[EM] SARC definition improvement

[MIKE OSSIPOFF]

>Your comment ("You'd be surprised how many people value majority rule.")
>makes sense only when you presume that everybody who values majority rule
>also values SFC, GSFC, WDSC and SDSC. My comment ("You would be surprised
>how many people value majority rule without valueing the Condorcet
>criterion.") says that I don't agree that everybody who values majority
>rule also values SFC, GSFC, WDSC and SDSC.

Now you're saying that standards on which criteria are based
mean nothing because maybe someone will agree with the standard
and not like the criterion. Are you saying that you don't
agree that SFC, GSFC, WDSC & SDSC measure for the majority rule
standard. If so, what criteria would you suggest instead?
You haven't mentioned any on the list, but why don't you write
some that measure for majority rule better than those 4.

You can speculate that people who believe in majority rule might
not want a majority to be able to get its way without drastically
insincere strategy, but what is the value of such speculation?
It would be more productive to say why you don't believe that those
4 criteria measure compliance with the majority rule standard,
and to state what other criteria you believe would do a better
job of that.

It's too easy to say "Maybe people won't like that" instead of
suggesting criteria that you think would be better.

>
>******

> > > To my opinion, this demonstrates that SARC doesn't do what it
> > > was designed for.
> >
> > It was designed to find out with which methods undominated voting
> > can defeat favorites or elect last choices. It does what it's
> > designed for. The fact that you can devise some way of voting
> > in PC or Plurality that won't do that doesn't matter unless you
> > can promise that everyone will vote as you suggest.  Because,
> > otherwise, you haven't shown that the violation won't happen.
>
>But why was SARC designed to find out with which methods undominated
>voting can defeat favorites or elect last choices? If the aim is to
>make the voters be not deterred from showing up, then I have to answer
>that my example above demonstrates that SARC doesn't do what it was
>designed to do.

SARC wasn't designed to make voters not be deterred from showing
up. At no time did I say that it was. In fact, wasn't there
agreement on this list that no one would stay home to avoid a
possible adverse result? Right now, for instance, we're using
Plurality here, and it violates SARC. Even if there were enough
honest progressive voters so that Nader could win if some of the
Gore-voting Nader-preferring voters stayed home, I don't think any
of those voters will say "If I show up & vote for Gore, I might
make Nader lose. I'd better stay home." Either he goes & votes for
Gore, or he goes & votes for Nader. If he's interested in the
outcome and wants to try to affect it in his interest.

Your worry is groundless that voters will fear to show up because
they might cause themselves an adverse result.

The purpose of SARC is as I have stated. To find out with which
methods undominated voting can defeat a favorite or elect a
last choice. We have more traditional adverse results criteria, which,
when worded about a voter worsening his outcome, stipulate that he
votes sincerely. I don't think that's a realistic assumption, and
so I changed the assumption of sincere voting to a more realistic
assumption of undominated voting.

>If the aim of SARC is to guarantee that a voter isn't
>punished for showing up and using a certain kind of strategy, then I
>have to answer that if you show up and vote strategically then it is
>your own fault when you are punished.

You're repeating the same argument without commenting on my
previous answer to it. As I said, one way you can defeat your
favorite, for instance, is by compromise strategy. But, as I said,
more than once, SARC compliance doesn't mean that the method
protects you from the possible consequences of voting someone over
your favorite. That complying method, rather, protects you from
having a strategic need to vote someone over your favorite. So
that, with a complying method, voting someone over your favorite
is a dominated strategy, and you won't do it.

As I keep telling you, you say that I want to protect insincere
voters, and I say that I only want to protect them from the
need to vote insincerely. This is what we mean when we refer to
people repeating the same argument without answering the
reply to that argument.

Aside from all that, SARC doesn't mention insincere voting.
With some methods you could vote sincerely and elect your last
choice. That can easily happen with IRV, for instance. You can
write an IRV SARC badexample in which everyone votes sincerely.
I'll post one upon request.

I don't understand what you think you prove when you find some
case where a method doesn't violate SARC. The idea is: It would
be nice if those failures couldn't happen. As I said, you can't
guarantee to me that people will vote as you suggest, in a way
that can't cause a violation. It's of interest with what methods
that can happen, and with what methods it can't happen.

I don't think this is complicated.  It's nice if people can't
defeat their favorite or elect their last choice, by showing up.
What part of that do you disagree with?

Would I like to extend that to all adverse results, instead of
just those involving 1st & last choice? Sure, but then the
criterion would be unmeetable.

>
>******
>
>You wrote (12 Sep 2000):
> > You're making the error of feeling as if a method meets a criterion
> > because you can find some case where it doesn't fail the criterion.
>
>In so far as I wrote in my last mail that Approval Voting meets SARC
>and PC violates SARC, I don't understand your comment.

You know that PC violates SARC, but you feel as if PC
as-good-as-meets SARC, because you've found a case where
PC doesn't violate SARC.

