Abstract:
We will discuss the simplest possible kind of dice game: two players roll
one die apiece and if the results differ, the winner is the player
whose die has
rolled the higher value. People have played such games (and much more
complicated ones too) since antiquity, but the idea of playing a simple dice
game with different dice seems to have been introduced in living memory --
Gardner's Scientific American articles attribute the idea to Efron
[unpublished]. Gardner's discussion centers on the two following
observations:
-
In such games the dice need not tend to have equal numbers of wins and
losses in the long run, even if their rolls have the same expected value.
For instance, the 3-sided die (3,3,4) is stronger than the 3-sided die (2,2,6).
Moreover, the relative strengths of dice need not obey transitivity.
For instance, the die (1,4,5) will lose to (2,2,6) more often than not, though
it is stronger than (3,3,4).
-
The second observation naturally suggests the celebrated theorem of
Arrow. Indeed, dice games provide a natural way to judge multicandidate
elections; if several candidates are assessed by n voters then each
candidate
has an associated n-sided die, which simply lists the scores the various
voters have given that candidate. The relative strengths of two candidates
may
be assessed by playing a dice game with their associated dice.
We will discuss several recent results about these dice games, and also
several unsolved problems. Most are related to the fact that surprisingly
few pairs of distinct dice tend to have equal numbers of wins and losses
in the long run.