How to vote in Approval

[Forest Simmons]

On Tue, 2 Apr 2002, MIKE OSSIPOFF wrote:

>
> Rob LG said:

<snip>

>
> Rob LG continued:
>
> When an election is definitely zero-information, finely-grained Cardinal
> Ratings is undoubtedly the champ. In fact, when no strategy is possible, I'd
> rather the sincere CR winner win than the Condorcet winner.
>
> [Mike comments]:
>
> But strategy is possible even with 0-info. CR is an excellent method, with
> 0-info or nonzero-info. I don't know if everyone will rate sinderely. In a
> 0-info public
> election, the voter who wants to maximize his/her utility expectation
> will give maximum points to all of the above-mean candidates, and
> minimum points to the rest.

This is true, but as Richard Moore once pointed out, if the CR values are
constrained by limiting the variance, then the best zero information
strategy is to use sincere strategy.  [The sincere CR values are adjusted
by dividing by their variance, so the relative CR values are unchanged.]

So Rob LeGrand isn't too far off.

Personally, I prefer the sincere zero info approval winner over the
sincere CR winner, because of the caulking of most trickle down leaks in
the real world.

Forest