Re: Proposed constitution fix for n+1-majorities
* Steve Langasek (vorlon@debian.org) [090109 21:58]:
> On Wed, Jan 07, 2009 at 12:39:54AM +0100, Andreas Barth wrote:
>
> > this is a proposal to fix the n+1-bug in the constitution:
>
> > In A.6.3.2, this sentence is changed:
> > An option A defeats the default option D by a majority ratio N, if V(A,D)
> > is strictly greater than N * V(D,A).
> > to:
> > An option A defeats the default option D by a majority ratio 1, if V(A,D)
> > is strictly greater than V(D,A), or (for N > 1) if V(A,D) is at least N *
> > V(D,A).
>
> > Background: In case of majority requirements N > 1, we require that more
> > than N * people vote in favour of the option than against (instead of "at
> > least N *"). This isn't a real issue for normal GRs, but it is an issue for
> > the tech ctte.
>
> While it may not be a "real issue" for GRs, the proposed change nevertheless
> does change the supermajority requirements for GRs. I don't object to this
> myself, but is it going to be generally acceptable to developers to change
> the GR procedure because of a bug that's specific to the TC?
Well, the bug also happens with normal GRs, only that (if we look at
the large number of DDs voting) it normally doesn't have any effect.
(It could also happen if e.g. 150 DDs vote against a constitution
change, and 450 DDs vote in favour - but usually numbers are a bit
more off, so 451:150 is already enough, which is 3.007:1, whereas with
the tech ctte requireing 4:1 instead of 3:1 is a by far large change.)
> The constitution does already specify that the TC chair has a casting
> vote, although it's not altogether clear what this means for
> condorcet+supermajority. I think a rational interpretation would be that if
> an option defeats the default option by exactly N:1, the chair's vote
> determines whether that option is considered to pass the majority
> requirement.
I'm not sure whether I'm so happy with that. Hm. I think I have to
think about that.
Cheers,
Andi
Reply to: