3
to minimize the grants duration; the mathematician, of course,
wanted maximum duration. This table on the instruction sheet indi-
cated how the grant would be awarded.
FOUNDATIONSCHOICE
4
Using the information that the foundation people wanted to
minimize the duration of his grant, the mathematician saw that if he
picked R2 his minimum grant would be for one year, if he picked
R1 it would be for no less than two years, but if he picked R3 it would
be for at least five years. So he picked the application form that gave
him the maximum of these minimums, R3. The foundation people,
knowing that the mathematician would try to maximize the years
of his grant, saw that if they chose C1 the maximum grant would be
for nine years, C3 would be for eight years, and C2 would be for
only five years. So they chose the column with the minimum of these
maximums, C2. The intersection of R3 and C2 is five years, and both
the foundation people and the mathematician were satisfied. Acting
rationally neither could do better.
The mathematician and the administrators were playing a two-
person zero-sum game, described by mathematical theory as a game
having two players, a row player-in this case the mathematician-
and a column player-the foundation. In two-person zero-sum games,
one players loss is always the negative of the other players gain and
the sum of gain and loss is zero. The table of values, or payoff matrix,
lists payoffs from column to row player under all game conditions.
Each player makes only one move and that without knowledge of the
other players move. It is assumed that both players are rational, and
will try to do their best within payoff matrix restrictions.
Such games can be solved for payoffs that are mutually satisfac-
tory to both players. This payoff matrix, whichcontains negative num-
bers, is a little more complex than the one used for the mathematician-
foundation game, but negative numbers represent the row players
payoffs to the column player in the same way that positive numbers
COLUMN PLAYER
5
represent the column players payoffs to the row player. If they are
thought of as dollar values, this game is no more difficult than
the mathematician-foundation game. Readers are invited to find its
solution, which includes the column that the column player should
play, the row the row player should play, and the resulting optimum
payoff. Answers will be given in the April, 1969, issue of THE
SCIENCES.
THE THEORY OF CONFLICT
6
orists to include the culled knowledge of psychology and sociology
in their models. The reply is likely to be a variation on the theme
that for the most part, this knowledge is not in a sufficiently pre-
cise form to be incorporated as assumptions in a mathematical
model. (Games and Decisions) Game theory is a beautiful abstrac-
tion, its critics conclude, whose practical value is relatively small.
But game theory has been successfully applied in both economic
and warfare situations. Economists at the American Telephone and
Telegraph Company, for example, use game theory to help decide
on the optimum financial strategy for making short-term investments
of the corporations liquid assets. Mathematician Dr. Rufus Isaacs in
Diflerential Games (John Wiley & Sons, 1965 ) describes games de-
veloped to solve such military problems as the optimal trajectories
of ICBMs and how, ideally, a general divides his armys efforts be-
tween destroying the enemys source of weapon supply and directly
attacking military targets. In the contest between opposing armies-
or nations-game theory indicates optimal strategy. When forces of
any type are approaching one another, Dr. Isaacs writes, how best
should each distribute (or conserve) his firepower? Too early action
is wasteful because of the small hit probability; too late action risks
the enemys firing first and decimating too much potential fire before
it is put into action.
PLAYING AGAINST NATURE
7
of Surgery at Alfred Hospital in Melbourne, Australia, cites an
example of a physician playing a two-person zero-sum game with
Nature. A child is admitted to the hospital with acute abdominal
pain which is first central and then in the right iliac fossa. There is
fever and tenderness in the right lower abdomen. The differential
diagnosis lies (for simplicity) between acute appendicitis and acute
non-specific mesenteric lymphadenitis. Often there will be confirma-
tory signs of one or the other but not infrequently the surgeon is left
with the feeling that although he is doubtful about the existence of
appendicitis he cannot be sure. He will then operate prepared to have
a live mistake rather than a dead certainty. The payoff matrix for
this game is:
NATURE PLAYS