[AFS3-std] Re: First order of Business - AFS Standardization Process

[Russ Allbery]

Jeffrey Hutzelman <jhutz@cmu.edu> writes:
> Andrew Deason <adeason@sinenomine.net> wrote:

>> Another comment that was not brought up on the list anywhere yet, but
>> that I have mentioned to a couple of others, is about the voting system
>> used to elect chairs. A Condorcet method would be nice instead of the
>> simple single-vote mechanism used now, in order to avoid some of the
>> common pitfalls of such simple elections.

> Note, however, that because we have two chairs, there is an edge case in
> which a single-winner system is not sufficient.  So, if we choose such a
> system, we will need to make modifications to eliminate those cases.

Condorcet gives you a ranked set of choices, not just a winner, so it's
particularly good there (although I don't know what happens if you have to
apply the cycle-breaking strategy; that's never happened in any Debian
vote).

> In general, any system we choose needs to be easy to understand well
> enough to particpate, to operate elections, and potentially to verify
> results. Bear in mind that most of us are not game theorists :-)

Condorcet is pretty trivial from the voter perspective: it's just ranked
preference voting.  It's more complicated than "pick one," but it's not
horribly more complicated than "pick two."  All the work is on the vote
counting side.

-- 
Russ Allbery (rra@stanford.edu)             <http://www.eyrie.org/~eagle/>