[EM] Score DSV JQ

[Kevin Venzke]

Hello,

--- En date de : Sam 29.8.09, Jameson Quinn <jameson.quinn at gmail.com> a écrit :
>> Issues:
>> 
>> 1. If you don't use Warren's methodology and
>> assumptions, it's not clear
>> that your results will be convincing to a Range crowd. (And
>> other crowds
>> don't care as much.)

> The part about Range partisans being wedded to Warren's
> assumptions I understand, though I don't necessarily
> agree. The part about other partisans not caring about
> utility seems stranger to me. Why not?

Mostly because other crowds don't consider utility very easy to measure.
It could still be interesting.

> Anyway, I'm proposing having each virtual voting group
> evaluate whether strategy will help them, given different
> levels of true information. I think this is feasible
> computationally, and I don't see how anybody in any camp
> could argue that finding utility in this case is not
> relevant.
> 
>> 2. When Range voters vote approval-style and Condorcet
>> voters use
>> reasonably sane strategies, Range/Approval is known to be
>> worse, as the
>> number of viable candidates increases. So it won't be
>> that novel to show
>> that your method is better than Range
>> here.
>> Where are you getting this?

Warren practically said this himself in his "Range Voting" paper from
years ago. But he doesn't find this very important because he doesn't
believe Condorcet voters will use anything like sane strategies.

The fact that Approval gets worse as the number of viable candidates
increases I guess I take mostly from my own simulations. I may have
overstated how "known" it is.

>> 3. Given the nature of the differences between Approval and
>> Condorcet,
>> it seems that Score DSV's consideration of ratings is
>> more likely to
>> hurt it than help it here.
> With honest votes, or considering strategy? I can't see
> why you'd say this. Score DSV is more like Range than
> your average condorcet system.

Considering strategy.

>> Well, here are some comments going over the page
>> quickly.
>> "If there's a Condorcet winner, all voters'
>> ideal strategy will be to
>> vote approval-style, and the Condorcet winner will win,
>> thus this method
>> satisfies the Condorcet criterion."
>> 
>> I wrote out a whole long thing here but eventually realized
>> that you
>> aren't ruling out non-Smith candidates from winning.
>> And that is why you
>> are talking about strategy above.
>> 
>> Fortunately or unfortunately depending on your perspective,
>> you have to
>> evaluate Condorcet compliance based on cast votes. If a
>> voted CW doesn't
>> necessarily win, then Score DSV isn't a Condorcet
>> method.
> 
> Ouch. That passage is obviously unclear. I meant
> "strategy" in the sense of "declared
> strategy". I was not considering any strategy at all
> from the actual voters on the ballots they would input to
> Score DSV, but virtual "declared" strategy on the
> output (imaginary) renormalized ballots, which are intended
> to be equivalent to (the probabilistic average of) their
> strategic Range ballots if their input ballots are honest
> and if they knew the true Smith set but nothing else. In
> other words: if there is a condorcet winner, the correct
> Range strategy for those who know that winner (and nothing
> else) is to vote approval-style for that person and all
> better candidates, thus Score DSV chooses the CW. It is a
> Condorcet method, even though it does not satisfy the Smith
> criterion (if there is no CW, it could potentially elect the
> condorcet loser, if that candidate had a high renormalized
> utility).

I don't understand. You need to be able to say "if there is a CW,
then the CW is elected." What is this talk about strategy? Either
the CW necessarily wins in this method or he doesn't. It can't depend
on what the voters choose to do.

>> I don't remember (and won't examine presently) the
>> precise wording of
>> SFC (strategy-free criterion), but Score DSV doesn't
>> seem to satisfy
>> the votes-only shortcut interpretation, because it can
>> elect B with
>> these rankings:
>> 
>> 49 b (a and c rated zero)
>> 24 a>b
>> 27 c>a
>> 
>> The criticism is that the A>B voters can give away
>> victory to B, when
>> 
>> assuming no order reversal, A might be the "sincere
>> CW" but B definitely
>> is not.
>> This case has a CW, so Score DSV would choose that winner.

> There is no condorcet cycle. You need at least 4 of the b
> voters to vote b>c for your example to work. Then your
> example is no longer covered by the SFC, which states:
> "If a Condorcet candidate exists, and if a majority
> prefers this
> candidate to another candidate, then the other candidate
> should not win
> if that majority votes sincerely and no other voter
> falsifies
> any preferences.
> In a ranked method, it is nearly equivalent to say:
> If more than half of the voters rank x above
> y, and there is no candidate z whom more than
> half of the voters rank above x, then y must
> not be elected." 

I don't follow. The voted cycle is A>B>C>A.

>> It doesn't satisfy the votes-only interpretation of
>> SDSC, because it
>> can elect B with these rankings:
>> 
>> 49 b
>> 24 a
>> 27 c>a
>
> Again, no, only if you change 28 total b and a voters to
> b>c and a>b, respectively, which puts you out of the
> purview of SDSC.

Sorry, I don't follow again. What do you say is accomplished by changing
the B and A voters this way?

Kevin Venzke