[EM] Majority Criterion poor standard for elections

[Abd ul-Rahman Lomax]

At 12:41 PM 10/24/2006, Simmons, Forest wrote:
>It seems to me that if there is a majority winner, then she should 
>at least have a chance of winning.  What if we chose by random 
>ballot from among all of the candidates that have a majority beat 
>path to the Range winner (with a final approval vote to ratify this choice)?

On the Range Voting list, I suggested that if the Range winner is 
different from the preference winner, that the Range winner be 
subject to a ratification vote. ("Shall the Range winner assume the 
office, Yes/No.")

If the motion fails, the Range winner is *eliminated*. The election 
is held again.

An alternative which I consider less satisfactory would be to have a 
runoff election between the Range winner and the preference winner. 
Some will argue that the preference winner will necessarily win this; 
however, this is actually not so obvious. In a two-person race where 
the voting public has knowledge that the Range winner is likely to be 
a unifying leader ("maximizes overall satisfaction with the election 
outcome:), those who had a weak preference for the Preference winner 
may very well change their votes....

In both of these procedures the preference winner "has a chance." 
Essentially, what these procedures would do is to provide a means of 
rectifying what might be called a Range failure, perhaps do to some 
quirk of the Range method. Perhaps the so-called 25% rule -- a rule 
which I consider inadvisable, but which Warren supports and which is 
still, I think, given as part of some "standard" Range method -- 
resulted in the election of someone who was not rated by a large 
number of voters, perhaps it was a write-in candidate or someone unknown.

(All this results from the possibility that blank votes will not be 
included in the Range averages. Essentially, there are two kinds of 
Range: sum of votes, and *average* votes. They are different if 
blanks are neglected, not included in the averages.)

But random ballot as a method simply reduces the expected utility of 
the election by the chance of the election of the preference winner 
times the difference in Range utility. In other words, it, on the 
average, will lower satisfaction with the outcome.