[EM] Bart, Bart, You Rang?

[LAYTON Craig]

Donald wrote:

>     Besides, B will not be any kind of a winner if your example were more
>realistic. Allow me to change your example to make it more believable.
>     49 A,  2 BA,  21 BC,  2 B,  2 CA,  22 CB,  2 C
>     Now, candidate B is not the Approval winner.  A is the Approval
>winner.  A is the Condorcet winner.  A is the Instant Runoff winner. A is
>the winner all around.

And:

>     I wish to thank you for rating the candidates, because when you rated
>them you opened up another chink that points to what I have long been
>saying, that is, there is a big flaw in the lower choices. They are of
>lessor value.

I'm not Bart, but anyway.  I haven't had much chance to practice simple
arguments against IRV, so feel free to delete this message if you don't want
to read them.

Ratings might be important, but are not the be all and end all of voting
systems.  If it were, we'd all be supporting Borda or some other points
system.  I'm suprised, actually, that you don't, given that the top choices
are apparently so much more important than the bottom choices. Yet you
advocate a method where bottom choices often decide elections (and are given
almost equal weight to the top choices).

You don't, because you want to get a winner with a majority.  You want to
select the most preferred candidate.  These are aims with which I most
definately concur.  Borda is unnacceptable because a candidate who is most
favoured by an absolute majority can lose.  For the same reason but to a
lesser extent (because it is much less likely), Approval isn't the most
desirable option either.

But why IRV?  You get a candidate with a majority over one other candidate.
So what?  You get a candidate with a majority over one other candidate in
almost every imaginable electoral system.  In Condorcet systems the winner
will always have a majority over one other candidate.  In fact, the winner
will have more of a majority over more other candidates than an IRV winner.
I have been convinced, that a straight Condorcet system is better than one
with a minimum quota of first preference votes, but a minimum quota is still
okay (still much more sensible than IRV).  Why is IRV better than
Condorcet//Smith with a minimum quota?  First preference votes (if that's
what you're concerned about) are potentially more important in the latter
system, because, in IRV, a candidate can win with a very small number of
first preference votes, or even none at all!