[Election-Methods] Smith--When is extreme rating suboptimal?

[Abd ul-Rahman Lomax]

At 06:11 AM 7/26/2007, Michael Ossipoff wrote:

>I forgot to include this in "Smith exposes our false statements":
>
>Smith said that sometimes extreme rating is suboptimal in RV. So, Smith,
>when is that so?

First of all, the original statement was general, and so Smith simply 
gave a simple counterexample from a small election, a common practice 
on this list. Ossipoff does it himself. But now he specifies his 
statement more narrowly, and Smith's example, therefore, no longer 
applies -- if I remembered it correctly.


>We're talking about public elections, where there are so many voters that
>your own votes won't change the probabilities enough to change your
>strategy.

I've given a case, fully specified. How many times to I have to 
repeat it. Many voters. Zero knowledge, which means that all the 
probabilities are exactly equal. Three candidates, Range 2, our voter 
has utilities of 2,1,0.

Optimal voting strategy: Sincere. Expected utility of Sincere: 40/27, 
of Approval, 39/27. Not voting: 1.0.

>Smith should probably check the consensus before shooting his mouth off. Or
>at least refer to quotes that have been posted.

With a counterexample, we do not worry about consensus, do we? The 
emperor has no clothes.