[EM] Re: What exactly is an "Election Method"?

[Paul Kislanko]

Bryan Ford in response to my post about splitting the analysis of methods
into two pieces:

>>You're quite right to point out that both IRV and Condorcet (and Approval
and 
most other "alternative" methods) critically depend on changing the
balloting 
scheme to something other than simply "choose one candidate".  But I don't 
think it's quite true that "_any_ improvement over plurality needs some 
[sort] of ranked-ballot input".  For example, the Candidate Proxy scheme
that 
has been discussed before on this list, as well as my very similar
Delegative 
Voting scheme (http://www.brynosaurus.com/deleg/elect), use ordinary 
plurality-style ballots but nevertheless I think at least represent a 
significant improvement over plurality (independent of how they might
compare 
to IRV or Condorcet).<<

Quite true. I would argue that from a theoretical perspective, though,
there's a different assumption behind proxy systems that essentially makes
the process different. Without passing any judgment, I just note that
defining the transfer of votes for a defeated candidate to one still in the
running to be handled by the candidate I voted for rather than something
that I explicitly entered still provides the same "additional information"
that can't be inferred from pure Plurality.

>>One of the main reasons I suggested Delegative Voting in the first place
was 
precisely because it requires no change whatsoever in balloting or vote 
counting technologies, or the attendant effort in re-educating voters in how

to cast ballots.  In that sense Delegative Voting or Candidate Proxy could 
represent a "path of least resistance" from plurality to something at least 
substantially better if not necessarily ideal.<<

I think this is worth more study. It is true that the ballots would look the
same, but the implications of the vote are different. It is not at all
obvious in my neck of the woods that most voters would choose "let a
politician decide how you'd have voted if you knew your guy was going to
lose" would be more acceptable than "Ok, who's your second choice?" 

My concern about proxy systems is there's an extra level of complexity that
might cause my sincere preferences to be different than in a direct
election. This may not be a bad thing but the effect would be hard to
analyze. Personally, I think it would make me more likely to make a "protest
vote" for a fringe candidate, but I think most people (around here anyway)
would tend to move back toward just straight party-line votes.

Regarding:

>So I would argue quite strongly that if you use a pairwise matrix as your
>tally mechanism, you allow each voter to fill in their pair-wise
>preferences. If you don't, the system is not "transparent" and can't be
>backed-up by voters' ballots. That would be a lot harder to sell (but it's
>trivial to implement)

>>Very cute idea - for us voting geeks anyway - but somehow I'm not quite
sure 
it would fly with the general public... :)<<

Whether the general public would adopt it would depend upon the
implementation. It is much easier to deal with pairs of candidates than it
is to accurately order a list according to one's sincere preferences. Take a
5-candidate election - I have 120 choices for my ranked ballot. If you list
out the 10 pairs and ask me to select X, Y, EITHER, NEITHER I can generate
the same information you have to infer from a ranked ballot in seconds and
you run no risk of inferring something that might not have been my
preference.

If I say NEITHER at any pair, the most I have to deal with is 4 questions. I
haven't had to go to the trouble of picking one of 120 choices for a ranked
ballot, weighing the implications in the method for each choice (and
certainly no one does this - by the time they're at the point of deciding
between 4th and 5th they are flipping coins or making random marks...) 

I think it would be easier for the voter to do it that way, and much easier
to analyze the effects of the tallying method without having to make
assumptions about the translation from 120 different ranked ballots into a
20-bucket set of pair-wise sums.