Conference PaperPDF Available

Information-Velocity Metric for the Flow of Information through an Organization: Application to Decision Support

Authors:

Abstract and Figures

The rapid flow of information through a command-and-control organization strengthens the agility of the command and increases the efficiency of the organization. Therefore, a metric is proposed for the speed of useful information as it flows in an organization, enterprise or command structure. Information is assumed to be useful if a member of the organization considers it to have the potential to reduce uncertainty, and thus become an important contribution to a decision process. Information velocity depends on information flow and time, which is sometimes described as "getting the right information to the right person at the right time." Efficient information flow is the goal of net-centric warfare and is needed for making faster, better decisions.
Content may be subject to copyright.
Cover sheet for the 14P
th
P ICCRTS
“C2 and Agility” 15-17 June 2009, Washington, D.C.
Paper No. 016
Title:
Information-Velocity Metric for the Flow of Information
through an Organization: Application to Decision Support
Track 8 topic: C2 Assessment Tools and Metrics
Authors: Jeff Waters, J.D., Ritesh Patel, B.S, James Eitelberg, M.S.,
Gunnar Ramstrum, M.S., and Marion Ceruti, Ph.D.
Organization: Space and Naval Warfare Systems Center Pacific (SSC Pacific)
Address: 53560 Hull Street, San Diego, CA 92152-5001, USA, (619) 553-4068
Email: HTUjeff.waters@navy.milUTH, , HTUritesh.patel@jpmis.milUTH, HTUjames.eitelberg@navy.milUTHHTU,
UTHHTUgunnarramstrum@yahoo.comUTH, Umarion.ceruti@navy.mil
Marion Ceruti, Ph.D., Point of Contact
(619) 553 4068
Filename: Waters.etal.PaperID_016_InfoVelMet_ver38.CR.doc
Report Documentation Page
Form Approved
OMB No. 0704-0188
Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington
VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it
does not display a currently valid OMB control number.
1. REPORT DATE
JUN 2009
2. REPORT TYPE
3. DATES COVERED
00-00-2009 to 00-00-2009
4. TITLE AND SUBTITLE
Information-Velocity Metric for the Flow of Information through an
Organization: Application to Decision Support
5a. CONTRACT NUMBER
5b. GRANT NUMBER
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) 5d. PROJECT NUMBER
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)
Space and Naval Warfare Systems Center Pacific (SSC Pacific),53560
Hull Street,San Diego,CA,92152-5001
8. PERFORMING ORGANIZATION
REPORT NUMBER
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)
11. SPONSOR/MONITOR’S REPORT
NUMBER(S)
12. DISTRIBUTION/AVAILABILITY STATEMENT
Approved for public release; distribution unlimited
13. SUPPLEMENTARY NOTES
In Proceedings of the 14th International Command and Control Research and Technology Symposium
(ICCRTS) was held Jun 15-17, 2009, in Washington, DC
14. ABSTRACT
see report
15. SUBJECT TERMS
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF
ABSTRACT
Same as
Report (SAR)
18. NUMBER
OF PAGES
45
19a. NAME OF
RESPONSIBLE PERSON
a. REPORT
unclassified
b. ABSTRACT
unclassified
c. THIS PAGE
unclassified
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std Z39-18
Information-Velocity Metric for the Flow of Information
through an Organization: Application to Decision Support
Jeff Waters, J.D., Ritesh Patel, B.S, James Eitelberg, M.S.,
Gunnar Ramstrum, M.S., and Marion Ceruti, Ph.D.
Space and Naval Warfare Systems Center Pacific (SSC Pacific)
53560 Hull Street, San Diego, CA 92152-5001, USA, (619) 553-4068
HTUjeff.waters@navy.milUTH, , HTUritesh.patel@jpmis.milUTH, HTUjames.eitelberg@navy.milUTHHTU,
UTHHTUgunnarramstrum@yahoo.comUTH, Umarion.ceruti@navy.mil
Abstract: - The rapid flow of information through a command-and-control organization strengthens
the agility of the command and increases the efficiency of the organization. Therefore, a metric is
proposed for the speed of useful information as it flows in an organization, enterprise or command
structure. Information is assumed to be useful if a member of the organization considers it to have
the potential to reduce uncertainty, and thus become an important contribution to a decision proc-
ess. Information velocity depends on information flow and time, which is sometimes described as
“getting the right information to the right person at the right time.” Efficient information flow is the
goal of net-centric warfare and is needed for making faster, better decisions. Everyone is a decision
maker on some level. Modern organizations demand optimal agility, creativity, and innovation.
