[EM] Primaries?

[Adam Tarr]

There's really only one issue left worth discussing in this thread, so I'll 
cut the rest out:

Dave Ketchum wrote and I responded, et cetera:

>>>>>>>      Puzzle:  Assuming the above leads to Condorcet in the primary, 
>>>>>>> to select two candidates for the general election - WHY NOT?  the 
>>>>>>> arguments are not necessarily the same as related to electing two 
>>>>>>> officers for PR.
>>>>>>
>>>>>>Not necessarily, sure, but I don't think that Condorcet is clearly 
>>>>>>the best method to elect two candidates.  It seems likely that it 
>>>>>>would end up picking two candidates from the center of a party, and 
>>>>>>nobody from a wing (think Kerry and Edwards, in stead of Kerry and 
>>>>>>Dean).  But there have been some stabs taken at Condorcet-flavored 
>>>>>>proportional representation.  The best attempt is probably this one:
>>>>>>http://groups.yahoo.com/group/election-methods-list/message/10308
>>>>>>It's pretty complicated, but worth the read.  Try to sell that to the 
>>>>>>public, though...
>>>>>
>>>>>As I said above, we are not doing PR, so almost certainly would not 
>>>>>find such complication worth the pain.
>>>>
>>>>Probably not, but this does not imply that pure iterative single-winner 
>>>>is the best approach, either.  A good compromise (in my opinion) would 
>>>>be the sequential variant of the method described in the link.  So, 
>>>>first you find the CW, then you find the best two-candidate slate with 
>>>>the CW in it, then you find the best three-candidate slate with those 
>>>>two candidates in it, and so on until you've generated as much of the 
>>>>order as you need.
>>>
>>>You seem determined to add unneeded complications.
>>Well, there's complications, and there's unneeded complications.  The 
>>simple fact is that, while Condorcet is an excellent (I would say the 
>>best) method for choosing one winner, it is a terrible method for 
>>choosing two or more.  It is likely to produce a set of "clone" 
>>candidates, all representing the center faction of the electorate.
>>Using some method of PR, no matter how crude, is more likely to get good 
>>results.
>>I'll happily agree that sequential CFPRM is still really complicated, and 
>>probably more complicated then we need.  But how about single 
>>non-transferrable vote?  How about cumulative voting?  How about STV-PR?
>>All of those are well-understood and relatively simple methods, and all 
>>would be dramatically better than using Condorcet to elect two or three 
>>people.  As I said, I think Condorcet is the very best single-winner 
>>method.  But would a Democrat who supported Howard Dean or Dennis 
>>Kucinich be happy if the two Democrats in the general election were Kerry 
>>and Edwards?  Because that would be the most likely outcome if you use 
>>Condorcet.
>>We surely don't need perfect PR in a primary, but some degree of PR is 
>>necessary, or producing more than one winner is often almost pointless.
>>-Adam
>
>AGAIN, we are NOT doing PR, and it is not clear to me that what might look 
>to you like clones have to be a bad thing.

So, you think it would be acceptable, even desireable, to have three 
candidates from a party's centermost faction enter the general election, 
even when the party has a broad range of viewpoints?

Say (somewhat but not totally arbitrarily) that the Democratic candidates 
for president could have been ordered from least liberal to most liberal as 
follows:

Lieberman<Clark<Kerry<Edwards<Gephardt<Dean<Kucinich<Brown<Sharpton

Furthermore, let's assume that Kerry was the Condorcet winner, and 
furthermore let's assume that preferences by most voters follow a 
predictable pattern of moving away from their favorite.

If we use successive Condorcet elections to pick three nominees, then the 
other two winners (other than Kerry) will be either Edwards and Gephardt, 
Edwards and Clark, or Clark and Lieberman.  This leaves the entire more 
liberal third of the Democratic electorate without any representation, and 
it leaves the candidate with the second-strongest core support (Dean) 
eliminated.

This shouldn't bother us in a single-winner situation, but in the case of a 
multi-winner election it begs the question of why we are even bothering 
with electing more than one nominee, if all we are doing is getting the 
three most similar candidates.

Moreover, this could constitute really poor strategy.  If the general 
electorate has different preferences than the primary electorate (obviously 
very likely) then producing a slate of candidates with diverse stances is 
more likely to produce a good result for the party, in general.

Again, none of this argues that perfect PR is needed in a primary.  But 
completely ignoring PR issues is a mistake, too.

-Adam