Electoral College Reform starting with one or a few states

Mike Ossipoff dfb at bbs.cruzio.com
Mon Oct 21 17:04:45 PDT 1996


When you speak of transferring the electoral votes of dropped
unwinnable candidates, that implies that you've given them
electoral votes. There's no need to give electoral votes to anyone
, even temporarily, before the choice process is completed.

And these vote transfers are still committing your proposal
to a vote-tranferring method, or some kind of an effort to
hybridize a vote-transferring method with a pairwise-count
method. Obviously, in a pairwise method, there's no need to
tranfer votes from the dropped candidates, since the ranking
doesn't confer a vote--it indicates 2-candidate preferences.

This latest proposal of yours is closer to the original proposal,
but you haven't arrived there yet.

You say that, with 4 candidates, it's necessary to keep track
of 64 vote-sums. Well, that depends on the method, doesn't it.
With your Instant-Runoff, that's the case, and, with lots of
candidates it might be more efficient to just record the 
rankings themselves in computer memory.

But neither of these things is necessary with Condorcet's method.
All that needs to be recorded & sent to the central count computer
are the pairwise results for the various pairs of candidates.
So, with 4 candidates, there are 6 possible pairs of candidates,
meaning that there are 12 vote totals to record. For instance,
for the A & B pair, it's necessary to record how many ranked A
over B, & how many ranked B over A.

Aside from other tremendous merit differences, this far less
demanding computer memory requirement is an additional advantage
of Condorcet's method, compared to Instant-Runoff.

For example, in the current Presidential race, there are 8 candidates
on the California ballot. With Condorcet's method that means that
there are 28 pairs of candidates, & therefore 56 vote totals to
record at each precict.

But, with Instant Runoff, there are 109,552 rankings of various
lengths that can be made from 8 candidates, and if you want to
record how many people voted each of those rankings, then you've
got to record 109,552 totals instead of 56.

With more candidates, as would be likely in a rank-balloting
election, it would soon be unfeasible to record a total for
each possible ranking, and it would soon be much more efficient
to simply record every one of the 100,000,000 rankings. That's
if you're using Instant-Runoff, which requires the continued use
of the rankings, unlike Condorcet's method, which only requires
those 56 pairwise vote totals, for the 28 pairs of candidates.

And if you wanted to reduce your number of possible rankings to
record, by dropping the last choice from rankings that ranked
everyone, you've still got 69,232 possible rankings to record.


But, from what you've said previously, you'd record all 109,552
of them.


Mike
 






-- 




More information about the Election-Methods mailing list