[EM] Condorcet Internet Service poll on political party platforms is now receiving votes.

Michael Ossipoff email9648742 at gmail.com
Mon Apr 1 21:07:29 PDT 2013


The nomination period has concluded, and now the poll is receiving votes.

I said that I'd chosen Condorcet-IRV (referred to by Armytage's
article as "Benham") for the poll's count method. It's one of the 4
rank-counts offered. But actually, anyone looking at a poll result can
have it counted by any or every one of the 4 count methods implemented
at the website. At the page that shows the results, in the right
margin, near the top, is a list of the 4 count methods, with a circle
to the left of each method name. Clicking on one of the circles will
mean that the results shown will be the results
by that count method.

But of course there nearly always will be a CW (as is the case so far
in this particular poll). And so, because all 4 of the implemented
count-methods choose a CW if there is one, then it would be quite rare
for the methods to choose differently from eachother.

Most likely, the only way to have a top-cycle instead of a CW would be
for a number of voters to use or attempt Condorcet offensive strategy,
or chicken-dilemma defection. With Condorcet-IRV, such defection would
backfire every time. That's how Condorcet-IRV differs from the other 3
count-methods.

The results page starts with an output-ranking of the alternatives
(party platforms, in this case). First the count is done, among all of
the alternatives. The winner gets the #1 place in the output-ranking.
Then a count is done, among all of the alternatives except for the one
that has already won. The winner of that count gets the #2 place in
the output-ranking.  ...etc. Counting among all but the N alternatives
that have already won, the (N+1)th alternative in the output-ranking
is chosen.

The #1 alternative in the output ranking is highlighted.

And if that alternative is the CW, then "Condorcet winner" will be
written after that alternative's name.

Of course, when it says that there is a CW, then of course it doesn't
matter which count method is used. So I suggest that, when looking at
the results, just leave the count-method setting at its
default--Schulze. No point bothering to change it unless you find that
there isn't a CW.

If there isn't a CW, then I suggest setting the count method to
Condorcet-IRV, because that's the one that I announced in the poll
introduction. If voters expected Condorcet-IRV, then they might rank
more sincerely, and that sincere ranking could be penalized if the
count is done by a different method--when there isn't a CW. Therefore,
for this poll, the announced count method, Condorcet gives a more
relevant result, truer to the intent of sincere-ranking voters who
expected that count method.

Michael Ossipoff



More information about the Election-Methods mailing list