Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 329
9
State of the Art with Respect to Military
Needs
I
n this section we review the state of the art in individual, organizational,
and societal modeling against the military modeling needs outlined in
Chapter 2 and discuss the major shortfalls in meeting those needs. The
five representative problems described in Chapter 2 are used as a structure
to organize the review.
DISRuPT TERRORIST NETWORKS
One potential use of cultural and organizational models is to fuse par-
tial and uncertain information from multiple sources to develop a model of
the network structure of a terrorist organization and to use that model to
evaluate alternative strategies for disrupting that organization, for example,
through disconnecting leaders or interrupting recruiting. This goal can be
supported by many of the modeling approaches described earlier, and each
has its own limitations for attacking the problem. Table 9-1 summarizes the
capabilities that would provide advantages for each modeling approach and
the major limitations of the approach in addressing this problem.
Network models provide a promising approach for this problem, but
a general limitation across the network modeling approaches is the lack
of data for model development. Lack of data is the primary challenge
for using models to understand and disrupt terrorist networks. The data
availability problem is compounded by the classification levels for existing
data, the control of those classified data by multiple organizations, and
the inconsistencies in data structure and content. For example, the same
individual may be identified by different names in multiple databases. The
OCR for page 330
0 BEHAVIORAL MODELING AND SIMULATION
TABLE 9-1 Modeling Approaches and Limitations for Disrupting
Terrorist Networks
Problem 1: Disrupt a terrorist network.
How can we fuse uncertain and partial information from multiple sources to identify the
dynamic network structure of a terrorist organization? How can we then best disrupt this
network?
Advantages of Approach Major Limitations
Conceptual models indicate how removal Need ethnographic data on terrorists to see if
of a leader will affect the network. models apply in the specific culture.
Cognitive/affective models could predict Extensive time and effort are required to
reactions of members of network if leader develop specific models from open sources.
is removed.
Organizational models of existing network Data are not available. Existing data are in
could predict impact of changes in network multiple databases controlled by different
on performance. organizations, with inconsistent structures
and contents.
Link analysis could identify network. Link data are difficult to acquire.
Social network analysis models can predict
how changes in leadership will affect
network structure and change power and
centrality.
Dynamic network analysis models could Requires resource and activity data, which
predict who the emergent leader will be if are very difficult to acquire.
current leader is removed.
System dynamics models could predict High-level model would not predict for
whether change in leadership might lead to individuals.
an increase in violence in the community.
Model could be tested in massively Behaviors in the MMOG might not resemble
multiplayer online games (MMOGs). those in the real world.
All approaches could make No facility for rapid development of models
recommendations for action. or provision of easy-to-understand guidance
to war-fighters.
more specific the predictions that could be made by the model, the more
data are required. A way to mitigate this problem (see Chapter 11) would
be to make unclassified representative databases more widely available for
model development and evaluation.
There is an additional challenge in how to test and validate these
models and how to clearly communicate the uncertainty surrounding model
forecasts. If models could be developed more quickly and easily and in
closer collaboration with both subject matter experts and the ultimate
users, they could be a more useful thinking tool for decision makers.
OCR for page 331
STATE OF THE ART WITH RESPECT TO MILITARY NEEDS
FORECAST ADvERSARy RESPONSE TO COuRSES OF ACTION
Models could be used to forecast the responses of adversaries to friendly
force actions over a range of responses, with estimates of the likelihood of
each. This is especially needed in urban operations and operations other
than war in an asymmetric warfare environment in which conventional
methods for predicting adversary behavior are often not relevant. For
example, what are the likely reactions of both noncombatants and local
insurgents to friendly force movements, and can those reactions be affected
by the diffusion of information or disinformation? Table 9-2 summarizes the
contributions that could be made by models and their major limitations.
In the example of forecasting enemy response to a disinformation
campaign about troop movements, network models are clearly applicable
conceptually—the limitations are in the details that would make the model
useful for a specific cultural context. First, we lack theory regarding cultural
differences in the believability of a message as a function of its source. Multi-
agent models could predict information diffusion patterns, but they would
require data on transmission links (media channels, literacy rates, etc.) that
are specific to the location and culture being modeled as well as ways to
predict the believability of the message when received. Finally, there are
no well-defined outcome variables for assessing the validity of the model’s
TABLE 9-2 Modeling Approaches and Limitations for Forecasting
Enemy Response to Disinformation
Problem 2: Forecast adversary response to blue actions.
