atarr at purdue.edu
Sat Feb 8 23:59:35 PST 2003
Mike Ossipoff wrote:
>I suggest that Eric not add margins-counting to his website, because
>, for one thing, having it would create some doubt for the user about
>which to use, requiring the website to add discussion about the relative
>merits of the 2 defeat-measures. The user would have to study the
>difference and relative merits of the 2 defeat measures, and answer an
>extra question at the input form. Also there's the unnecessary work of
>adding that meritless option to the programs. Simpler to just not have it.
>If someone wants an interactive count program to count margins Condorcet,
>they can use Blake's program.
Moreover, anyone can use a winning votes-based program to serve as a
margins-based count, simply by splitting every voter with an unexpressed
preference into two half voters, each of whom express one of the two
possible preferences. This eliminates all equal rankings or truncation, so
that every ballot is fully expressed. After this change, you will have the
same result from margins or winning votes at this point. This result will
not necessarily match the winning votes-based result of the original votes,
but it would match the margins-based result from those votes.
If I recall correctly, Rob LeGrand said that this is how the program on his
web site works. It is designed to take fully ranked ballots, so he splits
voters into half voters as necessary to make every ballot fully rank before
figuring out the winner.
If you are inputting a pairwise matrix, rather than a set of ballots, it's
still easy to convert a tally to margins. Just add an equal number of
votes to transposed positions on the matrix, until every transpose pair has
the same total number of votes.
Since no equivalent procedures exist to make a margins program emulate
winning votes, it strikes me as more egalitarian to write a winning-votes
based program. This way the user can decide (through very simple
manipulation of their input) whether they want margins or winning votes,
even if there's no check box. Of course, since I like winning votes more
than margins, I wouldn't bother with this manipulation.
For more information about this list (subscribe, unsubscribe, FAQ, etc),
please see http://www.eskimo.com/~robla/em
More information about the Election-Methods