[EM] Falsifying rankings. Woodall's criteria. Your method.

[MIKE OSSIPOFF]

Chris--

I'd said:

[in regards to "completing truncated ballots]

"Should"? Why? Because Woodall says so?

Woodall, or anyone else, of course has the right to make whatever rules they
want for a method that they propose. But any claim that truncated rankings
should be "completed" needs justification. Just asserting a claim like that
won't do."

You replied:

"Symetric Completion" is the title of a criterion/standard that he has 
worded more like a standard.
The "should" is a value that is epressed by the standard, not Woodall 
himself.

I reply:

What, so a method meets the "Symetric Completion Criterion" if it 
symetrically completes truncated ballots?

But any twit can come up with all sorts of criteria, and it doesn't mean 
anything unless it measures for a widely-accepted standard, or at least one 
that the proponent can convince people of the importance of.

I hereby define the Random Modification Criterion. A method meets the Random 
Modification Criterion if it randomly scrambles each ranking before it 
counts them.

You continued:

(Woodall is not big on "justifying" standards/criteria

I reply:

Now why doesn't that surprise me? :-)

You continue:

, but rather in showing which combinations
are possible and which are not.)

I reply:

Why should anyone care which unjustified criteria are mutually compatible or 
incompatible?

You continued:

Having said that, I like the criterion. Why should a truncated ballot be 
treated differently from its
symetric completion?

I reply:

For the same reason why an aardvark should be treated differently from a 
giraffe? Because they're not the same?

You continue:

To claim that (if there are 3 candidates)that "1A truncate" isn't equivalent 
to
"1A 2B 2C" is absurd.

I reply:

I have never claimed that "1A truncate" isn't equivalent to "1A 2B 2C". 
That's because I have no idea what "IA truncate" and "1A 2B 2C" are supposed 
to mean.

Absurd? Aren't we the assertive one. Why is it that the people who least 
justify their statements are the ones who state them the most assertively?

You continued:

What other reasonable interpretation is there? One voter likes filling in 
boxes
more than the other.

I reply:

Did you determine that by ESP? If a ranking doesn't give support to a 
particular candidate, could it be that the voter didn't want to give support 
to that candidate.  Not just in wv, but in several new methods recently 
proposed here, defensive truncation makes it impossible for a 
majority-defeated candidate to win by offensive order-reversal.

I'd said:

"If truncated rankings were falsified in that way, changed into somethiing
that the voter didn't vote, many would rightly object that
ballot-modification isn't democratic."

"Falsified" is absurd.

I reply:

Because you say so, right?

You modify a ranking so that it's different from what the person voted. 
Everyone knows what it means when we speak of altering or falsifying a 
document. Your "symetrical completion" adds candidates that the voter didn't 
put there. That falsifies that ranking by the accepted meaning of the word.

"Ballot-modification" is not exactly the right phrase.

I reply:

Is that what the ultimate arbiter of rightness says? You've added candidates 
to the ranking that weren't there when the voter gave you the ballot, and 
you say you haven't falsified or modified the ballot.

You continued:

FPP (aka Plurality method)
and IRV both meet Symetric Completion.

I reply:

I'll have to take your word for that, at least until you define the Symetric 
Completion Criterion.

A guess: Does a method meet that criterion iff it gives an unchanged result 
after a truncated ballot is "symetrically completed"?

Is that one of those criteria that Woodall doesn't justify? :-)

I'd said:

"And, with the best methods, rankings falsified in that way would hamper
those methods' ability to deter offensive order-reversal by defensive
truncation."

THIS is the statement that "needs justification".

I reply:

Sorry, but I'm not going to justify it to you. It's common knowledge on this 
list, and has been for years. I'd re-demonstrate it for you if your attitude 
justified the time-requirement.

You continued:

Condorcet completed by symetrically completed
reversed-rankings IRV Elimination, meets Symetric Completion while WV does 
not.

I reply:

Is that the method that you proposed here a few days ago?

It meets Symetric Completion, does it? I must have missed it when you told 
us why it's important to meet that criterion, and so would you repeat that?

Maybe the reason why it meets Symetric Completion is because it 
"symetricallly completes" rankings

About that method that you named above, and defined a few days ago: You 
define an elaborate method, show one or two examples, and either you expect 
us to determine its properties for you,
or you seem to believe that showing one or two examples shows us how good 
the method is. Or you just dump your elaborate method on us, saying "here it 
is". Your example(s) say nothing about how badly your method can fail in 
some different example. And no, don't ask me to show you a failure example. 
It would be prohibitively time-consuming to find failure examples for every 
elabortate method that could be defined. You, the proponent of it, are the 
one to demonstrate its properties. And not with one or two examples, but by 
demonstrating that it meets some desirable criterion. And unless that 
criterion is a popular one like Condorcet's Criterion or the Majority 
Favorite Criterion, you should say something about why it should matter to 
us if a method meets that criterion.

A method meets a criterion if there's something good that it will always do, 
or something undesirable that it will never do. Showing that, in one 
example, your method does or doesn't do something is meaningless.

You continued:

So how about a few
examples of WV doing a better job of "deterring offensive order-reversal"?

You see, this is what I was talking about: "A few examples" show nothing. WV 
has been demonstrated to meet criteria that tell its strategy properties.

If you believe that your method is better than wv, or as good, then it's up 
to you to demonstrate that your method meets some meaningful criterion that 
wv doesn't meet. And if you can't show that your method meets the criteria 
that wv meets, then your claim that your method is as good as, or better 
than, wv requires that you show why the criteria that your method meets and 
wv doesn't meet are more important than the criteria that wv meets and your 
method doesn't meet.

Mike Ossipoff

_________________________________________________________________
There are now three new levels of MSN Hotmail Extra Storage!  Learn more. 
http://join.msn.com/?pgmarket=en-us&page=hotmail/es2&ST=1