[EM] Optimal methods for multimember elections

[Forest Simmons]

On Fri, 13 Dec 2002, Alex Small wrote:

> Forest Simmons said:
> > Each voter marks one candidate on the ballot.  These candidates become
> > proxies (for the voters that marked their names) in an Election
> > Completion Convention.
> >
> > If there are n seats to be filled, and there is a subset of n candidates
> > each of which receives more than 1/(n+1) of the vote, then automatically
> > they distribute all of their proxy votes to self, and they win the
> > seats.
> >
> > Otherwise, since each proxy knows how many votes are controlled by every
> > other proxy, the candidates with common interests can get together and
> > decide on how to distribute their votes for best effect, i.e. so that
> > not too many votes, if any , get wasted.
>
> Am I correct in assuming that the candidates would essentially use
> something akin to cumulative voting?
>
Yes, cumulative voting with perfect information and the opportunity of
coordination of votes that cannot be done as easily in the non-proxy
(stand alone) version of Cumulative.

> This seems to be similar to closed party list systems, except that with a
> closed list the vote sharing among like-minded candidates has been decided
> in advance.  If the list gets enough votes for one seat the winner will be
> A.  If they get enough for 2 seats then B will also win, etc.  With your
> proposed system, there is more uncertainty.  I don't see any advantages
> favoring proxy PR over closed party list.

You're right, they are equally simple, and for large numbers of seats
lists would definitely be better.  Lists have history so they wouldn't be
considered experimental.

However,

(1) It seems to me that closed lists build walls, while cumulative builds
bridges.

(2) Proxy gives you the opportunity to support your favorite candidate
directly rather than just vote for his party. Perhaps you don't even like
the other candidates of his party, and he is third down the list.

> However, I do see a few advantages favoring proxy approval over regular
> approval (although I admit to preferring regular approval).

I prefer regular, too, but the four points I mentioned to Steven Brams
make proxy an easier sell.

Forest

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em