Predict the likely response of noncombatants and local insurgents to friendly force
movements, basing, logistics, and courses of action. Can disinformation be used to partially
protect our intentions? What is the most effective point of insertion of the disinformation?
Advantages of Approach Major Limitations
Persuasion theory suggests the need to No models for how individuals are likely
disconnect message from the source. to transform/distort messages as a result of
Research exists on direction of change in cultural and cognitive factors.
message with diffusion. There are no individual models to predict
interpretation or believability of messages,
especially taking cultural factors into account.
Use multiagent model of networks to Multiagent models would require models
decide where to drop information into of culture and technology to estimate speed
rumor mill (e.g., viral marketing). of transition. Need to link multiple types of
models together.
Use model to maximize predicted diffusion Straightforward conceptually but there is a
of information. lack of empirical country-specific data.
OCR for page 332
BEHAVIORAL MODELING AND SIMULATION
predictions. Clearly the uses to which the model is to be put, including the
details of the cultures and location of interest, must drive both its structure
and content. As in the previous example, the major recommended mitiga-
tion strategies (see Chapter 11) include the development and dissemination
of detailed datasets and the development of a close collaborative relation-
ship between model developers, subject matter experts, and model users.
SOCIETAL FORECASTINg
There is an urgent need for models that can forecast attitudes and
behaviors at a societal level as a function of alternative courses of action
particularly for diplomatic, information, military, and economic (DIME)
campaigns. For example, models are needed that can forecast the stability
of civilian governments and the incidence of violence as a function of these
DIME factors. Table 9-3 summarizes the possible contributions and limita-
tions of models for this goal.
Agent-based models (ABMs) on a large scale appear very promising
for modeling societal behaviors and forecasting responses to diverse DIME
courses of action (COAs). However, there are a number of limitations that
currently constrain what can be done with these models. First, such models
are time-consuming to build and often require data collection on a massive
level. Second, computational power limits the cognitive complexity that
can be built into individual agents if tens of thousands of agents are to be
included in the model. Third, predicting the response to integrated DIME
COAs requires multidisciplinary expertise in military planning, economics,
and political science. The need to integrate models across disciplines is the
primary challenge faced in this problem area. Theories and models in these
diverse disciplines are not currently integrated, and experts in these areas
often have little opportunity or incentive to collaborate. Hybrid federated
models that combine different models at different levels of detail for differ-
ent factors offer considerable promise for tackling large-scale societal pre-
diction, but these models are more an idea than a reality at present and will
require the multidisciplinary development of architectures and standards
for federation. To mitigate these limitations, Chapter 11 recommends an
extensive multidisciplinary research program focused around common chal-
lenge problems and datasets. Finally, as with other models, the predictions
of these models will have a high degree of uncertainty. Model development
must include collaboration between end users and diverse subject matter
experts to ensure that the predictions provided are relevant and that their
limitations are well understood.
OCR for page 333
STATE OF THE ART WITH RESPECT TO MILITARY NEEDS
TABLE 9-3 Modeling Approaches and Limitations for Societal Forecasting
Problem 3: Societal prediction.
Forecast the effects of alternative diplomatic, information, military, and economic (DIME)
courses of action on the attitudes and behaviors of residents in areas of interest.
Advantages of Approach Major Limitations
System dynamics models can predict the Sufficient theory is lacking to identify the key
effects of changes in DIME factors on variables to be included in the model and to
outcome variables of interest, such as level specify the connecting links between actions
of violence. and outcomes.
ABMs that capture resource use and ABMs need both theory and data to make
economic interactions as well as social useful predictions. Large-scale societal ABMs
links can predict the effects of economic are time-consuming and costly to develop.
factors within a social and cultural
context.
ABMs can include tens of thousands of Agents must be cognitively simplistic to make
agents to capture complex interactions at large-scale ABMs computationally feasible.
the societal level.
Historical data can be used to develop Prediction of the future, not the past, will
ABMs that predict societal effects. involve inherent uncertainty.
Federated models could integrate multiple There is a lack of infrastructure,
types of models at different levels of detail architectures, and standards for federated
to capture diverse DIME factors. models. DIME factors are studied by different
disciplines and few integrated models exist
that attempt to combine them. There is little
theory or data on how DIME factors interact.
MMOGs can provide an environment for The environment created by the MMOG may
data collection on a large scale to support not reproduce the key elements of a real-
model development and testing. world society.
CROWD CONTROL TRAININg
Virtual training environments offer an opportunity for troops involved
in peacekeeping operations to learn best practices for crowd control. Such
training will require models of noncombatants that respond to trainee
actions in a way that is realistic and appropriate for a specific location
and cultural environment. Table 9-4 summarizes the state of the art in this
area.
