[EM] Condorcet/Range DSV

Jameson Quinn jameson.quinn at gmail.com
Thu Jun 25 12:53:45 PDT 2009


I left out one good aspect of this system. It is additive - ie, it can be
counted locally. You need to keep n(n-1)(n-1)/2 tallies - the Condorcet
matrix, plus the result of each candidate renormalized against every other
pair of candidates. If there were ever more than 3 candidates in the Smith
set, you could reconstruct the multiway-renormalized totals from linear
combinations of the summed 3-way-renormalized totals.
If you are willing to do one and a half counting rounds (that is, one
definitely and then another one if there's no Condorcet winner), then you
keep n(n-1)/2 tallies in the first round and n tallies in the second round.

Generally, I'd support ballot access hurdles to keep it down to around 4-6
candidates per election. That's 18-90 tallies for one-round counting, or
10-36 for one-and-a-half-rounds.

The system is also pretty easy to explain. "You rank each candidate from 0
to 100. If one candidate beats each of the others one-on-one, they win. If
not, you ignore all the loser candidates who only beat other losers. Then,
to give everybody's vote full power, you set their favorite winner candidate
to 100 and their least favorite winner to 0, leaving the spacing of the
choices in between the same. You add together each candidate's rating on all
these fixed-up ballots and whichever candidate has the highest overall
rating wins.

"As a voter, unless you think there will be no clear winner AND you know
exactly who will beat whom, there's no reason to be anything less than
perfectly honest, because the system fixes up your ballot to give it full
power in the decisive runoff. If you do try to beat the system and you're
wrong, it will backfire; the safest ballot is your honest vote. You don't
even have to be too careful about giving everyone exact ratings, because it
only matters if there's not just a *two* way tie but actually a close tie
between at least *three* strong candidates, which is very rare."

(When Condorcet methods have more of a political track record, we'll be able
to start saying how rare. With random ballots between 3 candidates, it's
about 9% probable; since most elections do not have 3 equally-strong
frontrunners, it should be much rarer than that in practice.)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20090625/b1381363/attachment-0004.htm>


More information about the Election-Methods mailing list