Hi Michael,
Michael Meeks wrote (24-11-10 22:01)
On Wed, 2010-11-24 at 17:07 +0100, Cor Nouws wrote:
So - I havn't got to looking at this in detail yet; but I strongly
recommend a 'fair' voting scheme - such as used by GNOME - ie. STV. This
Can you pls tell what STV is?
Single Transferable Vote; cf.
http://en.wikipedia.org/wiki/Counting_Single_Transferable_Votes
Thanks, I see we can study quite some mathematical variations on this
subject ;-)
it is a fair way of electing people, such that:
makes it very difficult for a contributor with 51% of the votes to get
100% of the seats [ which 1st past the post assures ].
this cannot happen.
And what has a fair voting scheme to do with 100% each year or 50%
each year?
I tried to explain that. Basically, the problem is one of quantisation;
let me provide a few examples. The fewer people up for election, the
less fair the election is.
With STV - if we have 100 people for election, and 51% of the electors
choose a Foo-company candidate as first choice (and no-one else) - then
we end up with 51 (or perhaps 52) Foo-company guys.
However - if we have 1 person for election, and 51% of the electors
choose a Foo company candidate. We get a a single Foo-company elected
guy - right ?
So - in the limit, as we move towards smaller and smaller pools of
people we are voting for, the rounding errors get worse - and the
'fairness' (where by fairness, I mean matching the distributed intent of
the voters to who is actually elected) decreases.
This is essentially an artifact of not being able to elect fractional
people ;-) So lets take two lots of 4 vs. one lot of 8 people being
elected. Lets fix the vote both times: one lot 'A' have 66% of the first
choice vote: the other 'B' the remaining 34%
to vote for elected
4 3 A, 1 B
4 3 A, 1 B
8 5 A, 3 B
Soo ... in the two lots of 4 case, we get one fewer non-'A' person. At
this point someone of course will fault my mathmatics in this particular
instance ;-) suffice it to say, that (by considering the limits) this
kind of rounding problem has to exist.
As such, the bigger the number of candidates to elect, the more fair
the representation ultimately.
OK, that is clear, thanks.
However, in your example and view, you start with the idea of
candidates representing company/section/partition A or B. Rather then
random community members.
Then, there is the admin overhead of elections, and the problems of
getting people to vote more regularly.
Both voting each year for 100% of the seats, or each year for 50% of
the seats, result in one election per year. With less counting in the
letter case ;-)
Ok - but if you only elect half each year, then we have to have two
year terms, which seems -incredibly- long in a fast moving software
industry.
It is long in terms of software, indeed. Two years means probably 500
versions of LibO, of which at least 5, 6 stable versions ;-)
But that can only be reached when we have experienced people on board
that do not change every 6 months. Also for documentation, qa etc,
experience is important.
Maybe you can further explain what you think to gain with 100% of the
seats each year?
And perhaps you can help me understand what we loose ? :-) Particularly
when the foundation is in its inception, and growing fast, it seems
crazy to me to have two year terms.
No answer, but OK, I'll give mine fist ;-)
When I initially wrote about this suggestion, I referred to all I
have seen and joined wrt boards, councils etc. where three years is most
common (95+%).
I also made a not about our fast IT-world, and - sorry - did not
refer to my probably state of being old fashioned.
So it is simply for continuity, a bit stability.
If our future practice shows us that it is not workable, dull, slow,
boaring, suffocating and such, I am all for a change.
Looking at all proposals that fly rapidly around, I would prefer a
bit stability first. The next issue then would be, which form of STV or
other voting scheme would be most appropriate. So the voting scheme
should support our desired elections scheme, not the other way round.
Regards,
Cor
--
- giving openoffice.org its foundation :: The Document Foundation -
--
Unsubscribe instructions: E-mail to [email protected]
List archive: http://www.documentfoundation.org/lists/steering-discuss/
*** All posts to this list are publicly archived for eternity ***
