[EM] IMHO, IRV superior to approval

[Brian Olson]

On Jun 5, 2004, at 12:20 PM, Stephane Rouillon wrote:

> It may not fit "1-person 1-vote", but clearly IRV lets all voters 
> express a
> preference at the last round in the sequential version,
> while approval restricts that "right or opportunity" to voters who 
> placed their
> approval cut-off between the last two contenders.
> You can say that the exclusion is random and thus it meets even "equal 
> power to
> voters", or better "equal esperance (in the
> mathematical sens) to voters", but it is a symptom of a less effective 
> selection
> process, and thus a less effective electoral method to me.
> If we assume a voter is "right" 51% of the time, using a random set of 
> 90% of the
> voters can only diminish the probability to select the best option.
>
> Please convince me, I always try to see things differently.
> Steph

In my simulations of zero-information honest voting, Approval achieved 
higher social utility than IRV.


Strategic voting and voting strategy aside, Approval for all it's 
efficiency still leaves me unsatisfied with the complaint "I want to 
approve X and Y but I  want to say I like X better than Y!" I think 
that psychological factor would keep me from getting too cozy with an 
approval ballot. (And since the world is like me, it would be a 
stumbling point for them too. ;-) )

I've always felt though that the argument about the 'likelihood of 
being the deciding vote' is irrelevant. Votes are commodity, not 
unique. Once it goes into the ballot box there is no 'my vote'. They 
are unordered and faceless.

To somewhat reverse what I said above, if an Approval ballot is taken 
and it comes down to two choices I approved of, then I should be 
reasonably happy. If it comes down to two choices I disapproved of, 
then I have participated in the system and tried to elect someone else 
but lost.

Somewhere it must have been said that straight Cardinal Ratings is 
equivalent to Approval given the proper voting strategy. It's in my 
best interest to saturate my votes to the maximum and minimum rating 
allowed. If that is even at all representative of my sincere 
preferences then it's much more important to get one of those I approve 
(and not get one of those I don't approve) than to distinguish within 
them.


Brian Olson
http://bolson.org/