ternary operator vote

[Laura Creighton]

> Gareth> I strongly agree that something very like approval voting is
> Gareth> the way to go. However, I think pure approval voting isn't
> Gareth> quite right here. After all, everyone who's using Python right
> Gareth> now already regards the language we have as *acceptable*...
> 
> The term "acceptable" is vague, of course.  What approval voting
> really asks voters to do is to rank-order the alternatives, draw a
> line between the alternatives that the voter wants and the ones that
> the voter doesn't want, and vote for all of the alternatives above
> that line.
> 
> As Laura Creighton pointed out, that procedure has the disadvantage
> that many people will pick "no change" by reflex.  

I never said, 'by reflex'.  What I said was that people will vote
the ones they like or no change, because that is what they want.
Which means that only those who are hell-bent on getting any change,
and refuse to put 'no change' on the card, will not have 'no change'
listed.

Here is a quote from electionmethods.org.

     Critics of Approval voting point out that it does not allow
     voters to specify preferences among their selected
     candidates. But the option to specify preferences is of dubious
     value if those preferences are not reliably counted. In Condorcet
     voting all preferences are systematically counted, but in IRV
     they are not, and the uncounted preferences can be the most
     crucial. Although Approval voting does not allow voters to
     express all their preferences, it gives them more control than
     IRV over which preferences are actually counted. Hence, although
     Approval voting is not as effective as Condorcet voting, we
     believe it is actually more effective than IRV, even though it is
     much simpler.

Thus all the people who draw their line just below 'no change', ie 
most people, will have just cast a ballot where 'no change' is a winner,
and so would the people who vote no change first, and then their 2
least hated alternatives (if I had to I could live with this) and
then draw the line.  A vote which is A B nochange <line> and
B A nochange <line> and nochange A B line _all count the same_.

I confidently predict that no_change will be above the line in the
vast majority of Pythonistas votes.  You have just handed me the
election.

> The way to deal with
> that is to force everyone to state the full rank order and take it
> into account in a way that ensures fairness.
> 
> That's what Condorcet voting is about; see electionmethods.org for details.

I still don't think this is what we want.

I've been here.  This is how the 'Riotous Good Living' Party made sure
that all undergraduate Engineers were properly equipped with the
necessities of life (and you might be surprised what we considered
necessary).  'Budget Surplus' was not a word in the Riotous Good
Living Party's vocabulary.  Things worked fine as long as the people
who voted 'save it' did not mind so much that it nearly always got
spent.  When the Engineers suffered a major crisis, and had to fund
their own newspaper, that changed. This method of voting no longer
suited.  'Save it' was no longer just one alternative that you liked a
little more than 'more beer'.

We didn't use the beatpath method.  I will have to check and to see
if this makes a difference.  

Fortunately, there is this wonderful site: 
http://www.ericgorr.net/condorcet/

We can use it to stuff matrixes of imagined voting data into the
algorithm for Basic Concorcet, Beatpath Winner, and Ranked Pairs
and see if what we get out produces a result that we think is 
fair.

Can somebody go prepare some, and post and discuss the results?
I've got a paper to write, and besides, though I am working quite
hard to save the pro-ternary camp from a thorough trouncing at the
hands of the Approval voting method, I am hardly non-partisan.

Laura

> 
> -- 
> Andrew Koenig, ark at research.att.com, http://www.research.att.com/info/ark
> -- 
> http://mail.python.org/mailman/listinfo/python-list