[EM] Oops! MMPO with summed disapprovals as an opposition elects B in ABE. FBC/ABE roundup.

[MIKE OSSIPOFF]

Forest--

The Approval bad-example:

27: A>B
24: B
49: C

B's negative score is 49, the number who rank C over B, and also the number who don't rank B

C's negative score is 51, the number who rank B over C, and also the number who don't rank C

A's negative score is 73, the number who don't rank A

B has the lowest negative score, and wins.

You described a good, diplomatic way to avoid the ABE problem in high-res Score Voting. Of course, in public elections,
that strategy could be implemented probabilistically in Approval.

But that requires information about how many votes C will get.

To me, an FBC/ABE method is one that automatically avoids the co-operation/defection problem,
not requiring predictive information on the part of the A voters.

What kinds of FBC/ABE methods are proposed so far?

Three kinds:

1. MMPO and MDDTR:

Advantages: 

Simple and brief definition. A voters can unilaterally establish coalition defeat of C.

Disadvantage: 

Criticizable by Mono-Add-Plump or Kevin's MMPO bad-example. Those criticisms don't 
describe genuine problems. They don't amount to strategy problems for voters. They don't prevent the
electorate from getting changes that they want. But they could be used by opposition to distract
voters from the important considerations.

2. Conditional methods such as MMT, GMAT, AOC, MTAOC, MCAOC, AOCBucklin, AC, MTAC, MCAC 
and ACBucklin:

Advantages:

Avoids criticisms of #1. Though MMT doesn't meet Mono-Add-Plump, that criticism is
easily answered for MMT. The other conditional methods don't have even that criticism, or the
Kevin's MMPO bad-example criticism eitiher.

These methods are simple, and follow from Plurality and Approval in a simple, obvious and natural 
way. Their avoidance of the co-operation/defection problem, too, is obvious, simple, natural and 
straightforward, as is its motivation.

These methods can be offered as _options_ in an Approval election. For example, all of them other
than MMT and GMAT can be offered together as options in an Approval election.

For the methods other than MMT and GMAT, the conditionality can be optional by candidate.

Disadvantages:

One disadvantage: Chris doesn't like them.

Chris doesn't like them because he evidently doesn't like coalition or conditionality (though any
method which, in the ABE, defeats C only if the B voters co-operate is conditional too.

The rule-explicitness of the conditional methods' conditionality is what makes their avoidance of
the co-operation/defection problem simple, straightforward and natural.

Chris doesn't like MMT because of its noncompliance with Mono-Add-Plump,
though that criticism is easily answered.

3. Methods using Kevin's tied-at-top comparison:

Several have been suggested. No one has claimed that any of them have the desired properties.
They're all speculative. 

Advantages:

At least some of them elect A in the standard ABE. In other words, like MMPO and MDDTR, the A 
voters can defeat C by unilateral coalition support. They probably avoid the criticisms that are
used against MMPO and MDDTR.

Disadvantages:

Too complicated and wordy of definition.

They're only speculations, as of this time.

I suggest that the conditional methods are the winners.

Mike Ossipoff




 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20120111/7253458b/attachment-0003.htm>