Should points equal the number of voters? (was Re: Housekeeping

[Steve Eppley]

Donald D wrote:
-snip-
>   ONE: Why do I want the Coombs' points to be equal to the number
>        of voters?
>
>QUESTION ONE answer: If all the voters are equally represented in
>the Coombs' vote then the Coombs' vote will be equal to the number
>of voters.  Equality in representation is "relevant to democracy".

Donald appears to be inventing a new definition of "equality in
representation", which is normally the concept behind proportional
representation and has no relevance to single-winner elections. 
There's no compelling reason which supports Donald's redefinition.

>If the Coombs' tally is less than the number of voters that means
>that some voters are not represented in deciding which candidate to
>drop. 

Assuming we're not counting as "voters" those people who ranked no
candidate for this particular office, this case isn't possible in
either of the two variations of Coombs we're talking about.

>If the Coombs' tally is more than the number of voters that means
>some voters are represented with more weight than others. Some
>voters could have two or three or more times weight than others.

The "one person, one vote" principle means no voter may cast multiple 
votes for (or against) any candidate.  Allowing a voter to cast one 
vote for (or against) multiple candidates is *not* a violation of 
"one person, one vote."

This voter's ballot violates "one person, one vote":
   Dole=    2 points
   Forbes=  0
   Clinton= 0

This voter's ballot does NOT violate "one person, one vote":
   Dole=    1 point
   Forbes=  1 point
   Clinton= 0

Consider, for example, the way citizens commonly vote on initiatives
on a state ballot using the Approval voting method.  Sometimes there
are conflicting initiatives on the ballot; when that happens only
one of the rivals may pass, and the one which passes is the one
which is most approved.  (I.e., the one with the largest sum of
Yeses minus Nos.)

Here in California, we had some rival initiatives on the ballot in
November 1996.  There were two health care "reform" initiatives
(propositions 214 and 216).  At most one of these could win, and I
voted Yes on both of them.  Did I violate "one person, one vote"? 
Not at all, since I voted only once for each of them.  What would
happen if almost every voter voted Yes on 214 and Yes on 216?  The
number of Yes votes in that single-winner election would be nearly
twice the number of voters.  This might bother Donald's aesthetic
sense, but not a criterion of democracy.  Not "one person, one vote."

>     The Coombs' tally and the number of voters must balance -
>something like double entry bookkeeping. 

No, they don't have to be equal to satisfy "one person, one vote."

>Otherwise we could have results something like having more votes
>cast in an election than there were voters. If we had an election in
>which two hundred voters went to the polls but three hundred votes
>were cast we would be correct in saying that something was amiss. 

No one has been going around claiming there is an unmatched number
of votes in the ubiquitous Approval elections.  This is because
"total votes should match the number of voters" is a bogus criterion,
having nothing to do with democracy.

>Our interest in having the votes cast be equal to the number of
>voters would be more than being "aesthetically pleasing on the
>surface". The same is true with the Coombs' vote - it is a question
>of being honest.

If Donald were correct that something would be amiss by not
splitting a voter's equal votes, then surely many people would be
demanding that my votes for Props 214 and 216 be split, giving only a
half vote to each.  Obviously, no one has been criticizing Approval
as dishonest, so obviously Donald has made a mistake.

---Steve     (Steve Eppley    seppley at alumni.caltech.edu)