There
is a good chance that over the holiday period many of us will have indulged in
one or more games with our relatives and friends. The aim of the standard board
games or card games is normally to be the sole winner, although some games
require cooperation such as Save the
Whales, Pandemic or Max.
The game theorists are into
other sorts of games, which delve more into our psyche. One of their favorites
is the Prisoners Dilemma game in its
iterative version. In case you’ve forgotten the game, it’s very simply the
situation beloved of the writer’s of police dramas. There are two prisoners.
Each has the choice of sticking to a prior agreement to say nothing, in which
case they both get one month in jail. If one defects and rats out the other,
he/she walk free and the other gets a year inside.
We should note that if each
rats on the other, they both get three months in jail. The logic goes that
during iterative play, the players both become defectors and cooperation is an
unstable condition.
Those with large computers
to hand don’t have to troll round their police stations or student bars to find
real players, but can do this with simulations of large populations. Wang, Han
and Han at the Chinese Academy of Sciences in Beijing are among the most recent
to try a variation (1).
Their simulation used a
large population in which everybody could interact, but any pair only play each
other once. They showed that a population that started out with 50% of the
population trying to be cooperative rapidly turned into one with only about 15%
trying to be cooperative. This was the baseline.
The fun began when they
stuffed stooges into the mix who were instructed to cooperate if the other
players record indicates that they are probable cooperators, otherwise they
defect. The governors of the game only make the opting records of the players
available to their stooges.
The addition of stooges was
termed “soft control” and if there were a significant number of stooges in the
population, more cooperation was the result. The amount of cooperation was a
function of the number of stooges, though a large number are required to make a
large difference.
Other variants have been
reported with additions to the population of agents who can add feedback, or teaching,
to the population, but this would be a form of “hard control”. Soft control is
taking place without the population realizing it in the case studied here and,
of course, our Ad Agencies are already aware of its effectiveness.
Please note that this
blog is migrating to
in 8 days.
