[EM] BeatpathWinner/CSSD and FBC
MIKE OSSIPOFF
nkklrp at hotmail.com
Fri Mar 4 04:40:12 PST 2005
I haven´t found an FBC failure for wv Condorcet with 3 candidates. I´ve
checked the following scenario: Three candidates: A, B, & C. B is CW.
The names of the candidates are also used to represent the number of people
who have that candidate as favorite. Except that C representst the number of
people who have C as favorite and don´t use defensive strategy, and C1
represents the number of people who have C as favorite and do use defensive
strategy.
The A voters use offensive order-reversal against B. The C voters, as I
said, don´t use any defensive strategy, and rank sincerely. The C1 voters
use defensive strategy to try to save B´s win.
Sincere rankings:
A: ABC
B: BAC
C: CBA
The B voters, in the scenario that I checked, rank sincerely, BAC.
Voted rankings in the scenario that I checked:
1.
A: ACB
B: BAC
C: CBA
C1: (C=B)A
2.
A: ACB
B: BAC
C: CBA
C1: BCA
Though it´s possible that I made an error, I found that if the C1 voters
can´t save B by equally ranking B and C in 1st place, then they can´t save B
by ranking B over C. So that scenario doesn´t give an FBC violation for wv
Condorcet.
At least 2 of the inequalities that have to be satisfied for the C1 voters
to not be able to save B by equal ranking, but to be able to save him by
favorite-burial, contradict eachother. Unless I made an error.
I tried the same scenario for margins Condorcet, and again found that if the
C1 voters can´t save B by equally ranking B with C, then they can´t save B
by ranking B over C.
I emphasize that even if it turns out that wv and margins pass FBC, margins
still fails WDSC and margins is still falsifying. And wv, including PC,
passes SFC, WDSC, & Nonfalsifyingness. And SSD, BeatpathWinner/CSSD, SD,
and RP additionally pass GSFC, SDSC, and probably Expressiveness.
As I said, there could be an error, but probably not, but even if it´s true
that wv and margins can´t fail FBC in that scenario, that doesn´t
necessarily mean that they can´t fail FBC with 3 candidates. Maybe if the B
voters would defensively truncate, ranking only B, it would be possible for
the C1 voters to save B by ranking B over C, even when they couldn´t save B
by ranking B equal to C.
And even if it turns out that that scenario too doesn´t give an FBC failure
for wv &/or margins, that doesn´t mean that one couldn´t be found when there
are more candidates.
Could it be shown that if the failure can´t be found with N candidates then
it can´t be found with N + 1 candidates--or, more specifically, that if it
can´t be found with 3 candidates then it can´t be found at all?
Mike Ossipoff
_________________________________________________________________
On the road to retirement? Check out MSN Life Events for advice on how to
get there! http://lifeevents.msn.com/category.aspx?cid=Retirement
More information about the Election-Methods
mailing list