[EM] Re: Efforts to Improve on CR's Strategy

[Brian Olson]

On May 20, 2004, at 6:03 PM, MIKE OSSIPOFF wrote:

> Here's a conceptual example that I think better illustrates the problem
> that I observed. Suppose you vote in an election in which there are 6
> candidates and you have no idea how anyone else votes. Your sincere CR
> profile for candidates A ... F is
>    SincereCR: A(0.7), B(0.5), C(0.3), D(0.1), E(-0.1), F(-0.3)
> (This assumes signed CR's, with an approval cutoff of zero.) What I 
> call
> "ExaggerateCR" simply applies a linear transformation so that the max
> and min CR's are +1 and -1:
>
> I reply:
>
> If you do anything other than mutliplying all of a particular voter's  
> ratings by the same factor, then you'll get something that's 
> meaningless.

I think the operation being applied to each rating of a voter is
f(r) = m * r + b

By choosing m and b for voter, the highest rating scales to 1.0, the 
lowest to -1.0 and everything else proportionally in between.

> And if you multiply different voter's ratings by different factors, 
> then you don't have a valid CR count. Not that you're necessarily 
> doing a CR count.

It's not straight-CR, but it's still useful. I'd say it makes sure 
everyone has the same voting strategy for CR, which adds a measure of 
fairness. This particular variation can still be taken advantage of. 
The proper vote is 1.0 for all choices with positive utility and -1.0 
for all others. That maximizes my expected utility. But, the experiment 
as I understand it was applying various voting systems to honest 
preferences.

Brian Olson
http://bolson.org/