2009/2/24 Thomas Lord <[email protected]>:
> It's sort of funny that all of the
> information that goes into the results
> of the election has been available for
> a while now yet nobody has bothered
> to both run the program to find the outcome
> and posted about it, even if unofficially.

Well actually, I did run it, but more for experimental purposes as I
am engaged in a side debate over voting systems :) I noticed at least
one odd anomaly in the three-seat case (for our ballots) that takes a
bit more space to elucidate than I really felt like writing up. But
since you asked...

If you follow the process of the count through the transcript that
Alan's code provides, you will notice that Marc Feeley is elected
outright with a full quota on the first round. Will Clinger came in
second by the number of first-preference votes *but* Jonathan Rees
turns out to be the second one elected on transfers. And in fact,
Clinger leads for second place right up until Anton Vanstraaten is
eliminated, electing Rees on the penultimate round.

At this point, one may well ask, "So what? Isn't that the way it's
supposed to work?", and indeed that's just fine, until you start
looking at other sizes of the SC (which I was doing because I was
curious about who might be on an expanded SC). So just for grins, I
ran a two-seat election, theorizing that perhaps Feeley/Rees should be
the result based on interpreting the previous count.

I was wrong. Feeley was still the first elected, but not until the
fifth round. And Clinger filled the second seat, leading all the way.
There is a sensitivity to the number of seats (and consequently the
initial quotas) in the counting algorithm that I found moderately
surprising.  And after running all the elections up to seven seats,
each larger elected set was a proper superset of the immediately
preceding, smaller election, so I suppose you could say that the
ballots and the algorithm determine a total ordering of the candidates
by the community, but it seems less psychologically true than
mathematically true. Also I would have to wonder how much people's
conception of the committee size affected their votes *and* how much
the short ballots (and there were a few that did not rank all the
candidates) affected the outcome.

My conclusion: that my interlocutor who has been badgering me about
Restricted Range Voting has a point. There are some pathologies in
STV, and we may actually have seen them (e.g. partisan voting effects
in the Rees/Clinger count-off) in this election.

And for the terminally curious (but lazy :) here are the results for
all the election counts I ran. The names are in the order that the
members are chosen by the counting process.

2 seats: (feeley clinger)
3 seats: (feeley rees clinger)
4 seats: (feeley clinger shivers rees)
5 seats: (feeley shivers clinger rees hanson)
6 seats: (feeley rees clinger shivers hanson vanstraaten)
7 seats: (feeley clinger rees shivers hanson vanstraaten dybvig)

Raw code: http://www.r6rs.org/steering-committee/election/pr.scm
Raw data: 
http://www.r6rs.org/steering-committee/election/preliminary-ballots.scm

david rush
-- 
GPG Public key at http://cyber-rush.org/drr/gpg-public-key.txt

_______________________________________________
r6rs-discuss mailing list
[email protected]
http://lists.r6rs.org/cgi-bin/mailman/listinfo/r6rs-discuss

Reply via email to