[EM] Bucklin strategy (was: Fwd: [ESF #1126] Another bullet vote argument)

[Abd ul-Rahman Lomax]

At 10:48 AM 5/14/2010, Jameson Quinn wrote:
>Basically, I'd summarize my feelings about Bucklin in light of this 
>strategy analysis as:
>
>Bucklin is no more than sugar-coated approval. And that makes it the 
>best system available in a wide range of circumstances.

Probably. That is, more or less, my point. But hybrid systems can be 
designed that allow improvement over raw Bucklin. I don't see 
straight Approval as being politically practical, even though it is 
the cheapest useful reform.

It isn't merely "sugar-coated," though. The additional flexibility 
isn't merely "empty calories."

It is, rather, itself, a hybrid system, a hybrid between approval and 
ranked systems and a similarity to simulation of repeated ballot 
without eliminations.

With a good ballot design, Bucklin can allow both Range and Condorcet 
analysis, and I believe that the rules can be designed so that normal 
"strategic voting" encourages sincere disclosure of preferences, thus 
the Range and Condorcet data should be good.

Thus, for example, if Bucklin is used in a runoff system with good 
runoff rules, a Condorcet winner might never be passed over without 
the knowing consent of the electorate (in which case the candidate is 
no longer the Condorcet winner). I've argued that in a runoff between 
a solid Range winner and a condorcet winner (from the primary), the 
Range winner will have the advantage, normally, because of 
differential turnout and motivation.

(The disparity between the winners requires that the preferences 
creating a condorcet winner be weak compared to those for the Range 
winner, but the supporters of the Condorcet winner outnumber those 
supporting the Range winner. In this situation, two factors help the 
Range winner: it's easier to convince someone with a weak preference 
to change their vote, and those with a weak preference between the 
two candidates are less likely to be motivated to vote. But, on the 
other hand, it is possible that normalization error caused the 
*appearance* of a different Range winner, which is why such a 
conflict might be presented to the voters in a runoff. We don't 
actually know until it's tested.)