[EM] possible improved IRV method

[eric at ericgorr.net]

Quoting Eric Gorr <eric at ericgorr.net>:

> Allen Pulsifer wrote:
>> Going back to the example I gave, the Condorcet majority that elected D was
>> made up of everyone who did not rank A the highest, i.e., the Not A's, and
>> they all coalesced around D.  In reality a coalition like that would never
>> happen and if it did, it could not be held together.  In fact, in the
>> example I gave, the only thing that allowed D in the party was the A voters:
>>
>> 166:A>B>D>C
>> 166:A>C>D>B
>>  83:A>D>B>C
>>  83:A>D>C>B
>>   etc.
>>
>> If only a small portion of those voters realized that their strongest
>> head-to-head competitor was Candidate D, and then proceeded to downgrade him
>> to last place, then under any system D would be shut out and A would win.
>> That is what would happen in reality, and IMHO, in order to have healthy
>> governance, that is what should happen.
>
> This is a wonderful example of the fundamental flaw with this class of
> argument that I mentioned before.
>
> Why assume that only the A voters would attempt to manipulate the
> election by changing their votes?
>
> If the B & C voters, who certainly prefer D over A, get a hint that the
> A voters are attempting to manipulate the election, they can bury A and
> promote D and give D the victory - 500 to 499, which was the sincere
> outcome to begin with.

Of course, if things go really wrong, the A voters could end up giving  
the victory way to candidate C or B. For example, if the A voters all  
bury D and then B&C voters would see this promotion as an opportunity  
to win by burying A. Then, if the D voters, knowing they can no longer  
win and being rather upset with the dishonesty of the A voters, also  
bury A, you can end up with something like:

166:A>B>D>C
166:A>C>B>D
  83:A>B>C>D
  83:A>C>B>D
  83:B>C>D>A
  83:B>D>C>A
  83:B>D>C>A
  83:C>B>D>A
  83:C>D>B>A
  83:C>D>B>A
  2:D>C>B>A
  1:A>B>C>D

Granting C the victory as the new Condorcet Winner.

Which, of course, brings me back to another point I made earlier that  
when people start to vote strategically, the potential for _anyone_ to  
win in such close elections increases dramatically...which, again, is  
leading me to a currently unprovable conclusion that a good Condorcet  
Method encourages sincere voting.