[EM] Kevin, 19 March, '05, 0330 GMT

[MIKE OSSIPOFF]

Kevin--

I'd said:

>You suggested that there could be a method in which a majority, who have 
>transitive strict preferences among all the candidates, could ensure that 
>some Y won't win, by alternately voting ">" and "=" in their rankings. You 
>said that's a silly way of voting, and that, because a silly way of voting 
>could make Y lose, thereby compllying with WDSC's requirement, that meant, 
>you said, that WDSC's requirement must not be reasonable.

You say:

No, I think it's a silly way to *have* to vote. And if WDSC passes a method
that requires a voter to vote in a silly way, then maybe WDSC could be made
more strict.

I reply:

I don't deny for a minute that WDSC could be made more strict. Then it would 
be a different criterion, a more strict one, a stronger one. SDSC is an 
example of a criterion that's stronger than WDSC.  SFC, too,  is more 
difficult to comply with than WDSC is.

But, as I already told you, it's good to have a gradation of criteria with 
different demandingness. In that way, we have a way to measure for several 
different desiderata, some of which may be easier to get than others. Yes, 
I'd like methods that meet more than WDSC. Approval and CR additionally 
meets FBC. PC additionally meets SFC. No method that I propose meets only 
WDSC. But if there's such a method, it would still be an improvement over a 
method that doesn't even meet WDSC. I considser WDSC to be the bare minimum 
for a tolerably acceptable method. But all the methods that I propose meet 
more than WDSC.

Well, though I call WDSC the bare minimum requirement for an acceptable 
method, I guess I'd have to agree that, by itself, WDSC wouldn't really be 
enough, when we consider that there are so many methods that do better than 
just meeting WDSC.

But WDSC, by itself, is still useful for showing that a method is completely 
inadequate. A method such as Plurality, IRV, or Margins.

Any number of other criteria could be written that are stronger than WDSC, 
including one that specifically requires that that majority mentioned by 
WDSC additionally not have to vote in a way that is silly. You'd have to 
define silly, or course. If you want a criterion like that, then I encourage 
you to write one.

I must admit that my requirements are a bit more modest than yours.

I want voters to not need to reverse their preferences in order to make some 
"greater-evil" lose.

Under the conditions described in WDSC's premise conditions, a majority can 
be free of that need, with a complying method.

Not only do I want them to have that need, but many others don't like it 
when voters are strategically forced to reverse their preferences.

Now, even though you haven't named a method that requires those voters to 
alternately vote ">" and "=", it bothers you a lot that such a method could 
pass a criterion that requires them to not need to reverse a preference.

What you need, then, is something quite different from WDSC. You need to 
define the No Silliness Criterion. Go for it. In the meantime, though, I'll 
settle for methods that don't require those voters to reverse a preference.

When you write the No Sillilness Criterion, will you specifically say that 
the members of that majority shouldn't need to alternate ">" and "=" in 
their rankings, or will you write a more general definition of silly voting? 
  For instance,  maybe you'll  define silly voting as voting that has ">" 
and "=" in some regular, recognizable, or easily-describable pattern? It's 
up to you!

Well, you're breaking new ground, and I encourage you
to pursue what's important to you.

If I could make a suggestion for your No Sillliness Criterion, you might 
consider SDSC. It requires that that majority additionally not need to vote 
equal two candidates that they don't prefer equally. As I define "vote 
equally", it applies only to candidates whom that voter votes over someone.

Is SDSC  what you're searching for, as a substitute for WDSC? If so, then 
you could consider SDSC to stand for "Silliness Defensive Strategy 
Criterion" :-) , though, as I name the criterion, SDSC stands for Strong 
Defensive Strategy Criterion.

I'd said:

>Now, maybe you'd like to hypothesize a method with which, under certain 
>conditions, the _only_ way that that majority can keep some Y from winning 
>would be for them to alternately vote "=" and ">", or to use ">" and "=" to 
>spell out "WDSC" in Morse code. Silly? Sure. Would that be a silly method? 
>Of course. Would that mean that the guarantee that those voters don't have 
>to reverse a preference in order to make someone lose is not a meaningful 
>or reasonable guarantee. No.

You say:

I don't agree with you.

I reply:

Ok, then you must be  saying that guaranteeing that those voters won't need 
to reverse a preference isn't a meaningful or reasonable guarantee.

That's quite a claim. Saying that you'd prefer to require that some 
additional kind of voting also not be needed isn't enough to justify your 
claim that a guarantee that those voters needn't reverse a preference isn't 
a meaningful guarantee. For that to not be a meaninful guarantee, you'd have 
to say that it doesn't mean anything to you whether or not voters need to 
reverse a preference. Well, maybe it doesn't. I don't claim to know what's 
important or meaningful to you.

But you have a right to want more, and to write a criterion that requires 
more, and to tell us which methods meet it and which don't. Or would you 
rather just complain because I haven't written your criterion that you want?

You say it isn't a reasonable guarantee. So, then, is it an unreasonable 
guarantee? :-)

Mike Ossipoff

_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today - it's FREE! 
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/