Algorithm •At
every
tick,
randomly
pick
n
cells •Compare
features
of
the
culture
with
each
of
its
neighbours •If
the
trait
difference
for
the
same
feature
between
2
cultures
are
<
t,
randomly
select
either
culture
and
copy
the
trait
of
the
feature
to
the
other
culture
The
more
similar
2
cultures
are,
the
more
likely
there
will
be
cultural
exchange Probability
of
cultural
exchange
=
1
-‐ difference
between
2
cultures
÷ 96
What
do
we
want
to
measure? •Average feature
distance
tells
us
how
far
apart
(different)
the
cultures
are •Uniques tells
us
the
number
of
unique
cultures
at
any
given
point
in
time •Changes tells
us
how
vibrant
the
cultural
exchanges
are
Observations •Eventual
equilibrium
is
only
a
few
dominant
cultures •Dominant
cultures
can
be
quite
different
from
each
other •Smaller
areas
results
in
faster
equilibrium
and
smaller
number
of
dominant
cultures •A
culture
that
is
more
dominant
at
a
point
in
time
doesn’t
mean
it
will
be
dominant
in
the
end
Algorithm •At
every
tick,
check
every
cell •If
at
least
n
number
of
its
neighbours are
of
the
same
‘race’,
do
nothing •Otherwise,
randomly
pick
an
empty
cell
and
move
there
Parameters •Acceptable number
of
neighbours (n) •Number of
‘races’
in
the
grid
(r) •Percentage
of
vacant
cells
(v) •Policy
limitation – cannot
have
more
than
(l)
number
of
neighbours of
the
same
race
Observations • Segregation
happens
even
if
there
is
weak
preference
for
neighbours of
same
type • The
weaker
the
preference,
the
less
segregated
(smaller
clusters) • The
stronger
the
preference,
the
more
segregated
(larger
clusters) • At
a
threshold,
stronger
preference
results
in
an
unstable
but
non-‐ segregated
state
(occupants
always
moving
to
another
cell) • Number
of
races
have
no
impact
of
segregation • Number
of
vacant
cells
have
no
impact
on
segregation • Policy
enforcement
has
limited
impact
on
segregation
(stronger
policies
result
in
unstable
state)
V
=
value
if
at
least
1
volunteer C
=
individual
cost
of
volunteering A
=
overall
cost
if
no
volunteers N
=
number
of
people
involved p
=
probability
of
getting
at
least
1
volunteer 1-‐ p
=
probability
of
getting
no
volunteers V
– C
=
(pN-‐1)V
+
(1
– p) N-‐1(V
– A)
V
=
value
if
at
least
1
volunteer C
=
individual
cost
of
volunteering A
=
overall
cost
if
no
volunteers N
=
number
of
people
involved p
=
probability
of
getting
at
least
1
volunteer 1-‐ p
=
probability
of
getting
no
volunteers p =1− C A " # $ % & ' 1 N−1
Observations •Decrease
individual
cost
of
volunteering •Increase
the
overall
costs/impact
of
not
volunteering
(??) •Increase
difference
between
individual
cost
and
overall
costs •Reduce
number
of
‘players’ •Increasing
number
of
players
have
negative
or
no
impact