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

Re: Constitutional amendment: Condorcet/Clone Proof SSD vote tallying

> > Correct me if I'm wrong, but: what Manoj's May 15 proposal
> > implements logically equivalent to your suggestion?

On Sun, May 25, 2003 at 02:54:08AM +0200, Markus Schulze wrote:
> As far as I have understood Manoj's May 15 proposal correctly,
> an option is dropped unless it _directly_ defeats the default
> option with the required quorum.

Actually, no.

For example:

Quorum of 10, ballot A, default (D), votes:

31 A D
28 D A

Here, A does not defeat D by 10, but still satisfies the quorum

In essence, the quorum requirement is very much like the practice of
requiring "seconds" for an option.  The major difference between this
concept of option quorum and seconds is quantative (seconding happens
during the discussion period, and if we require too many seconds that
would stall discussion), and we use the ballot collection mechanism to
determine which options (thus, for example, in a leader election the
mechanism is anonymous).

On the other hand, options not meeting quorum can't be in the beat paths
which defeat other options -- so, in that sense, those beat paths are

> I suggest that it should be
> sufficient that this option _transitively_ defeats the default
> option with the required quorum.

As stated, this changes the quorum requirement from a concept of "minimum
level of approval required to conduct business" to a concept of "required
margin of victory required against the default option".

In earlier discussions, it was felt that the "margin of victory" concept
was too heavily biased in the direction of the status quo.

That said, I'll note that your explanation, further down, actually
indicates you had a different idea in mind.

> In Situation 1 in my last mail, the quorum is 38. According
> to Manoj's May 15 proposal, candidate D is disqualified since
> only 24 voters strictly prefer candidate D to candidate C.
> According to my proposal, candidate D is not disqualified
> since 38 voters strictly prefer candidate E to candidate C
> and 42 voters strictly prefer candidate D to candidate E.

Oh... I think I see.  You're not looking at defeats, but minimum votes
in favor from any element on the beat path.

The problem I have with this concept is that it doesn't seem to be related
to how we've defined the default option in the first place -- instead of
"rank something above the default option if you think it's a good idea",
it's ... well, I'm not sure what it is.

> Manoj's May 15 proposal would choose candidate E. In the next
> elections, when candidate E is the default option, Manoj's
> May 15 proposal would choose candidate D. My proposal would
> choose candidate D immediately. In my opinion, the advantage
> of my proposal is that it doesn't lead to unnecessarily
> frequent changes of the status quo.

I think I disagree.

Ok, it's true that in your example, E was ranked second last (with C,
the default, ranked last).  But, given that not many voters elected to
participate, and given that those who did vote couldn't muster enough
agreement that any of the other options were better than doing nothing,
I think this is a worthwhile outcome.

More fundamentally, however: you've assumed that while only 24 people
thought D was a good idea in the first election, that 42 would think it's
a good idea in the second election.  In other words, your starting point
is thinking of the default option as an outcome, rather than a refusal
to choose an option.  At the same time, you've assumed that the large
majority of voters who didn't participate in the first election would
continue to ignore the second.

The way I see it, either these options just aren't all that important
(in which case there's no reason for 18 people to suddenly decide to
change their minds about whether or not D is a good idea), or they are
important (in which case there should be many more people participating
in that second election).

In neither circumstance does the extra "decisiveness" gained by your
approach yield a positive result:

For the case that these options aren't that important, it's harder to
explain to people what the default option means.  [It no longer means
postponing agreeing on some decisions, except for cases where people
can come to some sort of agreement on the overall ranking of options.]

For the case where these options are important, we're achieving a decision
before people have realized that they care.


Reply to: