Insincere Equal Rankings

[Hugh Tobin]

Mike, I am glad you have resumed posting on this list, and thank you for
elaborating on Markus's point about incentives to insincere equal
rankings at the top of one's ballot.  I had responded rather shallowly
to that point 13 months ago, to the effect that those with sincere first
preferences would not want to sacrifice them.  Now I see that in
Condorcet with a "votes-against" tiebreaker, equal first rankings could
be a rational strategy where one's preference between two favorites is
relatively weak and the pairwise races are so close that a circular tie
is possible, but one cannot predict which of one's favored candidates is
likely to lose to one's least favorite (in a 3-way race, if one did know
which of one's favored candidates was at risk of losing to the worst
evil, then instead of using "upward truncation" one would vote for the
other of one's favorites as first choice,  trying to make him or her
Condorcet winner, reversing true preference if necessary).   Markus, I
am sorry it took this long for me to get your point.

No Condorcet tiebreak system can eliminate the risk that one might be
punished for voting for one's true first choice, because there is always
the risk that such a vote will cause the need for the tiebreak, when
otherwise one's second choice would have won outright.  But with a
margins-of-defeat tiebreaker, if one has no way of knowing which
direction the circular tie runs, then I think one must conclude that
choosing between one's two favorites is just as likely to prevent the
worst outcome in a circular tie as to cause it, and could cause one's
favorite to win the circular tie over one's least favorite (by reducing
a margin of defeat), so one should simply vote a sincere ranking.

Perhaps insincere equal rankings at the top of ballots would not be
widespread in practice with Smith//Condorcet[EM], even if the ballot
form made them easy (as I have proposed). But the incentive does provide
another reason to question the [EM] tiebreaker.

I am not convinced by arguments dismissing such an incentive on the
ground that no method will help those who fail to support a compromise
that they need. If we know what compromise we need, and what majority we
are a part of, then we can elect the compromise candidate under the
prevailing plurality system. IMHO, a major goal in electoral reform is
to allow the voter to vote based upon his preferences without having to
guess which of them he must sacrifice in order to cause an acceptable
compromise to win.  Another, at least in my mind, is to give voters a
wider range of choice, among candidates who provide different mixes of
policies than the major parties offer and who appeal to voters based on
ideas and reason, rather than by positioning themselves on a
one-dimensional spectrum.  In short, I think successful reform should
make many elections sufficiently "chaotic" that it may indeed be
difficult to know a priori who is the needed compromise for whom, and
there may be natural circular ties.  Therefore I value a system in which
a voter with no information about likely votes of others always would
vote a sincerely ranked ballot -- not because he would never be punished
for it, but because the risk of being punished for insincerity is
greater.  

Markus, last year (5/6/97) you asserted that incentive to truncation was
a general problem with Condorcet systems, which I understood to mean not
limited to the [EM] tiebreaker.  Would you agree there is an important
difference between a possibility of being "punished" for providing true
information about one's preferences, and an incentive to rational voters
to conceal preferences?  Would you agree that the incentive exists only
when, based on the information that the voter has about the probable
votes of others, the risk-adjusted expected value to the voter from
truncation is greater than from ranking all candidates?  Do you see any
incentives to truncation, or insincere votes, by those voters who lack
information regarding probable votes of others, in Condorcet with a
margins-of-defeat tiebreaker?
 
-- Hugh Tobin