I invite you to propose a criterion that differs from SARC by
saying that a method passes if it's possible to find a way to
vote that won't defeat favorite or elect last choice. But that's
your criterion, not mine. I don't know if any method would fail
that criterion, and, if not, it wouldn't be useful.

>
>******

>I suggest beat-path GMC as an alternative to SFC, GSFC, WDSC and
>SDSC. (In so far as you consider beat-path GMC to be too strict,
>you can also use BC.)

Now that's better. Any rank method that meets BC meets SFC, GSFC,
WDSC & SDSC. The reason to use those 4 instead of just BC is
that those 4 majority defensive strategy criteria have obvious
purpose & motivation: A majority being able to defeat someone
without some specified degree of insincere voting. For people
who don't want sincere voting to be punished (you, for instance
don't), the purpose & desirability of those criteria is obvious.
We don't name BC at the website because its purpose isn't obvious.

Being able to defeat some greater evil without an unnecessary
degree of insincere voting obviously reduces the lesser-of-2-evils
problem, because that's what that problem is about. It also
accomplishes that goal that you expressed: protecting sincere voters.

>
>******
>
>You wrote (12 Sep 2000):
> > In a reply to Markus I spoke of minimizing need for insincerity.
> > When I say that I want criteria that measure how well a method
> > minimizes need for insincerity, I mean that I'd like to
> > minimize the degree of insincerity needed, and it would be good
> > for that protection to extend to as many voters in as many
> > situations as possible.
>
>I consider the way you measure the need for insincerity to be
>unrealistic. You ask about the needed insincerity when a majority
>of the voters strictly prefers candidate A to candidate B and
>this majority wants to prevent candidate B from winning. I consider
>this way to measure the need for insincerity to be unrealistic
>because when someone votes strategically then he does so to
>maximize his own utility expectations and not because he wants
>some "rightful" candidate to win.

The criteria make no mention of any particular candidate winning.

If you want to write a criterion about utility expectation
maximization, you have my encouragement. But my criteria aren't
the only ones that aren't about utility expectation maximization.

Do you want to throw out every criterion that isn't about
utility expectation maximization?

Maybe you want all strategy criteria to be about that. Again,
you're criticising without making a constructive situation.
If there's a utility expectation maximization criterion that would
be a good replacement for one of our defensive strategy criteria,
then why won't you write it here?

You say maybe a better one could be written, but instead of
writing it, you advocate dropping the existing defensive strategy
criteria, which would result in there being none.

Suggesting a criterion about utility expectation maximization
is one thing, but saying that other strategy criteria are
unrealistic is a claim that needs justification. When, in
accordance with a strategy to maximize utility expectation,
someone votes for Gore, he's trying to maximize his utility
expectation, but he's also trying to make Bush not win.
You haven't demonstrated that those 2 kinds of goals are
incompatible.

Realistic? Ask anyone why they vote for the Democrat instead of
for the candidate whom they like more, and they'll tell you it's
to keep the Republican from winning. I've heard that more times
than I'd care to estimate. Not once has anyone told me it's to
maximize their utility expectation. Of course someone might feel
that keeping the Republican from winning is the best way they
can use their vote, and their best utility expectation strategy.

You're really grabbing at straws now, Markus. Speaking of what
it take to keep someone from winning would be unrealistic if
people didn't vote as the do in order to keep someone from winning.
But millions of people say they do, and it's an acknowledged problem.
Your claim about realism seems indefensible. Can you defend it?

>
>******

> > Why is Markus posting this outdated message from Bruce?
>
>I don't agree with you that Bruce message is "outdated." Bruce wrote
>in great detail where he had problems understanding the wordings of
>your "lesser-of-two-evils" criteria. It is true that in a reply to
>Bruce you explained in great detail the wordings of your "lesser-of-
>two-evils" criteria. But nevertheless you still use the same
>wordings. Therefore someone who reads today's wordings of your
>criteria will still have the same problems like Bruce.

Only if they imagine something that I didn't say. I didn't say
that all 3 majorities should be able to simulataneously use their
SDSC power to defeat all 3 candidates. I said that each one of
those majorities should have a way of voting that will defeat
a certain candidate, under those conditions.

By the way, it isn't that I explained otherwise incomprehensible
wordings to Bruce. I told him why his interpretation was contrary
to the literal meaning of what I'd said. I explained to him that
he'd invented something that I hadn't said. That isn't the same
as admitting that the wording was ambiguous and supplementing it
for Bruce.

You can help by trying to find actual problems in the criteria.
But saying that you or someone misunderstood it, without being
able to point to ambiguity in the wording, that doesn't prove
anything, because, frankly, you're motivated to find a problem.
That's what makes you so helpful in finding actual problems.
We want to know of any problems that you can find.

But if the worst that you can say is that you misunderstood it,
without showing ambiguity; or that a criterion of a different
type could be written (but without writing one), what that shows
is that you can't find a genuine problem with the criteria.

Mike Ossipoff

_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.

Share information about yourself, create your own public profile at 
http://profiles.msn.com.