[EM] 05/13/02 - The Education of Poor Richard:

[Richard Moore]

Craig Carey wrote:
 > -------------------------------------
 > At 02\04\10 17:50 +1200 Wednesday, Craig Carey wrote:
 > ...
 >  > >At 2000\12\25 06:37 +1300 Monday, Craig Carey wrote to
 >  > >instantrunoff-freewheeling:
 >  > >...
 >  > >>Here is (6.2): ((AB)->(B), Cw->Bw: Allowed but vote wastage):
 >  > >>
 >  >                   (AB)<-->(B)
 >  > >>    (6.2):    -------------
 >  > >>               AB   2    1
 >  > >>               B    0    1
 >  > >>               BC   1    1
 >  > >>               CB   2    2
 >  > >>              -------------
 >  > >>           AV:      C    B    :   (AB {C+)-(B+ )
 >  > >>         IFPP:      C    B    :   (AB {C+)-(B+ )
 >  >     pref-Approval:    B    B
 >  >            FPTP:     A|C  B|C
 >  > >>
 >  > >>        Total =  5.   Quota = 1.6666..
 >  > >>
 >  > >...
 >  > >
 >  > >
 >  >*>IFPP is partly derived here: http://www.ijs.co.nz/quota-13.htm ]
 > ...
 > -------------------------------------
 >
 > I can't recall the exact definition of IIA, but I recall that that
 > example eliminated the method (once the additional mathematics is
 > done: i.e. the derivation of the ideal 3 candidate preferential voting
 > method).

Craig, it seems as if you are arguing that, because IFPP fails IIAC, IIAC
must be defective. Is that correct?

I can accept that you have a set of axioms that constitute your notion of
an ideal method, and that IIAC is not part of that ideal, but there is
nothing in the fact that IFPP fails IIAC to compel anyone to accept
that IIAC isn't desirable or necessary. You can reject IIAC in your
system, while a different system, representing a different notion of
the ideal, can have IIAC as an axiom.

It's like arguing that non-Euclidean geometries must be wrong because
the angles of a triangle fail to add up to exactly 180 degrees in those
geometries.

Or like arguing that non-IRV methods must be wrong because they have
a possibility of electing candidates that IRV eliminates.

   -- Richard

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