[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: RFD: informal proposal

On Mon, Nov 18, 2002 at 06:08:10PM +1000, Anthony Towns wrote:
> Again, I'm not inferring it, I'm assuming it: that 90:10 people are happy
> to remove non-free, and that 60:40 people would rather remove non-free than
> handwave about it. If it weren't the case -- if, say, only 60:40 people were
> happy to remove non-free -- they should have voted:
> 	60 A B D
> 	30 B D A  ***
> 	10 D B A
> and their opinion "No, don't remove non-free" would have been accurately
> expressed, and B would have won since A didn't achieve its required
> supermajority, and B was the next best compromise. (A eliminated since
> it doesn't beat D, B beats D 90:10, the fact that A beats B 60:40 is
> irrelevant)

Hmm, I think that's where we differ: D means "further discussion", and
I think that someone who definitely doesn't want A would want to settle
that issue, not discuss it more.  Further discussion would lead to
another vote, which leaves no way for B-but-not-A supporters to vote
"definitely not A, now shut up about it".

Are we perhaps differing on the interpretation of "handwave about it,
don't do anything"?  I expect that someone who prefers a "don't do
anything" option over a "do something" option is, in fact, in favour
of not doing anything.  This is precisely a vote for the inertia that
a supermajority requirement is supposed to represent.

Hmm, "further discussion" and "none of the above" are not equivalent
this way, but we represent both of them as "the default option" and
treat them the same way.

Maybe what we really need in mixed-supermajority votes is to make
explicit which options support which other options.  To illustrate
the need for this, consider this vote, with results identical to

   A - rename Technical Committee to Non-Technical Committee
       and change its role accordingly (2:1 required)
   B - leave Technical Committee alone
   D - Further discussion

I think you'll agree that in this case, it's not reasonable to count
a B A D vote as support for the supermajority requirement for A.  (If you
don't agree, then please explain, because then I'm really missing something
in your argument.)

However, the only difference between this vote and your example is in
the semantics of the options.  We can't have an algorithm that decides
this vote differently (given identical results for A, B, and D) unless
it takes the meanings of B and A into account.

One way to fix this could be to separate the "default option" and
"superminority" roles, so that there would be an explicit "Maintain
status quo" option that does not invite further discussion.  Voters
can then explicitly rank it against the supermajority options,
and supermajority options are only scaled when they are compared
to that special option.

Richard Braakman

Reply to: