[EM] difficulty of interpersonal comparisons in utility

[Brian Olson]

On May 22, 2004, at 9:54 AM, Kevin Venzke wrote:

>  --- Brian Olson <bql at bolson.org> a écrit : >
>> Well said. This argument goes back at least as far as the canonical
>> work by Kenneth Arrow. In laying the axioms on which his conclusions
>> lay, he argued that you can't compare utility _between_ people.
>>
>> I say otherwise. We do implicitly compare utility between people. We
>> declare them all to be equal. That's why the ideal has been "One Man,
>> One Vote". (Unless you're a shareholder where the system is 'one 
>> share,
>> one vote')
>
> Do you mean to say that everybody being equal shows that we implicitly
> compare utility between people?  And that this shows that election 
> methods
> can do so?

Yes, we implicitly balance interpersonal utility.

An election method could do so providing it had some way of recording a 
person's utility and provided that they have accurate self-knowledge, 
accurate knowledge of the options available, and incentive to record 
their utility honestly on the ballot.

>> Given that we _can_ compare utilities between people, Rating systems
>> become the natural basis, rather than Ranking systems.
>
> If the above is sufficient to show that we can compare utilities 
> between
> people, why is it more natural to have a rating system than a ranked 
> one?

A rating would be a more natural expression of Expected Utility derived 
from a candidate. Rankings can only express the utility Borda-style 
point assignment would use to approximate candidate utilities, OR 
rankings express "greater than" relationships between the expected 
utilities of the candidates, but have no expression of the relative 
magnitudes of the expected utilities.

Brian Olson
http://bolson.org/