single-winner ballot with restricted rankings?

Steve Eppley seppley at
Sun Feb 16 11:55:14 PST 1997

                                                [bcc: Mike O]

I've been mulling over possibilities for *practical* ranked ballots
for single-winner elections, limited by my woefully incomplete
knowledge of ballot technologies.

Here's an option which violates Arrow's "universal domain" axiom,
but perhaps it wouldn't be a bad compromise:

Example 1:  a U.S. lefty's ballot, circa 1996

    RANK THE CANDIDATES     (Best)       (Worst)
    Candidate                1  2  3  4  5  6
    -----------------       ------------------
    Clinton   (Dem)                X
    Nader     (Green)        X
    Dole      (Rep)                      X
    Brown     (Lib)                   X
    Buchanan  (Rep)
    Jackson   (Dem)          X
    Gephardt  (Dem)             X
    Forbes    (Rep)                         X
    Alexander (Rep)                      X

It's okay for the voter to rank candidates equally.  Candidates left 
unranked will be treated as if the voter ranked them last (7th, in 
the above example.)

Key aspects of this format:
1. The number of distinct rank positions is sharply limited by the
width of the ballot.  
2. Common punchcards can be used so computers can quickly tally the 

My concern: would the limited number of rank positions significantly
impact the anti-spoiler, anti-lesser evil, anti-tactical, majority
rule properties of Condorcet's method?  How many positions is enough
to have a healthy multiparty democracy?

The truncation resistance property of Condorcet's method will 
probably be a factor.

If my brief thinking about this isn't screwed up, 3 or 4 explicit
positions might be adequate.  The candidate(s) perceived as the
"lesser evil" centrist compromise (i.e., expected to win if
unrestricted rankings were allowed) could be ranked, and the
"greater evils" could be left unranked:

Example 2:  a U.S. righty's ballot, circa 1996 (even narrower format)

    Candidate                1  2  3 
    -----------------       ---------
    Clinton   (Dem)                X
    Nader     (Green)        
    Dole      (Rep)             X
    Brown     (Lib)          X           
    Buchanan  (Rep)                X
    Jackson   (Dem)          
    Gephardt  (Dem)          
    Forbes    (Rep)          X
    Alexander (Rep)             X

Example 3:  a U.S. centrist's ballot, circa 1996 (narrow format)

    Candidate                1  2  3 
    -----------------       ---------
    Clinton   (Dem)          X
    Nader     (Green)        
    Dole      (Rep)             X
    Brown     (Lib)          
    Buchanan  (Rep)                
    Jackson   (Dem)          
    Gephardt  (Dem)                X
    Forbes    (Rep)                X
    Alexander (Rep)             X

Perhaps by the time we achieve widespread preference voting in this
country, voting booths will have computers and the practicality
issue will be moot.  In the meantime, this format may be useful
for demonstration purposes.  And its practicality might boost the
reform movement. 

More information about the Election-Methods mailing list