[EM] Condorcet complicated?

[bql at bolson.org]

On Thu, 14 Oct 2004, Paul Kislanko wrote:

> "Likewise, IRV can suffer spoilers, a problem Condorcet avoids by reading
> all the ranking in each ballot."
>
> Condorcet does not reference ballots, Condorcet depends upon the pairwise
> matrix which cannot be mapped back to ballots.

I've never heard of anyone wanting to take results and map them back to 
ballots. A summation of single votes cannot be mapped back to ballots.

Anyway, I think the original point is that Condorcet style methods consider
a voter's whole expression at once, instead of only considering a tiny
fragment (whichever choice is first at some stage).

Methods with this wider view are capable of finding better solutions that
more narrow methods miss.

Another way: Condorcet can find a compromise choice that IRV could miss.
A compromise choice might make 80% of the people mostly happy while
IRV would find the choice that makes 60% happy and 40% unhappy.

Brian Olson
http://bolson.org/