MATHEMATICAL RECREATIONS
by fan Stewart
Monopoly Revisited
tn the April column 1 deserthed 3
‘mathematical model of the board
game Monopoly. At the start of
the game, when everyone emerges from
the GO position by throwing dice, the
probability of the first few: squares he-
ing occupied is high, and the distant
squares ace umoccapied, Using the cor
cept nf Markov chains, | showed that
this initial bunching of probabilities
timately evens out sa that the game is
fairs everyone has an equal chance to 0<-
cupy any square an to buy that prop-
ety. This outcome is tc, however
only when certain simplifying assump-
tions are made, Monopoly enthusiasts
were quick to point out that in the weal
game, the long-term distribution of
probabilities isnot even,
So what are the true probabilities?
‘The Markov chain method can alsa he
applied to the real game; Have to warn
yon, however: that the analysis is com=
plex and requires substantial computer
assistance. Let me frst remind you how
Markov chains are used for Monopoly.
Aplayercan be is any one of 40 squares
oon the board, which, for convenience,
vve number clockwise from zer0 10 39,
starting with GO (which is zero}.
Given any two squares A and B, there
isa quantity called the transition prob-
ability—the probability that player
who starts from A will reach B a the
conclaston of his or ber turn at throve
ing the dice. If this move is impossible,
then the transition probability is zero,
“There are 40 x 40 = 1,600 ransition
probabilities in al, and they ean conve:
hiently be encoded in a square matrix
(M with 40 horigontal sows and 40 yer~
tical columns. For example, the entry in
the sixth row and 10th column deseribes
the probability of moving from Read-
ing Railroad to Connecticut Avenue in
fone turn. The inital probebiliies for a
player are 1 for posinon Oand 0 forall
the rest; they can be encoded as 2 vee-
rorya (1,0, 0)
“The theory of Markoy chains tells us
that the evolution of this probability
distribution is given by tte sequence of
vectors ¥y My MPy, MPy and. soon:
each throw of the dice corresponds to
the marnix M operating on the vector ¥,
The resulting veerors can be calculaeed
by standard matrix methods, avaitable
on any good computer algebra. pack
age. Such packages can also ealeulate
the so-called eigenvectors and eigenval-
ues of M.A vector U is an eigenvector
swith eigenvalue cif Mu =¢ xu, where
can be a real or complex number:
Markov's key theorem is that the long-
‘erm probability distribucin is given by
the eigenvector whose eigenvalue has
the largest absolute value.
So in order to analyze the fairness of
Monopoly, all we need to do is cor
pure M and apply matrix algebra. For
‘my simplified model this was easy, but
for the real game we must also take into
Fy 4s
snc nko
AL rinows
=
HUMBER OF SQUARES Pron CLARENT POSITION
PROBABILITIES OF MOVING
a given nunber of squares in one tart, after accounting for rules
for throwing dice, peak at the number sevex. Ifa player throws double,
beor be gets to throw again, but three consecutive doubles meas Jail
ScUNIIFIE AstERICAN Ostober 1996
xO
&
Jal.
cand the marry ways to land brit,
‘makes Monopoly complex.
account multiple rolls of the dice, spe-
cial squares such as GOTO JAIL and ine
structions on cards thar players draw
when they land on CHANG: and COM-
MUNITY CHEST
Many readers sent me their analyses
ofthe guume, The most extensive were
{rom William J. Butler of Portsmouth,
RL, Thomas H. Friddell, 2 Boeing en-
aineet from Maple Valley, Wash., and
Stephen Abbott of the mathematics de-
partinent at St, Olaf College in Norch=
field, Minn, wha collaborated with his
colleague Matt Richey, Butler wrote a
Pascal prograen, Friddell used! Mathcad
and Abbott used Maple. The discussion
that follows is a synthesis of their re-
sults. (All models of Monopoly: make
assumpcions abour the degree of detail
tw be incorporated; there were insignifi-
cant cflerences in the assumptions made
by vatious cortespondemts.)
The first modification of my original
model is rake full account of the rules
for the dice. A pair of dice is chrown,
and if the esult isa double, the player
throws again, bur three consecutive dour
bles lands him or her in Jail. The throw
of the dice ia tiny Markov chain in its
‘own right and can be solved by the usu
al method. The resuit is 2 graph of the
probability of moving any given die
tance from the current position [see i-
lestration at left|, Notice that the most
likely lisence is 7, but that i is possible
to move up to 35 squares (by throwing
6.65 6,6; 6,5). Yer the probabilities of
Mathemetical RecreationsPROBABILITY (PERCENT)
5 OS
squane
LONG-TERM PROBABILITY DISTRIBUTION
shows thatthe Jil square 1s most likely to be occupied.
moving more than 29 squares are so
smvall that they fail 10 show up on the
raph, These resuits are incorporated
into M by appropriately changing each
individual entry.
Next the elfect of the GOTO ja
‘square must be included, The Jai les
pose « problem, because players can
lect to buy their way out oF stay in ancl
try to throw doubles to get out. (Or at
later stages, when Jail becomes a reftge
from high rents they ean stay in and
hope not to throw doubles!) The probs
abilities associated with this choice de
pend om the player's psychology, 40 the
process is nonMarkovian, Most corre
spondents ge around this poser by ax
suming thatthe player did not buy his
cr her way eat, Then Jal becomes not
so mmuch a single square as. Marko
subprocess—a series of three (viral)
squares where players move from Just
in Jail to In Jail One Tuen Alteady to
Mast Come Out of jail Next Tarn. The
GO TO JAth square itself has probabiliey
zero because nobody actually occupies it,
The next step isto modify M to ac
count for the CHANCE and COMMUN
TY GHEST cards, which may senda play-
er (0 Jail o¢ 10 some other positon an
the board. This refinement can be made
quite straight forwardly (if laboriously)
by courring the proportion of cards that
send the player o any given square. The
extra probability js then added to the
corresponding position in M.
A™ ‘SL George's sculptures in the May column stimulated a discussion of
how to make Uwee-dimensional
‘objects based on regular polyhedra,
Witiam J, Sheppard of Columbus, Ohio, sent detai's of Ns cunning method for
cutting a regular tetrahedron oF ootahedron trem salld wood, pining cut that
“sturdy, sold models are more corwenient than hollow medels made by tapine
together equilateral triangles.” His
methods can be found in the Journal!
of Chemical Education, Vol. 4a, page
1683; November 1967.
Nortnan Gallatin of Garrison, lows,
thas been working on Platonic solids
for a quarter of a century and has de-
veloped remarkable sculptures, some
‘made from mirror glass. The picture
at the right represents a thrae-