Open Stack

Mark Washenberger wrote:
> John Griffith wrote:
> 
>> I'd agree for the most part, I only wonder if there should be a
>> limit regarding company affiliation?  In other words it doesn't seem
>> like it would be overly useful to have the majority of members all
>> representing the same company.  That's the only concern that I would
>> have.

Members of the TC should never "represent a company". If they are seen
as driving the interests of a company while in their TC capacity they
should probably not be elected again. Ideally I would prefer to ignore
affiliation, because it introduces complexity when you hit the limits or
when people change affiliation (which seems to happen to us more often
than others). We used to have 30% of TC members from Nebula in the
previous committee and I didn't really felt like that was an issue.

But if that's a concern I guess we could introduce some kind of upper
limit (less than half ?) and the procedural complexity around how to
handle that limit if we happen to hit it.

> In theory, the condorcet proportional voting [1] approach should help to
> mitigate that concern. But its not something folks are familiar with and
> its not obvious how things would work, so there are some real reasons to
> be hesitant about adopting proportional votes.
> 
> Thierry, do you think we could possibly work with the maintainers of the
> voting/poll site to run a normal, non-proportional vote (if we chose a
> flat 11 election), but to save the data and let us view what the results
> would be if we did a proportional vote?
> 
> 1. http://www.cs.cornell.edu/w8/~andru/civs/proportional.html

Yes, you summarized pretty well the issue with CIVS proportional voting
option... On paper it looks like it would deliver the intended results,
but the fact that this algorithm is experimental and only used on CIVS
prevents us from making sure that it would still elect the "right"
people: we want it to be a safeguard in extreme cases, not change the
natural condorcet results for the general case.

I'll look into how we could experiment with it.

-- 
Thierry Carrez