[EM] Pseudo-election reform in California

[Brian Olson]

On May 31, 2004, at 6:47 PM, Dr.Ernie Prabhakar wrote:

> One of my favorite columnists, Dan Walters, is talking about a new  
> approach to California's politicized, gerrymandered primaries:
>
> http://www.sacbee.com/content/politics/columns/walters/story/9490344p 
> -10414306c.html
>
>>  The new plan would have all candidates for office run on one ballot.  
>> If none got an outright majority in the primary, the top two  
>> vote-getters, regardless of party, would vie in a general election  
>> runoff. The system is very similar to the way California fills  
>> nonpartisan local offices and vacant legislative and congressional  
>> seats in special elections.
>
> Has anyone heard of this?  It basically sounds like it eliminates  
> parties, but is still based on plurality.  Any thoughts on whether it  
> is likely to work, and/or encourage 'median' candidates?  The  
> justification for multiple rounds, I suspect, is that primary  
> campaigns will still tend to be lower-profile, and more sectarian,  
> compared to the fall general election.

For the 'local offices' which this already applies to, this made a  
difference in the recent election of the Mayor of San Francisco. The  
top two were the Democrat Party and the Green Party candidates. The  
Green guy very nearly won the whole show. IIRC, the Green candidate had  
the highest first round votes, with something like 35% G, 25% D, 15% R  
(+ others). Then in the (manual) runoff, it went pretty close, perhaps  
a bit closer than 55% D, 45% G.

Internationally, something like this system was applied in France  
recently, and it got the whole country engaged because the #2 position  
was a pretty radical and in the end widely distasteful candidate. The  
runoff came out 80+% to about 10%. Now, we certainly didn't need a  
runoff to tell us that. There could have been better choices for a  
best-two runoff. Better choices would have extracted more information  
in the second balloting about what kind of candidate the people wanted.  
Better choices would have been closer in results.

> At any rate, it is far from ideal, but still sounds like a huge step  
> forward.

I'd say 'tiny' rather than 'huge'. I should add best-two-runoff to my  
sims to check that. I bet it's worse than IRV. It's probably  
analytically provably worse than IRV.

> If they *were* willing to consider rank-order voting of some kind,  
> what would be the optimal method to use for selecting the top-two for  
> a runoff?

Eh? I still think one of our Advanced Election Methods selling points  
is that of the single balloting. No expensive, wasteful re-balloting  
runoffs. That's probably also a conceit the IRV advocates use "You  
understand what a runoff is, but those are expensive. Now you can have  
(ta dah!) Instant Runoff!"

Brian Olson
http://bolson.org/