Why plurality is not "wrong"

From: David Mertz <voting-project_at_gnosis_dot_cx>
Date: Fri Apr 16 2004 - 17:45:44 CDT

Jeff has been pushing the notion that plurality voting is somehow wrong
technically, not simply something he disagrees with politically.

Jeff's notion is simply misguided--Jeff's criteria for what he would
consider the "correct" outcome is not uniformly shared by all voters
(or probably even most voters). I'm not talking here about how Jeff
feels about Nader vs. Bush vs. Gore, which was a famous real-world case
where different tallying methods would have produced a different
result. Even if presented with the Alice/Bob/Charlie-style
hypotheticals, opinions differ on the "right" outcome.

For example:

On Apr 16, 2004, at 2:50 PM, Jeff Almeida wrote:
> the Republican primary for Denton County
> Sheriff last month (and with no Democrat running, it basically decides
> the
> position) was a five-way race where the "winner" polled 38%, and second
> and third place were separated by all of about 19 votes. Had this
> been a
> plurality, he'd be our new Sheriff-elect come November. However, all
> three other candidates endorced the second-place finisher in the
> subsequent runoff (which was held Tuesday) and said "second-place
> finisher" won handily

There are a number of problems with Jeff's "obvious" example.

For one, within the information he gives, it is ENTIRELY plausible that
Mr.38% would have also won under Condercet, IRV, or CR systems. AFTER
the first round was over, the bottom three candidates endorsed
Mr.2nd-Place; but those endorsements probably affected later voting
behavior. On the ORIGINAL voting day, the voter ranks could well have
been scattered around in such a way that Mr.38% would have come out
ahead (absent the subsequent endorsement information that voters got
only later).

So maybe Mr.38% really is the "fair" winner for Sheriff. Not knowing
the names or county, I obviously have no real opinion on the merits of
the candidates, but abstractly, it is far from intuitive that the
actual (delayed runoff) result was the -fair- result.

Moreover, differing intuitions about the meanings of rankings can
change outcomes. Plurality is really a degenerate case of CR
(cardinality ranking).

The usual CR system does something like assign ten "points" to rank#1,
nine points to rank#2, and so on. Maybe that's fine. On the other
hand, maybe I feel that my #1 candidate I like a whole lot; all the
other ranked candidates I like in a certain order, but far less that
#1. In my world view, fair points might be more like:

        #1 100 points
        #2 50 points
        #3 25 points
        #4 12 points
        #5 6 points

I could still rank them, but rank #1 is the most significant,
logarithmically. Well, probably under this weighting, old Mr.38% would
have still won. Or maybe not. We can't know, since no ranks were
recorded in practice. Certainly, in the abstract, log-scale points can
produce different results than linear-scale points. Which is "fair"?
If you assume everyone REALLY wants their top choice, and indifferently
dislikes the others... well, that's plurality.

These questions are inherently political. They're not technical. And
OVC should get involved in them, except to implement what jurisdictions
actually want done.

Yours, David...

---[ to our friends at TLAs (spread the word) ]-----------------------
Iran nuclear neocon POTUS patriot Pakistan weaponized uranium invasion
smallpox Gitmo Castro Tikrit armed revolution Carnivore al-Qaeda sarin
---[ Gnosis Software ("We know stuff") <mertz@gnosis.cx> ]-----------
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
Received on Fri Apr 30 23:17:09 2004

This archive was generated by hypermail 2.1.8 : Fri Apr 30 2004 - 23:17:29 CDT