[EM] The "Turkey" problem (to Dave)

[Bart Ingles]

Dave Ketchum wrote:
> 
> Coming back to "universally disliked", if this label is true this
> candidate is not going to get ranked high by enough voters to win.
> 

       Liked <-----------------------> Disliked
votes ------------------------------------------
49%   A                                    B   C
 2%   B      A                                 C
49%   C                                    B   A

You could call this a "King Solomon" example, in that since A and C
can't agree on a winner, someone unacceptable to either wins.

Another way to look at this is to think of the weaker B>C and B>A
preferences as "noise", and the stronger preferences as "signal". 
Condorcet amplifies the noise (or clips the signal) so that all have the
same weight.  I don't know about you, but I can usually stomach having
my own major preferences overruled by a larger faction's major
preferences, but would find it a bit sickening to have my major
preferences overruled by someone else's "noise".  



Kevin Venzke wrote:
> 
> One of the main things I was arguing was that truncation doesn't benefit
> the voter who does it (at least in cases where there is a CW).

Not entirely true, although this involves a prisoner's dilemma.  

If, in advance of the election, the vote count is not known, but A and C
are believed equally likely to win in a head-to-head contest, the A and
C factions can agree to truncate and block B from winning, thereby
giving A and C each a 50% chance of winning.  This increases the
expected utility of the outcome for both A and C, if B was a likely
winner on fully-ranked ballots.

Some believe a prisoner's dilemma strategy to be unworkable, since
either side can "cheat" the other by failing to truncate, but the cost
of being cheated seems low in this case.  If nothing else, it seems like
a bit of a dilemma to decide how to play the prisoner's dilemma.