[EM] Greatest Majority is the future of elections

[robert bristow-johnson]

On Sep 3, 2010, at 11:30 AM, Sand W wrote:

> I just joined this list and some of it is kind of bewildering.

i suspect that i am both a source and recipient of things  
bewildering.  but i like it better than ESF or any of the discussion  
lists associated with specific methods.

> 1-what is your goal for elections?  I would assume that it is to  
> have the best govt. which presumably can be identified as the system  
> of govt. supported by the most people.

it's an axiom of democracy, i think.  in the case of a simple binary  
decision, we give it to the simple majority (also the plurality or  
"... the guy with the most votes ...") for, essentially utilitarian  
reasons.  assuming every voter is equal (the true meaning of the "one  
person, on vote" concept, that's another place where the FPTP/runoff  
traditionalists are wrong and have usurped a label that doesn't really  
belong to them) and has equal weight, we give it to the majority  
candidate to minimize the amount of disappointment in the election.



> 2- Since there will probably be more than one exclusive/competing  
> policy/candidate supporter by more than a majority we can call this  
> goal "GREATEST majority voting."

one place where i disagree (philosophically) with Warren about Score  
Voting is that, quantitatively, we should not measure the degree of  
disappointment from different voters differently.  voters can take  
satisfaction when a candidate they hate is defeated and when the  
candidate they supported is.  there can also be a degree of  
satisfaction when a candidate other than the one they primarily  
supported is elected (or they could be neutral about it).  but when we  
add up the degree of satisfaction from all the voters, we should not  
be telling any voters that their satisfaction/disappointment counts  
any less than some other's satisfaction or disappointment.  if you  
tell voters that their effectiveness (in a race between the candidate  
they want to see elected and another candidate) is reduced by marking  
any candidate less extreme than 00 or 99, many might just decide to do  
that (and it's a strategic decision), and Score could devolve to  
virtually Approval vote.  and if the voter's favorite candidate is  
important to the voter, the motivation to bullet that favorite will  
devolve it more to Plurality (which is fine if there are only two  
candidates in the race, and nearly all of us think is not fine for 3  
or more).

> 3- Thus the goal would be determine the best or at least better  
> structure of govt./voting to require the greatest majority support  
> possible.

now it depends on what you mean by "greatest majority support  
possible".  if it's the same as the least disappointment amount of  
voters possible, then i might be able to say that, *if* there is a  
Condorcet winner, the choice of candidate with the greatest support  
(sorta "maximin" or the minimax of voter disappointment) is the  
Condorcet winner (candidates other than the CW have a max  
disappointment that is positive, only the CW is negative).

> 4-This would probably be a combination of local representatives and  
> executives elected by ranked ballots, "ranked pairs" if possible,  
> but that counting the system is not realistically practical and the  
> IRV counting system is realistically equivalent.

maybe.  in the case of 3 candidates in the Smith set, it's a pretty  
simple Rock-Paper-Scissors thing, every candidate has a positive max  
disappointment.  then considering only the max disappointment of the 3  
Smith candidates, the candidate that loses minimally to the others, is  
the one with the least minimax disappointment and a semantic case can  
be made that that's the candidate with the greatest majority support.   
that's not exactly what would happen in any general case of a  
Condorcet cycle (where there is more than 3 in the Smith Set), but i  
think that such is so unlikely, because i believe that cycles  
themselves would be uncommon if Condorcet were widely adopted; it  
seems likely to me that in the few cases that there is a cycle, nearly  
all would be between 3 candidates.  this is one reason why i might  
support RP over Schulze, because they don't elect different candidates  
unless possibly when the Smith set gets bigger than 3.

> I suggest the ranked voting forum for debate: http://4gmv.org/

why not bring the debate here to this forum?

--

r b-j                  rbj at audioimagination.com

"Imagination is more important than knowledge."