Too late this night for fancy words, but hopefully I can express some
useful thoughts.
On Aug 30, 2011, at 4:52 PM, Richard Fobes wrote:
Here is what I've just written for the new section titled "Multiple
rounds of voting":
----------- begin ------------
In highly competitive elections, multiple rounds of voting are
needed to eliminate the weakest candidates so that attention can be
focused on electing one of the most popular candidates. Our
supported election methods work as described for two rounds of
voting if the first round of voting elects a single winner from each
political party, and the second round chooses from among those
winners.
FPTP has a serious problem because it cannot let a voter vote for more
than one candidate - and voters can want to vote for more than one -
and to say which are liked better than others. Methods we are
promoting, such as Score and Condorcet, give the voter needed power.
With such methods rounds become less needed since voters can better
express their desires in the main election. Likewise, when there are
to be rounds, more of the weakest can be discarded before the round
since we know better which of the weakest might believably win.
The last sentence above is about primaries. FPTP desperately needed
such to avoid multiple candidates from a party competing in an
election. Once voters understand they can vote for more, and indicate
their preference via rating or ranking, primaries will lose much of
their backing - thus, possibly getting discarded with FPTP.
Note that parties could have a single candidate and not have need for
a primary, even in FPTP days.
However, different counting methods are needed if the same voters
vote in both rounds. There are election methods that handle such
cases, and they use the better ballots we support. However, we have
not yet analyzed this category of counting methods sufficiently to
express support for any specific methods.
Assuming primaries still exist, I see no need for that round being
unlike the main election, even noting that some voters would be voting
in both.
We do strongly agree that single-mark ballots must not be used in
any round of voting. More specifically, just as the candidate with
the most first-choice votes is not necessarily the most popular, and
the candidate with the fewest first-choice votes is not necessarily
the least popular, the candidate with the second-most first-choice
votes is not necessarily second-most popular, and the candidate with
the second-fewest votes is not necessarily the second-least popular.
Whatever makes single-mark evil needs explaining.
As a Condorcet backer I have to choke. As an example assume that
everyone considers V acceptable, and A, B, and C are each first choice
for 1/3 of the voters, If they all rank V as second choice then, for
each of the three groups, V will get twice as many V>x as x gets of
x>V. for being liked better than V.
Also we agree that "open primary" elections are not fair. In this
approach, the candidates who are identified as "most popular",
regardless of political-party affiliation, progress to the next
round. This approach fails to consider that the majority of voters
who support the most-popular candidate are likely to be the same
majority of voters who support the second-most popular candidate --
unless the counting method specifically compensates for this
redundant influence. The remaining voters, who may almost be a
majority, can end up with only getting to choose between the two
candidates who are preferred by the majority. Expressed another way,
the words "most popular" are ambiguous in the context of choosing
which candidates deserve to progress to another round of voting.
If I cannot kill having primaries I would vote against open.
----------- end ------------
I'm sure I'm missing some important additional considerations, but
they aren't coming to me at the moment, so I'll tap into your brains
to help refine this section.
Of course we aren't offering a fair way to handle French
presidential (?) first-round elections (in terms of which two
candidates should move on to the final runoff election), but we have
nothing specific we would agree on, right?
Richard Fobes
----
Election-Methods mailing list - see http://electorama.com/em for list info
