Why I Don’t Rob
Banks for a Living
Can game theory ‘prove’ that online robbery is irrational?
By Nicole Immorlica
DOI: 10.1145/1925041.1925042
Why be nice when the mean kids get all the candy? In many economic situations, people face a choice between being “nice” (cooperating with others for mutual benefit) and being “mean” (seeking immediate personal gain at the expense of others). This article uses game theory to mathematically prove that being nice is
not merely a personality trait, but also a rational decision that maximizes a person’s wealth and
happiness throughout his or her lifetime. This may help explain the continuing success of online
anonymous markets such as eBay, which rely on people to cooperate in order to function properly.
One of the most interesting developments is the creation of general-purpose markets for crowdsourcing
diverse tasks. For example, in Amazon’s
Mechanical Turk, tasks range from labeling images with keywords to judging the relevance of search results to
transcribing podcasts. Such micro-task
markets typically involve small tasks
(on the order of minutes or seconds) that
users self-select and complete for monetary gain. These markets represent the
potential for accomplishing work for a
fraction of the time and money required
by more traditional methods.
I am a computer scientist and an
economist, and so I believe the world
can be explained, or at least significantly illuminated, by careful mathematical modeling. Thus when I ask
“Why not rob a bank?” the answer I
am looking for is not “It’s immoral,”
or “It’s illegal,” but rather “It’s not rational.” That is, it doesn’t increase
our wealth and happiness. I believe if
it were rational, we would all be bank
robbers and morality would adjust accordingly. This viewpoint is admittedly
extreme, and even I don’t completely
buy it, but let’s run with it for a while
and see where we get.
C Abel:
D
Cain:
D
15, 35
20, 20
C
30, 30
35, 15
A MATHEMATICAL MODEL FOR CHEATING
The first step is to model bank robbing
as a mathematical formula. I’m going
to model this using games. The players
of this game are, generally speaking,
you and me and every other person on
earth, but to keep things simple we will
squint our eyes and assume society
consists of just two people. Let’s call
them Cain and Abel.
Here Abel (the row player) chooses a
row of the matrix. Row ‘C’ corresponds
to cooperating with Cain whereas row
‘D’ corresponds to defecting against
Cain. Similarly Cain (the column
player) simultaneously chooses a
column ‘C’ or ‘D’. These choices specify
a cell of the matrix. A choice of ‘D’ for
Abel and ‘C’ for Cain, for example,
corresponds to the bottom-left cell. The
payoffs for the players are the numbers
in the cell; Abel, the row player, gets a
payoff equal to the first number ($35 in
this case), whereas Cain, the column
player, gets a payoff equal to the second
number ($15 in this case).
The interesting point about this
matrix is that it is mutually beneficial
for both players to cooperate. In fact,
if both cooperate, this maximizes the
total sum of money received (and so, in
some sense, it is the best outcome for
society). However, neither will do so,
if he is rational and tries to maximize
his payoff. For if Cain believes Abel
will cooperate (i.e., choose the top
row), then Cain receives $30 if he
