[SPAM: #] [EM] Approval Strategy

Alex Small asmall at physics.ucsb.edu
Thu Aug 7 23:24:01 PDT 2003


Neal Finne said:
> I'd be reluctant to use any electoral system that requires voters to
> vote strategically. Still, within those systems, approval is a huge
> improvement over plurality or IRV.

There is a theorem (named after its discoverers, Gibbard and
Satterthwaite) which states that no ranked single-winner voting system for
3 or more candidates can simultaneously be

1)  Non-dictatorial
2)  Pareto Efficient
3)  devoid of any incentives for people to vote insincerely

Since the first two criteria are satisfied by any voting system that might
seriously be proposed, it follows that any ranked voting system suggested
for single winner public elections will occasionally give voters an
incentive to vote insincerely, aka strategically.

(If you want a precise definition of a ranked voting system, and an
explanation of why Approval is not a ranked voting system, write to me
personally.  The list already has had a ton of discussion on the seemingly
simple question "Is Approval a ranked voting system?")

Approval and similar systems at least have the advantage that they never
give incentives for voters to reverse two candidates.  Approval forces
people to insincerely "rank" some candidates equal, but there's never a
situation where you have an incentive to approve B but not A when your
preference is A>B.  (Brams and Fishburn dispute this in their book, but
the examples put forward require unusual circumstances where voters have
very incomplete information.  In the case of either zero information or
adequate information their examples break down.  It's only in a certain
"twilight zone" that a little bit of information can be a dangerous
thing.)

So Approval and related systems might be as good as it gets in regard to
minimizing incentives for insincere voting.



Alex





More information about the Election-Methods mailing list