[EM] Identifying ranked ballot sets

John john.r.moser at gmail.com
Mon Oct 1 16:18:03 PDT 2018


[Note:  I'm not on-list; please CC me on replies]

Somebody suggested a two-ballot set to me with three candidates, and
it turns out you can mutate this set and achieve the same pairwise
votes.

BAC
CAB

      LOSE
     A  B  C
W A  _  1  1
I B  1  _  1
N C  1  1  _

You'll note this set also derives from BCA ACB.

This expands to larger sets:

BACD = BDCA
DCAB = ACDB

      LOSE
     A  B  C  D
W A  _  1  1  1
I B  1  _  1  1
N C  1  1  _  1
  D  1  1  1  _

If we clear the first votes:

      LOSE
     A  B  C  D
W A  _  .  .  .
I B  .  _  .  .
N C  1  1  _  1
  D  1  1  1  _

(B)DCA = (B)CDA
(A)CDB = (A)DCB

I'm trying to identify a concise representation of a set of ranked
ballots.  At this time, I believe this may be:

1.  The pairwise race matrix (Win-Lose)
2.  The matrix of appearances at each ordinal rank

For BACD and DCAB, #2 is as such:

   1  2  3  4
A 0  1  1  0
B 1  0  0  1
C 0  1  1  0
D 1  0  0  1

It probably doesn't matter if you rank the truncated ballots by
counting all unranked as ties at that rank or by not counting them at
that rank.


More information about the Election-Methods mailing list