Bug#636783: TC supermajority bug
We discussed this on -project. I have been convinced that my previous
approach was wrong. I now think that we should do as follows:
* Move the "required majority ratio" test from A.6(3) to
the very end, after A.6(8).
* Change the test > to >=.
* The consequence of not meeting the majority ratio would be
specified at each place where a majority ratio is mentioned.
* Constitutional GRs and foundation document amendments: If the
required majority ratio is not met, the consequence would be that
the whole GR outcome is FD.
* Overriding the TC, and for TC overruling a Developer: If the
required majority ratio is not met, the consequence would be that
the whole resolution is to be regarded as an advisory statement of
opinion, and is not binding.
* Explicitly state in A.6(8) that if there is no elector with a
casting vote, the default option wins.
The results would be:
* In a split TC vote with no supermajority requirement, where FD is
in the tied set, the elector with the casting vote would be able to
choose any of the options, or FD.
* Supermajorities are a bare minimum.
* Evenly split 1:1 GRs (equal number of votes Y and N) still fail.
* A 3:1 GR which obtains votes 300:100 passes where previously it
would fail.
* We can no longer have a GR which produces in a pathological result
where the winning option is eliminated by the supermajority rule
leaving a minority view as the prevailing outcome.
* In a split TC vote with supermajority requirement, where any of the
options beat FD by the supermajority or more: barring pathological
situations, there will be one or more options which all beat FD but
which are tied or form a Schwartz cycle. Neither FD, nor any
non-3:1 options, are likely to be in the Schwartz set. The most
likely scenario is a tie about which version of an overruling
motion to use. The elector with the casting vote would be able to
choose between any of the options in the Schwarz set (ie, probably,
any of the tied overruling options) but not be able to choose FD.
I haven't yet drafted this up.
Ian.
Reply to: