[Election-Methods] IRV ballot is at least as fair as FPTP ballot

Kevin Venzke stepjak at yahoo.fr
Sat May 3 12:51:41 PDT 2008


Hi Stéphane,

--- Stéphane Rouillon <stephane.rouillon at sympatico.ca> a écrit :
> Again satisfaction analysis can be used to objectively determine which 
> of IRV and FTP
> produces the best outcome. Using enough election data, one could even 
> measure how often IRV may
> elect the candidate not favored by most voters. My humble estimation is 
> rarely (1/50 times).
> In comparison I estimate FPTP outcomes to be deficient (1/5 times) and 
> condorcet methods (1/200 times).
> I qualify a method to be deficient when another outcome would produce a 
> better global satisfaction.

To do this you would have to be clear and consistent with your
assumptions... Is "global satisfaction" the total utility (on some scale)
of the winner? When you guess that IRV fails by this standard 1/50 times,
are you considering real life elections, or random ones? If random then how
many candidates and is there any underlying policy space? etc.

Given real life elections I guess 1/50 may be accurate for IRV, but I don't
feel this tells the whole story. Strategic nomination and voting
incentives, as well as incentives created by institutions other than the
voting rule itself, would not seem to be considered at all by this measure.

Given random elections with even 3 candidates and the ability to truncate,
I guess IRV is much, much worse than 1/50.

Kevin Venzke

__________________________________________________
Do You Yahoo!?
En finir avec le spam? Yahoo! Mail vous offre la meilleure protection possible contre les messages non sollicités 
http://mail.yahoo.fr Yahoo! Mail 




More information about the Election-Methods mailing list