[EM] Answers to selected Steph statements

[MIKE OSSIPOFF]

I'd said:

 >
 >He's still CW. Truncation doesn't change who's CW. CW
 >is about
 >sincere preferences, not votes. Nor does it
 >change the fact that a majority of all the voters have
 >indicated on
 >their ballot that they prefer one candidate to
 >another.

Steph replied:


Once you face the S2 set of ballots "in the box",
how can you know if it comes
from an S1 set which has been truncated for strategical purpose,
or if S2 is the sincere ballot set?

I reply:

It doesn't matter, except that if that candidate is a sincere CW,
then we can apply SFC. If he's a member of the sincere Smith set
then we can apply GSFC.

Sure, we can't tell whether a short ranking is the result of strategic
truncation, lazy truncation, in-a-hurry truncation, or sincere lack
of preferences among some candidates. All I'm saying is that if the
candidate X in SFC is CW, then SFC applies, and if he's in the sincere
Smith set then GSFC applies. Then, with complying methods, we won't
have the majority rule violation and strategy problem of a Y victory.

Yes, wouldn't it be nice to protect majority rule and the power of
a majority to get its way without strategy, all the time. Well,
WDSC & SDSC are more general, in the sense that they don't require that
no one falsifies a preference, and they don't require that some particular 
candidate is the CW, or in the sincere Smith set.

SFC & GSFC merely tell some reasonable conditions in which complying
methods free a majority from strategic need. We have no need to go
through the ballot box trying to guess why each short ranking is short.
The reason why SFC & GSFC make more demands in their premises is
because they offer more--complete freedom of strategic need for the
majority referred to.

Before the election, the voter can at least be assured that a
majority-supported CW or sincere Smith set member won't have victory
taken from him because that voter didn't strategize to protect him.
That's all SFC & GSFC are intended to guarantee, but I believe that's
a lot. GSFC is a powerful strategy guarantee.

Steph continued:

If you dismiss the 2nd case,
you assume sincere preferences cannot be of the kind
A > B > C > D = E = F. Why not ?

I reply:

Sincere preferences of course can be of that kind.

I agree with the importance of protecting from truncation
because there is no way of finding back the original set
S1 if truncation occurs. To obtain this incentive, we need
a negative esperance from truncation.
It is the case, with both (rm) and (wv).
You seem to think (rm) produces a positive mean gain.
It is just less negative than with (wv).

I reply:

I've never claimed that wv methods punish truncation. I've never
said that truncation should be punished. It's often done sincerely.
In fact, it's a defensive strategy against offensive order-reversal
in the wv methods. Truncation is harmless in wv methods. Of course
no method can help the voter who doesn't support a compromise that
he needs. In rank-balloting elections, my ballot will never rank
all the candidates under existing political conditions. It will only
rank a few. There will be no Democrats or Republicans in my ranking.

Steph continued:

On the other hand,
it maximizes equity as minimizing the biggest (twarted?, thwarted ?)
overturned majoriy...Thus it maximizes the probability of having
no wrong pairwise comparison.

I reply:

Wrong by what definition? Blake makes that claim too. But it's meaningless 
to speak of the best candidate, or a right or wrong
pairwise comparison unless we have an operational definition of those
terms. An operational definition is an apply-able definition. A definition 
written so as to be usable for determining if the definition
is complied with, a definition that provides a test for compliance with
it. It would even be good enough if the test involves a "thought
experiment" that can't really be carried out. But Blake has never
given us such a definition of the best candidate, or a right or wrong
pairwise result. That's because there isn't one.

Steph continues:


It seems I cannot develop a semi-recursive evalution of results for
ranked-pairs as the number of voters increases. We will have to set
for a small number of voters as guideline, except if anyone has a better
idea...

I reply:

There are various ways to evaluate a method's results. We use
criteria for that, for instance, or the assurance or lack of assurance
Nash equilibria with no defensive order-reversal, or social utility
or similar social optimizations, such as Approval's maximization of
the number of voters who consider the winner so good that they'd have
preferred to directly put him/her into office instead of holding the
election, if they could have done so.

For evaluating rm in comparison to wv, I recommend the standards
of majority rule and getting rid of the lesser-of-2-evils problem.
The majority defensive strategy criteria are the criteria that have
been posted here for measuring for that very widely-accepted
standard. I also suggest the standard of sincere Nash equilibria
for comparing those 2 methods. I hope that some of these suggestions
are helpful for evaluating rm's results.

Mike Ossipoff


_________________________________________________________________
Send and receive Hotmail on your mobile device: http://mobile.msn.com

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em