO or MOAV//Smith//IRO to me. (It's not clear
which of these two Demorep intends, since he hasn't clearly
defined what it means to be "tied" in the head to head phase.
He didn't write "circular tie" but nevertheless I'd guess he
On what democratic standards or criteria does Demorep think
MOAV//Smith//IRO is better than MOAV//Smith//Condorcet (or that
MOAV//BeatsAll//IRO is better than MOAV//BeatsAll//Condorcet,
if that's the method he's advocating)?
>In any step where a candidate loses, the rankings move up
IRO, and Demorep and Donald, use a fallacious measure of "lastness"
to declare a candidate a "loser" which needs to be iteratively
eliminated. No reasonable argument has been made that "Iteratively
Drop One Candidate At A Time" is a serious criterion of democracy.
>The IRO feature of dropping the candidate with the lowest number
>of first choice votes would occur only if there are 3 or more
>candidates in a head to head tie.
May as well include the "2 tied candidates" case, for completeness' sake.
Why is //IRO better than //Condorcet?
>Mr. Black's book indicates that the possible fraction of circular
>ties goes way up as the number of candidates increases- which
>matters for single winner elections only for first place.
I think we're agreed that the choice of circular-tie-breaker is
Hopefully Demorep is prepared to explain why he thinks
MOAV//???//IRO is better on some serious standard or criterion than
MOAV//???//Condorcet, in accordance with the new EM behavioral
standard Rob L inserted in the list's welcome message.
>How many candidates would run for U.S. President, Governor in big
>states or Mayor in big cities (especially without an incumbent
Many, I hope. It's important to give the people many choices to be
ranked. (Corollary: it's important to use a voting method which won't
deter potential candidates from competing for fear of spoiling the
election to a greater evil.)
>How many of such candidates may be leftwing or rightwing extremists?
Who cares? Extremists won't win if a good majority rule method like
Condorcet is used. Does Demorep believe that only extremists will
run for such important offices?
The most powerful of the offices Demorep listed is President.
If Hitler were somehow elected Governor of a State, he still would
be constrained by the rights of citizens and the Republican Form of
Gov't guaranteed by the U.S. Constitution. The President too is
constrained, although s/he has enough power that enforcing the
constraints may be more difficult.
I think it's quite a stretch to believe there won't be any decent
candidates capable of beating Hitler pairwise competing for a
high stakes office. Hitler would not be selected by Condorcet,
nor by Smith//. Switching from the existing Plurality & Top2Runoff
methods to Condorcet or Smith//Condorcet would impose a significant
barrier to a Hitler or a Gingrich achieving executive power.
Demorep's alarmism and insistence on ignoring the KISS principle
---Steve (Steve Eppley seppley at alumni.caltech.edu)
More information about the Election-Methods