[EM] Duverger's "law"

[Abd ulRahman Lomax]

At 04:54 PM 8/31/2005, Warren Smith wrote:
>For
>evidence about Duverger law see
>    http://math.temple.edu/~wds/crv/Duverger.html
>and if anybody wants to supply me with some more datapoints (e.g. 
>from 2002-2005 elections)
>then I can add them to the picture there.

Duverger's law isn't a law. From wikipedia:
While there are indeed many FPTP systems with two parties, there are 
significant 
<http://en.wikipedia.org/wiki/Counterexample>counterexamples: 
<http://en.wikipedia.org/wiki/Scotland>Scotland has had until 
recently first-past-the-post and similar systems but has seen the 
development of several significant competing political parties. Many 
commentators regard the 
<http://en.wikipedia.org/wiki/United_Kingdom>United Kingdom's 
<http://en.wikipedia.org/wiki/Liberal_Democrats_%28UK%29>Liberal 
Democrat party, since the 2005 
<http://en.wikipedia.org/wiki/General_Election>General Election, as 
forming a 'third party' and creating a three-party system. 
<http://en.wikipedia.org/wiki/Canada>Canada and 
<http://en.wikipedia.org/wiki/India>India have multiple regional 
parties. Duverger himself did not regard his principle as absolute: 
instead he suggested that first-past-the-post would act to delay the 
emergence of a new political force, and would accelerate the 
elimination of a weakening force - 
<http://en.wikipedia.org/wiki/Proportional_representation>proportional 
representation would have the opposite effect.

However, it is clear that what Warren is calling "plurality" does 
encourage the formation and maintenance of two-party systems. 
However, from my point of view, this could also be seen as a failure 
of third parties from understanding how to grow in the plurality 
environment. Quite obviously, if a party encourages its members to 
waste their votes by voting for the party's candidates, while the 
party is weak, it will find itself in the vicious circle that Warren 
describes. And, in fact, the most dangerous time may be as the party 
approaches parity. However, there are other paths to a strong third 
party, which is for a caucus to form within an existing party, and to 
exercise influence without necessarily wasting votes. It's been 
pointed out that, from a policy perspective, Southern Democrats at 
the time of FDR were really a third party.

My own analysis of the situation tells me that third parties largely 
fail here in the U.S. because they and their members generally take a 
naive view of what a third party could do. They can be partially 
successful in that elements from their platforms can become elements 
of the platforms of one of the major parties (or both). However, they 
could also become stable forces, if they recognize that there are 
many successes short of having the party candidate win an election. 
Consider 2000: I have often pointed out that Nader could have 
negotiated with the other campaign(s) for an endorsement and a 
recommendation that voters supporting Nader should vote for, say, 
Gore, and only for Nader if they really could not support Gore even 
as the lesser of two evils. In addition, he would have asked all 
voters who did switch their vote because of his recommendation that 
they donate at least a small amount, say $5, or whatever, to a 
campaign fund set up just for this purpose, to measure, in a very 
tangible way, support that was not expressed in a vote. (This would 
have more than replaced the lost federal campaign financing funds, I 
think). As to the argument that the votes help keep the party on the 
ballot, if a party has enough votes to actually win an election, then 
it surely has enough people to get a candidate on the ballot. (I'm 
not familiar with state law on this subject, it may vary.)

As suggested by the Southern Democrat example and by others around 
the world, third parties may also form due to regional interests; 
these parties would be likely to form coalitions in national elections.

Duverger's "law" creates a pressure, it does not necessarily control.

And the conclusion that Warren keeps pushing as if it had been proven 
beyond doubt, that Range Voting is the only choice for third parties 
and that they had better push it or they might as well pack up and go 
home, has not been show, neither by his study nor by his additional 
arguments. At least not so far.

Range, *properly implemented*, could indeed help third parties, but 
so could Approval (granularity-2 range, not really a different method 
but Warren often treats it as such), Asset Voting (which might be 
legally difficult in public elections but which might work well in 
primaries and thus see possible wider use later), Condorcet, or, as 
is in fact shown by the chart on Warren's cited page, any kind of PR.