[EM] Effect of Voting Systems on Parties and Candidates

[Greg Nisbet]

Let's hypothesize about the impact various methods would have on society.

FPTP: If you live in the U.S., you see it every day. Two party domination is
fairly complete. Although this could be written off to the fact that America
started with a two-party system and that opposition is gerrymandered out of
existence. There are also annoying ballot access laws and sore loser laws
and whatnot that contribute to this. It isn't just our horrible voting
system. Anyway, main impact of this: two strong parties, a polarized
electorate and parties that only campaign in certain, rare competitive
regions.

IRV/STV: Two party domination too. It doesn't appear quite as bad because
parties in Australia at least appeaer to cooperate with each other through
vote swapping agreements. Full preferences and later-no-harm allow voters to
express these opinions without penalty, although their value is dubious.

Borda: Used in Kiribati, Nauru, and Slovenia at one point in history.
(Slovenia still uses it for their 2 minority members.) I couldn't get a hold
of anything for Slovenia's contests (oh well), but Social Choice in the
South Seas http://rangevoting.org/ReillySCSS.pdf explains the impact fairly
well in Kiribati. The most popular candidates were eliminated by political
maneuvering and inhabitants were annoyed (the country backslid to FPTP). It
is naive to think it will encourage candidates to cooperate at all, collude
yes, cooperate no. It will lead to one party domination probably as rich
parties use their "momentum" to crush any second parties.

Condorcet:
http://en.wikipedia.org/wiki/Schulze_method#Use_of_the_Schulze_method These
people use Schulze... but I'm not certain how it has impacted them. Even
though they don't satisfy later-no-harm, it seems pretty clear that Schulze
would encourage candidates to cooperate. You have to give your second choice
vote to someone, and it isn't likely to cause you lose if you give it to a
particular candidate (unlike Borda), so Condorcet politics would probably
form broad coalitions of parties that cooperate on various things.

Range: It is a positional method, but candidates can both benefit if they
support each other. E.g. if A and B agree to vote each other fairly highly
and attack C, both A and B benefit. I regard it as less likely to foster as
much competition as Condorcet, but it probably would have more overlapping
campaigning. With multiple parties, each could gain something from
campaigning in exactly the same area.

Approval: similar to Range, but less dramatic.

Bucklin: It's been done. Bullet voting galore. Massive backsliding. Voters
giving a second preference shot themselves in the foot.

Contingent Vote: It's been done in Sri Lanka and London Mayoral elections.
I'm not quite sure what the impact is, but I don't anticipate much
difference from FPTP. People probably won't waste their precious vote if
they use a truncated version of contingent vote. If the full version is
used, I anticipate better results, similar to TRS. It probably will
encourage cooperation, later-no-harm and all.

TRS: Doesn't lead to two party politics. It can produce very weird behavior
like Chirac vs Le Pen instead of Chirac vs Jospin. If you check the Range
Voting archive of weird behavior, TRS indeed has its problems. However, it
doesn't discourage the growth of new parties and the Range voting website
does claim that it does produce some positive effects in media coverage
relative to IRV. TRS obeys later-no-harm, so it will probably encourage
cooperation. Candidates will ally the top two finalists and campaign for
them, probably.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20081016/264175e4/attachment-0002.htm>