[EM] [CES #8967] Score Voting and Approval Voting not practically substantially different from Plurality?

[Jameson Quinn]

> I also believe that there is too much emphasis being given to
> combatting "strategic voting". With the exception of lesser eviling,
> which, I suppose, could be considered in this category, this is not
> such a big problem, and certainly should not be used as an excuse for
> supporting voting systems that prevent voters from expressing their
> views maximally, or which have terrible pathologies.


I suspect you intend to include Majority Approval Voting and other Bucklin
systems in the category "voting systems that prevent voters from expressing
their views maximally". But you have it backwards; in MAV, all votes have
maximal effect, pulling a given candidate's median towards or away from
winning. The fact that some votes aren't more powerful than others is a
good thing. It is in Score that some votes are (voluntarily) less than
maximally powerful; this is exactly the reason why Benn (largely but not
entirely correctly) worries that Score would or should reduce to approval.

I am not, of course, claiming that MAV is entirely strategy-free. As
Gibbard (and, less cleanly, Satterthwaite) proved, that is impossible.
Strategy failure in MAV is when someone rates the two frontrunner
candidates either both above both their medians, or both below both. If
there are enough of one kind (high or low) of such ballots which also
disagree with the finish order between the frontrunners, that can lead to a
participation failure. But in general, that would not happen between two
ideologically distant frontrunners, because almost all voters will
naturally (honestly) tend to vote one high and one low (unless they're
intentionally casting a "none of the above" vote, deliberately giving up
voting power in order to make a symbolic protest). Thus, it would only
happen between two similar candidates; that is, in a chicken dilemma
situation. MAV has relatively good resistance to the chicken dilemma, and
while it can in theory fail with purely rational voters, even a few honest
voters are enough to inoculate it against such failure. In sum, I think
that MAV's participation failure is just a theoretical, not a practical,
concern; in the real world, it would occur in a small fraction of elections
for a small fraction of voters. (I'd conservatively estimate that it would
be under 25% in both cases, so under 6% risk for a given voter).

Jameson
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20130625/13ce4841/attachment-0004.htm>