mixing approval and condorcet

[Raul Miller <moth@debian.org>]

We need to use something with properties somewhat like approval voting
if we're going to implement quorum and supermajority.

However, approval doesn't allow us to rank options we favor, while
condorcet does.  Thus, it makes some sense to use a system which combines
approval and condorcet.

If we are consistent about this, options with 1:1 majority will also
be dropped from consideration if they do not beat the default option.
We would use CpSSD to pick among any options which meet quorum
requirements and beat the default option.  This isn't really approval
voting, and it isn't really a condorcet method, but I think it offers
the best of both for us.

I have a vague memory of someone objecting to this, but I think the
objection was the "stability argument", and after spending quite a bit
of time on this issue, I think the stability argument is a huge waste
of time.

Does anyone have any other reasons/criteria for opposing this kind
of system?

[Meanwhile, I'm going to be coding up more test sampling conformance to
various criteria -- so far I've only implemented a test monotonicity.]

Thanks,

-- 
Raul