Various factors, techniques, and tools can help improve information velocity. Given the state of
technology at the time of this writing, a direct measure of information flow in command centers
does not appear to be possible in the near future. Therefore, the present study focuses on the timing
aspect of information velocity, which deals with time management in organizations in general, and
in command centers in particular. This paper describes an individual decision-making model that
attempts to fill this gap. It proposes formulas for measures, suggests a small, simplified set of ques-
tions to obtain a quick metric, and describes future research. The authors welcome comments and
suggestions and would like to collaborate with others on this effort. This work supports the GIS3T
interoperability test bed, which you are welcome to join.
Keywords: - Agility, assessment tools and metrics, C
P
2
P concepts, CP
2
P theory, decision model, decision
support, employee empowerment, entropy, industrial computer applications, infodynamics, infor-
mation sharing, information theory, time management, uncertainty.
4B1. Introduction
Information Velocity (v(info)) is the first time
derivative of the information flow. What is an
Information Velocity Metric (IVM)? The defini-
tion for the word “metric” is “art, process, sci-
ence or system of measurement.” “Velocity” is
the speed and direction of motion of a moving
body. “Information” has been defined as “facts,
data;” “knowledge obtained from investigation,
study or instruction;” and “something, such as a
message, experimental data or a picture, that jus-
tifies change in a construct, such as a plan or a
theory, that represents physical or mental experi-
ence of another construct” [1]. The last definition
is the one most relevant to the flow of informa-
tion because it relates to how new information
can change existing constructs by reducing un-
certainty regarding some aspect of these con-
structs. IVM is an attempt to measure the speed
and direction of information as it moves through
an organization.
Information velocity has two significant aspects
– information flow and time as illustrated in Fig-
ure 1. The study of information velocity is fo-
cused on factors that can reduce uncertainty in
the fastest manner. These aspects are important
in many different types of organizations but es-
pecially for command centers.
Figure 1. Taxonomy of Information Velocity
Ideally, information will move quickly and effi-
ciently through the organization - not just up the
organization’s chain of command, but also down
and across echelons in the organization. Today's
organizations, especially those focused on in-
formation processing, demand a free flow of in-
formation. The global war on terror demands a
focus on the need to share information across
federal, state, county, city, local, tribal and non-
governmental organizations and coalitions. The
IVM is intended to measure the speed of infor-
mation flow within an organization or enterprise
with a view toward improving the speed and
quality of making decisions thereby reducing
uncertainty. As is often said in the Department of
Defense (DoD), the IVM is intended to measure
the speed with which the “right information is
getting to the right person at the right time.”
This paper covers the following topics:
Information flow and uncertainty,
A derived expression for v(info) that relates
it to power and task tractability.
Approximate, causal, and effects-based in-
formation metrics,,
Advantages and disadvantages of the differ-
ent metrics,
Theory underlying the inclusion of the dif-
ferent metric components,
Proposed metrics based on theory, a review
of applicable literature, and an assessment
of cumulative experience in development of
situational-awareness tools in selected
command centers.
Future research to explore the efficacy of
the measures in a modeling-and-simulation
(M&S) environment, and explore the meas-
ures in an operational, organizational set-
ting.
2. Developing a Good Metric
The study of information could benefit from an
examination of the context in which a good met-
ric would be used. The definition of a good and
suitable metric is the first step in metric devel-
Magn
i
tude
Information Velocity
Direction:
Information should
flow in the right
direction to the right
person or network
site where the
information is
required for tasking.
Quantity: How
much information
can move?
Time management: How fast
can information move?
*
Magnitude
Direction:
Information should
flow in the right
direction to the right
person or network
site where the
information is
required for tasking.
Quantity: How
much information
can move?
Time management: How fast
can information move?
Information Flow:
Defined by entropy and
uncertainty reduction.
No direct, general metric
or calculation available.
Modeling & Simulation*
Ba
y
esian Networks*
Indirect methods:
Current paper
Future
e-
Current paper
Future work
Assumptions
App
roximations
*
V
elocit
y
opment. Ideally, a good and suitable metric is
one that is:
Measuring the right quantity,
Important and significant,
Relevant and useful,
Well defined and understandable,
Accessible and achievable,
Reliable and predictive,
Robust and comprehensive,
Sensitive,
Simple and easy to use,
Timely and cost effective,
Efficient to compute, and
Helpful to improve processes.
In contrast, a bad metric is one that is:
Measuring the wrong quantity,
Ill defined and incomprehensible,
Insignificant,
Too complex and hard to use,
Time consuming to compute,
Subject to errors and misinterpretation,
Ineffective and insensitive,
Misleading, and useless.
A good metric should yield the same value for
the same measurement data to within the toler-
ance or experimental error of the measuring de-
vice(s). Irrelevant modifications in the data
should not perturb it significantly. A good metric
is one that will help define and facilitate an un-
derstanding of the measured quantity. As ex-
pressed in quotations [2] attributed to Lord Kel-
vin, “To measure is to know.”
