[EM] approval voting and majority criterion

[wclark at xoom.org]

James Green-Armytage wrote:

> Let's say that 55% of an electorate are to the left of the center,
> and 45% are to the right of center. If a method goes ahead and
> elects a right of center candidate anyway, then we know something
> has gone wrong.

11 - L:10,C:0,R:0
44 - L:1,C:10,R:0
45 - L:0,C:0,R:10

What's "gone wrong" in the above example?

I see a relatively small passionately-left-of-center group, a relatively
large basically-centrist-with-a-very-slight-leftist-tendency group, and
another even larger passionately-right-of-center group.  Why should the
near-indifference (with respect to L/R) of the smaller middle group
outweigh the higher dedication of the larger third group?

> Given two groups of different size with irreconcilable opinions on a
> given issue, any value-neutral decision-making rule must logically
> chose to execute the wishes of the larger group.

...not if additional information (such as strength of preference for the
opinions) is available.  Your statement only holds true for rank (and
similar) methods.

-Bill

-- 
Ralph Nader for US President in 2004
http://votenader.org/