[PLUG] Excellent talk on preferential voting at PLUG North Tue 12/11...

[Fred Stluka via plug <plug@lists.phillylinux.org>]

PLUG folks,

John Karr gave an excellent talk on "preferential voting" at PLUG
North Tue 12/11.

I had heard of "Single Transferable Vote" aka "Ranked Choice Voting",
and had concluded that it was a pretty good idea.  I had written this
short summary of it:
- http://bristle.com/Vote/#single_transferable_vote

After hearing John's talk and chatting with a few very knowledgeable
attendees like Rich Freeman and John Ashmead (and others whose
names I didn't catch), I now know a lot more detail.

As I now understand it...

There are 2 basic forms of ballots for preferential voting:
- "Rank" (aka "Ordinal") - Voter picks 1st, 2nd, 3rd choice, etc.
- "Range" (aka "Cardinal" - Voter assigns preference score from
   a specified range like 1-5.  May give any score to each
   candidate, including multiple with the same score.  Not giving
   a score is the same as giving score 0.

There are several different ways to count the votes.  Mostly, they
say that a "Majority Winner" wins, if there is one.  For example, a
candidate who gets more than 50% of the 1st choice votes when a
a Rank ballot is used.  They differ mostly in how to proceed when
there is no such clear winner.

The simplest (and the only one I'd ever heard of before) is "Instant
Runoff Voting" (IRV).  If no candidate has more than 50% of the
1st choice votes, you drop the candidate with the fewest 1st
choice votes, convert the 2nd choice on those ballots to 1st
choice, and re-count.  If still no winner, drop the candidate with
the next fewest 1st choice votes, etc., until you have a winner.

This generally works MUCH better than our current system of
voting ("Plurality" winner) because it solves the "Spoiler" problem
(aka "Favorite Betrayal").  That is, voters can safely vote for a
long-shot 3rd-party candidate without risk of "throwing away"
their vote.  They're not missing out on the chance to vote for
their preference among the 2 major parties, to prevent the
strongly undesired other major party candidate from winning.

However, in certain cases, especially when a long-shot candidate
is not such a long-shot and actually has a real chance to beat
one of the major party candidates, it can still go wrong.  Rich
Freeman explained this very well before the talk started.  Here's
a short video that explains it like he did:
- "Favorite Betrayal in Plurality and Instant Runoff Voting"
    https://www.youtube.com/watch?v=JtKAScORevQ

Since IRV is not perfect, there are various other techniques that
can be used to count the ballots, discard weaker candidates, and
choose a single winner.  Unfortunately, none of them is perfect.
There are pros/cons to each of them.  But they pretty much ALL
work better than our current "Plurality" system.

John Karr discussed various criteria for evaluating such techniques
(Complexity, Consistency, Later Harm, etc.) and various techniques
(Condorcet, Smith Set, Borda, etc.).  He also showed some sample
data sets and how the different techniques chose different winners.
He used the open source Vote::count software that he's written in
Perl in his demos.

Be sure to catch his talk if he gives it again at another PLUG
location.  Was the talk video taped?  If so, someone please post
the video.  Also, are the slides available on-line?

Thanks for an excellent talk, John!
--Fred
------------------------------------------------------------------------
Fred Stluka -- Bristle Software, Inc. -- http://bristle.com
#DontBeATrump -- Make America Honorable Again!
------------------------------------------------------------------------

___________________________________________________________________________
Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug