[EM] favorite betrayal and 2-party domination in Condorcet(wv, =); and about DMC

[Abd ulRahman Lomax]

At 09:47 PM 9/5/2005, Warren Smith wrote:
>So.  from the point of view of US third parties,
>1. all Condorcet methods plausibly lead
>to 2-party domination, though we cannot be sure,
>2. all are more complicated
>than range voting, and
>3. range voting apparently does NOT lead to 2-party domination.
>
>Therefore, US third parties should push range voting in preference
>to all Condorcet methods.   (Doing the reverse would be idiotic/suicidal, name
>your favorite adjective.)

Mr. Smith has substituted his opinion for the "point of view of US 
third parties." He has yet to convince them. It is those parties 
which are looking at the situation and, so far at least, they don't 
see what he claims they should see.

In another post, he acknowledged that the probability of a Condorcet 
method leading to 2-party domination might be very small.

It is idiotic to claim that a plausible point of view is idiotic....

I agree with Mr. Smith that Range, properly implemented (and which 
might be as simple as granularity 2 range, i.e., Approval), is a 
superior method for a number of reasons. However, either of these 
reforms (Range or Condorcet, say, DMC) would be a vast improvement; 
it is difficult for me to conceive that they would harm third parties 
more than the status quo; the choice may boil down to political 
expedience, and my own view about that is that Approval is 
technically a tiny change and may therefore be easier. The proof is 
in the pudding, however.

Mr. Smith's claim that Range is better for third parties than 
Approval is based on his comparison of Range and Approval based on 
vote percentages in his poll. However, he has assumed that Range 
percentage ratings are equivalent to the same percentage vote in 
Approval, which in some cases could be far from true. For example, a 
2% Range rating might mean that not one single voter actually 
approved of the candidate, whereas a 2% approval vote means that 2% 
of the voters *did* approve of the candidate. To be able to compare 
Range and Approval, the Range ballot must have an Approval cutoff, 
where the voter indicates what percentage is considered acceptable. 
You could then determine true candidate support, as distinct from 
ratings, which would correlate with support in ways that could vary 
wildly from voter to voter.

Some proposed Range methods do present the ballot in a way as to 
imply an Approval Criterion. For example, the -1, 0, +1 Range ballot 
does imply clear approval at 1 and probably insufficient support to 
be considered approval at 0. Higher granularity with appropriate 
ballot instructions could improve this.