[EM] Better reply to Blake's example

[MIKE OSSIPOFF]

>46 D A B C -- A B C much lower rated
>19 A B C D -- D much lower
>18 B C A D -- D much lower
>17 C B A D -- D much lower
>
>IRV picks B.  Unless the A/B/C voters really get their act together, or
>have some kind of primary, D will win.  If one of A/B/C does win, who
>will decide which one?   A very small group of voters who were counting
>on other people to make sure D lost.
>
>Picking D is a failure of MMC and GITC (for a ranked method).  Is
>picking one of A/B/C a violation of DSC.

Actually yes. I was in a hurry when I replied yesterday. Let's say
that your rankings are people's sincere rankings. Let's say that everyone
but the D voters vote as you say they do. B wins in IRV. But 65% of
the voters prefer A to B. The A voters can't help A more than they
already are. The D voters could make A win instead of B, if they
ranked A in 1st place over their favorite.

This example is an example of IRV violating FBC, WDSC, & SDSC.

Whether it violates SFC & GSFC in this example too depends on
determining the CW &/or the sincere Smith set. I'll do that later
and comment tomorrow if I find a violation of SFC & GSFC here too.

So your example demonstrates what I said: MMC examples tend to typically
be IRV failure examples for the defensive strategy criteria.

I know: I'd said that they always are. Actually, the mere fact of
an MMC example doesn't specify enough detail to make examples for
the defensive strategy criteria. Your example gives those details
and shows how typical that failure is in MMC examples.

But let me contrive a criterion that will always be failed when
what we know about an election is that it's an MMC example:

FBC2:

It should never happen that, for some voters, voting someone over their
favorite dominates every other voting strategy.

Mike Ossipoff




>---
>Blake Cretney
>

_____________________________________________________________________________________
Get more from the Web.  FREE MSN Explorer download : http://explorer.msn.com