[Noisebridge-discuss] election software

[Andy Isaacson]

[tl;dr is at the end]

At tonight's meeting, Leif raised an interesting point that the
Condorcet voting method is not precisely in line with his conception of
what Noisebridge is about.  (FWIW, I agree with his argument, mostly, I
think.)  He suggested that we instead use "approval voting", which has
largely the same user experience but, he says, is more likely to choose
the candidate who is acceptable to the largest subset, rather than the
candidate who is preferred by the largest subset.

Several people thought this sounded like a good idea.

We decided, though, that in the interest of time we would go ahead with
a Condorcet process.  Shannon reminded us that last year, we had to use
an ad-hoc process, created during ballot processing, when we realized
that the Condorcet implementation we were using only gave us the single
winner rather than the top 5 winners.

The process we used, IIRC, was:
1. run the process to find the first winner, Abe.
2. Remove Abe's votes from the ballots.
3. run the process to find the second winner, Bob.
4. Remove Bob's votes from the ballots.
et cetera.

We documented the process at
https://www.noisebridge.net/wiki/2009_Director_Elections

Shannon proposed that we should add a feature to our implementation to
obviate this ad-hoc process.

I (and Leif?  I forget who else volunteered) are going to improve Ping's
Python Condorcet implementation to handle our situation better.

We'll publish a signed hash of the code no later than December 31.

Please clone from git://github.com/radii/condorcet if you'd like to
contribute.

Thanks,
-andy