The development of models for crowd control training is perhaps the
most advanced of the five representative problems considered. This can be
done now. The only major issue is whether the models can produce behav-
ior that is close enough to that of a real crowd in a specific environment to
provide useful training. The extent to which cultural factors create differ-
ences in crowd behavior in different locations deserves further study, but
OCR for page 334
4 BEHAVIORAL MODELING AND SIMULATION
TABLE 9-4 Modeling Approaches and Limitations for Crowd Control
Training
Problem 4: Crowd control training.
Use models of crowd behavior to create a virtual training environment in which soldiers can
learn to take the appropriate action.
Advantages of Approach Major Limitations
Cultural models can provide theory on Theory linking attitudes and behaviors may
how crowd members in a specific culture not be specific enough for the environment
are likely to react to different actions. for which training is needed.
Cognitive and affective models can Individuals may react differently as part of
represent individual reactions to soldier a crowd than they would alone. The effects
behavior. of cultural context on behavior are not fully
understood.
ABMs can capture the interactions among Cultural variability in crowd dynamics is not
crowd members that cause them to act completely understood.
collectively in ways in which they might
not have acted individually.
MMOGs can provide an interactive Model behavior should be reviewed by
environment for quickly testing model subject matter experts for believability.
behavior as well as an environment for the
training.
crowd behavior models can be implemented in virtual environments, such
as MMOGs, and their behavior can be reviewed by subject matter experts
in a specific culture to ensure that they are not behaving in unrealistic ways
that would result in negative training.
ORgANIzATIONAL DESIgN: FORCE COMPOSITION AND
COMMAND AND CONTROL ARCHITECTuRE
Because of the rapid changes in mission requirements, the military
services are moving toward modular expeditionary forces that are readily
reconfigurable for different types of missions. Making the best use of these
modular forces requires not only a recommended force composition (sys-
tems, equipment, units, and personnel) but also a command and control
(C2) architecture that is most effective for the force as constituted. The best
force composition and C2 architecture for a conventional military opera-
tion may be quite different from that for a peacekeeping or disaster relief
mission. Table 9-5 summarizes how organizational models could help in
this process.
OCR for page 335
5
STATE OF THE ART WITH RESPECT TO MILITARY NEEDS
TABLE 9-5 Modeling Approaches and Limitations for Organizational
Design
Problem 5: Organizational design: Force composition and command and control architecture.
Use organizational models to develop optimal force composition packages and C2
architectures for different mission types.
Advantages of Approach Major Limitations
Use organizational models to develop the Requires detailed data on the tasks to be
force composition, structure, and processes performed in the mission and the resources
that are predicted to best meet mission available.
requirements. Simulate organizational
performance for different structures in
different mission scenarios.
Use simulation and ABMs to identify the Requires detailed data on task information
points in the mission and the organization and workload requirements.
at which the most intensive cooperation
will be required, the points of maximum
workload, and the potential information
bottlenecks.
Use MMOGs as a testbed for May be difficult to replicate realistic mission
organizational structures. tasks and conditions.
Models have proven to be a useful tool for understanding, designing,
and testing organizations. Although people rarely think about “designing”
organizations in a systematic way, the military faces the need to do just that
as it develops new flexible, adaptive structures for rapidly changing mis-
sions. One example of an attempt at systematic organizational design based
on modeling is the recent effort at the National Aeronautics and Space
Administration (Carroll, Gormley, Bilardo, Burton, and Woodman, 2006),
using an ABM and a heuristic rule-based model. Modeling and simulation
can be used to develop and adapt force composition for changing mis-
sions and to suggest the best C2 structure for accomplishing the mission.
The best C2 architecture is especially challenging for coalition operations,
in which different types of forces may be involved, and for peacekeeping
and disaster relief operations, which require close coordination with non-
government organizations. The major limitation for this work is the need
for detailed information on the tasks to be accomplished in the mission and
the resources required to accomplish those tasks. The recommended miti-
gation strategy (see Chapter 11) is the development of common challenge
problems and datasets and the use of collaborative workshops to ensure
that that operational users and modelers have a shared understanding of
what can be done through modeling.
OCR for page 336
BEHAVIORAL MODELING AND SIMULATION
REFERENCE
Carroll, T.N., Gormley, T.J., Bilardo, V.J., Burton, R.M., and Woodman, K.L. (2006). Design-
ing a new organization at NASA: An organization design process using simulation.
Organization Science, , 202–214.
