[EM] "More often" (was: IRV and Condorcet operating identically)

[Markus Schulze]

Dear Craig,

you wrote (2 March 2003):
> Consider this (irv-wrong-winners, yet another wrong Mayor in
> office) See that there is a large support rise for A and that
> causes A to lose.
>
> >   ----------------------------
> >     A       19999      80004
> >     B           1          5
> >     BA      19997      19997
> >     CB      40002      40002
> >     DBA     20001      20001
> >   ----------------------------
> > Total:     100000     160009
> > AV Winner:      A          B
>
> That looks like a "classical participation axiom" problem (but
> it is not since B got 4 votes).

In your example, the fact that the winner is changed from candidate A
to candidate B is caused only by the addition of the 4 B voters. Your
conclusion that "the large support rise for A causes A to lose" is
false.

Actually, when IRV is being used and candidate A is the unique winner,
then a set of additional voters who strictly prefer candidate A to
every other candidate can never make candidate A lose.

Markus Schulze

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