[EM] approval voting and majority criterion

[Ken Johnson]

>From: "Steve Eppley" <seppley at alumni.caltech.edu>
>Date: Wed, 17 Mar 2004 20:06:08 -0800
>
>Ken Johnson wrote:
>  
>
>>Forest Simmons wrote:
>>    
>>
>-snip-
>  
>
>>>A more fundamental goal might be to go with the choice that 
>>>would be acceptable to the greatest number of voters.
>>>      
>>>
>>That sounds like Approval. The method follows directly and 
>>obviously from the statement of the goal, no formal proof
>>required.
>>    
>>
>-snip-
>
>Except it's not obvious what "acceptable" means.  It's not 
>defined in the rational choice theory model of individual 
>preferences, nor do I see how to derive it within that model.
>
>Furthermore, I see no reason to believe that "sincere 
>acceptability" would have anything to do with the voting 
>strategies actually employed.  Each voter would have a clear 
>incentive to approve their most preferred candidate, regardless 
>of whether that candidate is sincerely acceptable, and would 
>have a clear incentive to not approve their least preferred 
>candidate, regardless of whether that candidate is sincerely 
>acceptable, etc.
>
>--Steve
>
>  
>
The theoreticians may not know what "acceptable" means, but the concept 
is clear to the average voter. It is defined in the dictionary as
    1. Worthy of being accepted.
    2. Adequate to satisfy a need, requirement, or standard; satisfactory.
This is a primary advantage of AV - the concept is simple enough that 
you can explain it to a third grader, and it is compatible with existing 
voting machines and processes. On the other hand, I think Forest's 
suggested definition of the goal of elections is overly simplistic 
because it neglects strategy and doesn't recognize differing degrees of 
acceptability. So back to my original question, what is an appropriate 
definition of "the goal of single-winner voting" by which the merits of 
alternative voting systems can be evaluated?

Ken Johnson