[EM] proper Borda implementation

James Green-Armytage jarmyta at antioch-college.edu
Thu Sep 30 04:22:37 PDT 2004


>Hello,
>I'm not sure of the best way to implement Borda with ballots that don't 
>rank all the candidates.  Suppose that 4 candidates are running.  With a 
>ballot like:
>	A B C D
>A gets 4 points, B gets 3 points and so on.  But what about a ballot like
>	A B
>A and B should be treated as above, but what about C and D?  I see two
>options:
>(1) C and D each get nothing.
>(2) C and D each get 1.5 points (average of leftover points).
>Any thoughts as to which is better?

	Actually (although I am not a person who likes the Borda count at all)
I'd be inclined to give them each 1 point, if 1 is the minimum score on a
fully completed ballot. If not that, then I prefer (2). That way, if there
are two candidates, it doesn't make a difference whether you rank the
candidate whom you like less; if it did make a difference, I think that
would be pretty dumb.
	But, I'm curious about why you are asking the question in the first
place. Are you interested in using Borda for an election, or do you have
some other purpose in mind? I would be willing to accept Borda for some
purposes, e.g. where the voters aren't human beings (maybe machines
receiving and assimilating data) and/or the result isn't especially
important (all-star baseball picks, or whatever), but if one was electing
a mayor, a city council, a board of trustees, or even a student
government, I'd not like to see Borda there. Pairwise and STV are both
much more elegant.
	Why not Borda? My reasons are summarized in a rough essay which I wrote
last fall, and which can now be found at
http://fc.antioch.edu/~jarmyta@antioch-college.edu/voting_methods/antiborda.htm

my best,
James




More information about the Election-Methods mailing list