Lowest Majority loser tie breaker
dfb at bbs.cruzio.com
Mon Nov 4 14:21:10 PST 1996
DEMOREP1 at aol.com writes:
> As Mr. Ossipoff well knows the simple Condorcet method only indicates
> relative support.
The Condorcet circular tie solution is more about defeating majority-rejected
candidates, when its possible to do so (when not everyone is majority-
rejected). As opposed to trying to add-up support, according to some measure.
As I said, by counting votes-against, it's counting what the lesser-of-2-
evils voter wants counted so badly that s/he's willing to forfeit the
chance to vote _for_ someone. But with Condorcet, s/he can do both.
> He apparently would see no problem if the only candidates were Hitler, Stalin
> and Genghis Khan as long as Condorcet produced a winner detested the least.
Yes, if that's the best candidates the people can come up with. In an
election with a really poor selection I either wouldn't vote (as I'm
going to not vote on partisan races tomorrow) or I'd vote NOTB over
everyone, or vote "no" on everyone if one of those 2 options were
Look, I've repeatedly explained this to Demorep, & she still misquotes
me: I keep saying that I _don't_ oppose giving voters the option to
vote "y" or "n" on any candidate they want to, in addition to voting
their ranking. Likewise, I don't oppose including "NOTB" among the
alternatives. I even listed several ways of interpreting NOTB in
a recent posting. I even said I wish we had y/n or NOTB in the
current election, and that if we did, then I'd use it, even if we
had Condorcet's method.
Where Demorep makes her mistake is when she insists on hybridizing
y/n with Condorcet's method itself, rather than including it as an
added option. As I've already said, Demorep's suggested modifications
of Condorcet would take away its important & unique properties.
I even said that adding the y/n option, or maybe NOTB, might very well
be an extremely popular attraction for voter acceptance of the reform.
> He continually ignores my comments that any tie breaker is done only after
> head to head pairings are done.
Ok, I acknowledge that comment. But it isn't true, since I don't
oppose including y/n or the NOTB option, and immediately disqalifying
anyone who gets "n" from a majority, or who has "NOTB" ranked over hir
by a majority. Note that these candidates would be disqualified before
the pairwise-count. Whether defeats by them shouldn't be counted in
the pairwise count, including the Condorcet circular tie solkution
is another issue. I beleive that being beaten by a disqualified candidate
as at least as bad as, probably worse than, being beaten by
an un-disqualified one. So no, I wouldn't delete disqualified candidates
before doing the Condorcet count, though those disqualified candidates
would be barred from winning (because a majority had voted "n" on
them or ranked NOTB over them).
> My standard remains majority rule. His standard is obviously the lesser of 2
> evils (which in many cases would just mean a minority winner with a somewhat
> higher percentage approval than the ordinary plurality winner).
Mission Control to Demorep: I've repeatedly made it clear that my standards
are majority rule & the LO2E problem. I've repeatedly shown how Condorcet's
method does better by majority rule than does any other method. It's
absurd to compare Condorcet, without a separate y/n option, to
a method with a y/n option, especially since Condorcet advocates don't
oppose a y/n or NOTB option. (Aside from that, it would be so difficult
for a majority-rejected canddiate to win in ordinary Condorcet, that
there wouldn't really be much of a need for y/n or NOTB).
By "minority winner" you mean someone who doesn't have a majority?
A candidate with a majority would win in Condorcet's method or in
any pairwise method. But even when no one has a 1st choice majority,
we can still beat candidates who have a majority against them,
and Condorcet does that.
When you suggest y/n as a subsitute for Condorcet, you're wrong,
because y/n, like Approval, requires voters to divide candidates
into 2 distinct sets only: Approved & disapproved. What you're
missing is that you might approve one candidate more than another,
or disapprove one candidate more than another, and in that case
y/n just won't do, will it. y/n or NOTB is for designating candidates
that you absolutely reject. Fine. But that isn't enough.
Oh, majority rule is your standard? I've mentioned what I consider
to be a basic democratic principle, about majority rule. Tell
me if your method abides by this principle:
If a majority of all the voters indicate that they'd rather have
A than B, then if we choose A or B then it should be A.
There's a method (Condorcet) that will never unnecessarily violate
that principle. Can you say that for any of your methods?
If not, then in what way do you claim that 1 of your many methods
does better by majority rule? And don't say "By incorporating
a y/n option", because, as I said, Condorcet advocates don't oppose
a y/n option.
More information about the Election-Methods