[EM] Re: Minimal Improvements on Approval

[Kevin Venzke]

Mike,

Thanks for the reply.

 --- MIKE OSSIPOFF <nkklrp at hotmail.com> a écrit : > 
> You're probably familiar with this Approval improvement:
> 
> Voters cast an Approval ballot, but also vote for a 1st choice. If any 
> candidate is 1st choice of a majority, s/he wins. Otherwise the Approval 
> winner wins.

Yes.  I call that "MCA," although usually it's suggested here that any number 
of 1st choices be allowed.  The advantages of that change are obvious, in terms 
of sincerity.  But when that change is made, I think the method is nearly reduced 
to Approval:

>>I believe
>>this because the middle slot ("approved") is only useful if you think your
>>preferred candidates can win if they have an immediate majority, but not
>>otherwise.  I can't imagine a scenario where that makes any sense.

> Bucklin uses rankings, but is probably the best of the easily-counted rank 
> methods. Bucklin is probably the most easily-counted method that meets SDSC.

I agree with this.  I don't think Bucklin is good enough, though, because of
its clone problems, and because it is pretty bad at honoring pairwise contests.

> Your example:
> 
> 7 A>B
> 5 B
> 5 C
> 3 D>C
> 
> SSD, RP, BeatpathWinner/CSSD, Borda,  & Bucklin all choose B.

Yes.  I use MinMax scores as a rule of thumb.  I wouldn't accept any winner
other than B here.  The central problem (once it's granted that the approach
should be that the FPW and AW will be "finalists") is how to keep an unworthy
FPW from winning, when he beats the AW pairwise.

> I haven't adequately replied about the method proposals in that message. 
> Which ones would be the best ones to check out? What are the properties of 
> the better ones?

They all lose Approval's good properties of FBC and Participation.  (MCA keeps
FBC.)  They all intend to gain "some" Condorcet efficiency.  My hope is that
they're easier to hand-count.

The methods that I consider neck-and-neck are what I labeled "Withdrawable
Approval" and "No Worse Losses."  They're very similar in effect.  Let me
paste the definitions here:

Withdrawable Approval (WA):
>>[On the first count, find the FPW and AW.]
>>On the second count,
>>count FPW>AW (pairwise) votes as bullet-voting for the FPW.  (They are trying
>>to make the FPW win.)  AW>FPW votes are counted towards every approved candidate
>>except for the FPW (if he was indeed approved), because they are trying to get
>>anyone but the FPW to win the second count.  AW=FPW votes vote for every approved
>>candidate.  If the FPW wins, he is elected; if anyone else wins, the AW is elected.

No Worse Losses (NWL):
>>Elect the FPW if he beats the AW, and this victory is stronger than any
>>defeats suffered by the FPW to anyone else.  (Else elect the AW.)

Some differences:
1. NWL always elects the FPW or AW if they're the CW.  My simulations tell me that 
WA doesn't always, but I think it has to do with its treatment of equal rankings.
It might be fixable.

2. WA is better at counting (e.g.) FPW>AW ballots the way the voter would want.  
Someone who votes C>FPW>AW does not want C to beat the FPW pairwise, but that ballot
could do that in NWL.  (A mitigating factor, though: C>FPW has to be stronger than
FPW>AW in order to sink the FPW, and this ballot contributes to both strengths.)

3. I'm not sure which is harder to hand-count?  Keeping in mind that for NWL
the counter needs only to find the FPW's pairwise contests, and needn't count
candidates who could not possibly beat the FPW pairwise (based on approval, for
instance).

4. My friends do not think WA is intuitive.  They like NWL a lot better.


I would only consider "FPW & AW//MinMax" if it's considered unacceptable to
elect the AW when the AW has a worse MinMax score than the FPW.  I can accept
that.  And I wouldn't want to hand-count this method.

I suggested "Majority FPW vs AW" (i.e. FPW must have a majority-strength victory),
but I'm not so wild about it because 1. it's lousy at respecting an FPW>AW victory
even when the FPW is e.g. the CW, and 2. it doesn't succeed in guaranteeing that
the AW will have a better MinMax score than the FPW.


So I would plug the "No Worse Losses" method if I had to choose.


Kevin Venzke
stepjak at yahoo.fr



	

	
		
Yahoo! Mail : votre e-mail personnel et gratuit qui vous suit partout ! 
Créez votre Yahoo! Mail sur http://fr.benefits.yahoo.com/