Agent-Based Modeling Explores
Challenges to Theory, Science
in agent-based model formulation revolved around
social networks and geographic dimensionality,
with additional complexity arising from the over-
lay of social networks, networks of institutions,
and equation models hybridized with agent-based
modeling. Problems cast in terms of traffic and epi-
demics were further elaborated on by inclusion of
agents who think for themselves. The subject matter
ranged from villages of 200 to metropolitan areas
of 18 million. The chief commonalities included
the need for criteria to establish credibility, practical
ways to validate simulations, and astutely exploited
mathematical underpinnings.
Early on, it became clear that the use of
agent-based modeling exists along a continuum.
Exploratory models are at one end, with goals
related to examining the consequences of specific
theories or posited mechanisms for agent interac-
tions. At the other end, there are predictive models
that extrapolate the consequences of agent behavior
from a collection of specifications about the agents
and their abilities and modes of interaction. So,
Attendees from several disciplines converge to discuss agent-based modeling at some applications developed hypothetical scenarios
the recent NIss explorations Workshop.
based on assumptions and the alterations that might
follow specific interventions (that induce changes in
the model specification). Other applications investi-
gated differences between model-based (predictive)
patterns that emerged from an agent-based simula-
T
he NISS-sponsored Agent-Based Modeling:
tion with the patterns observed in reality to conjec-
Commonalities in Unconnected Problems,
ture about the mechanisms that lead to the reality.
second in the NISS series of Explorations
So what makes agent-based modeling credible
Workshops, opened with a “gaps in the state-of-the-
as a research tool? Obviously, answers must differ
art” address by Ben Klemens of the Brookings
for different research intentions. What are the infer-
Institute and proceeded to pose and illustrate sub-
ences to be drawn? An eventual equilibrium (stable
stantive and theoretical questions. Before the work-
function with estimable parameters)? Summary
shop ended, the new conjectures raised covered the
statistics of the process, itself? Mimicry of real
gamut from philosophy to the sociology of chickens
event(s) at a micro level? Statement, perhaps prob-
and from physicists’ use of Laplacians to social sci-
ability distribution-based, of the potential variation
entists’ use of GIS. By any measure of interdisci-
of outcomes? Comparative evaluation of different
plinary engagement, the workshop was an excep-
specifications that instantiate different theories?
tional success, and wide-ranging avenues for
Illumination of emergent behavior that is difficult
statistical advances were uncovered.
to extract from ‘noise’? Projections for hypothetical
The agenda included perspectives on agent-based
(unobservable) scenarios?
modeling from applied mathematics, military appli-
Similarly, validation of a particular agent-based
cations, social sciences, and statistics. Complications
model does not have a simple definition. When real
data are available, they can provide one component
6 AMsTAT NeWs JANUARy 2009
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80