A metric can provide an absolute measurement
or a relative measurement. For example, an
automotive gas gauge could indicate an absolute
measurement, e.g. 8.7 gallons, or could provide a
relative measurement, e.g. half full. Either for-
mat may suffice, but one may be better depend-
ing on the convention, the expertise of the users
or the purpose of the measure.
The second step in metric development is to de-
termine how well a quantity is defined and how
well its processes are understood. This is impor-
tant because it can affect the type of metric that
one can employ, e.g. relative or absolute meas-
urements. Consider three types of metrics, e.g.
those that employ:
A direct measure,
Causal measures,
Effects measures.
In the first case, if a quantity is well defined and
accessible to measurement, direct measures are
appropriate. Direct measures are especially im-
portant when:
They are not too costly,
They are not too intrusive,
The causes and effects of the measured
process are well understood.
Indirect measures, such as causal measures, are
useful and ap
propriate when:
They can be correlated to the process,
They help to define the process,
Direct measures are not easily em-
ployed.
Causal measures are especially helpful for ex-
ploring our understanding of the underlying
mechanisms of a process. Causal measures also
help to explain how to improve the process. Ef-
fects measures offer the advantage of helping to
achieve the desired effect. Yet the effects meas-
ures often assume a correlation between the
process and the effect, which is an assumption
that direct measures avoid. Considering the ap-
plicability of the three different types of meas-
ures (direct, causal, and effects) helps guide the
metric-design process and helps ensure that a
good metric will result
3. Characteristics of Agile, Creative and
Sustainable Decision Making
Making agile, creative, and sustainable decisions
in a command-and-control environment depends
on conditions that differ significantly from mak-
ing decisions in a standardized and relatively
static environment where sub-optimal decisions
do not lead to life-and-death consequences.
For example, consider the stereotype of a tradi-
tional manufacturing assembly line in an envi-
ronment where well-defined processes change
slowly. Workers make decisions according to
fixed rules that rarely change. Such an environ-
ment is characterized by the utmost standardiza-
tion and optimization where no one expects the
average worker to consider the big picture. Here,
the goal is not for workers to “think outside the
box,” or worry about other people's tasks. The
assembly-line worker must focus on limited,
fixed, and standardized decisions that are impor-
tant, but have a smaller scope. Here, the need for
agility and creativity is partitioned and compart-
mentalized for each worker. In the DoD some-
thing equivalent to this environment occurred
during the cold war. The enemy was static and
the emphasis on symmetric warfare encouraged
traditional thinking.
The stereotype described above stands in sharp
contrast with the demands for agility and innova-
tion in the current global business and military
environment. For example, the DoD and other
departments of the executive branch, such as the
Department of Homeland Security (DHS), are
focused on fighting the asymmetric, global war
on terror. In this environment, the conditions are
not as fixed and the processes not as standardized
as they were during the cold war. Here, the need
for honed optimization of narrow tasks is less
possible and less desirable because everyone
needs to be engaged in observing the environ-
ment. Agility and creativity on the part of all
employees is essential because good ideas and
innovation are not limited to a few experts. Suc-
cessful, agile decisions are best accomplished by
rallying all resources to the cause. (For additional
insights into the decision-making process, see,
for example, [3].)
The following observations pertain to an agile,
dynamic decision-making environment.
In the most successful modern organiza-
tions, people are empowered and encour-
aged to think independently, to develop
new ideas, and to share information that
sparks development. In this paradigm,
employees assess and reduce uncertainty
by passing useful information.
At some level, every employee is a deci-
sion maker. Roles are broader and have
more authority.
People typically make decisions that are
based on the decisions of others and often
not on raw data especially in organiza-
tions that process information. Thus, op-
timizing the individual decision-making
process also optimizes overall information
flow. For example, decisions of people in
support roles constitute the input informa-
tion for most high-level decision makers.
“Useful” information from an organiza-
tional perspective is any information that
an employee considers important to re-
duce uncertainty. Thus, a general defini-
tion of information usefulness should be
based on the employee’s assessment of
uncertainty before and after the informa-
tion is received.
A proposed standard for representing decisions, a
Common Decision Exchange Protocol, is defined
and recommended for this purpose in [4].
4. Decision-Making Process Model
This section describes a model focused on the
time-management component of information
velocity (Figure 1.) Significant research has de-
fined the basic decision-making process in the
general form, commonly called the Observe–
Orient–Decide–Act (OODA) loop. In this study,
this decision-making model has been revised
only slightly to capture the process from the per-
spective of information flow.
The decision-making process model is depicted
in Figure 2, where states of information man-
agement are not the same as the states of infor-
mation as described in [5], but rather, the states
of information management parallel the states of
information because information is aggregated at
progressively higher levels from one state to the
next. Data aggregation, data integration, and data
fusion in support of command decisions are
aimed at the rapid reduction of uncertainty by
minimizing the alternative decisions. In Figure 2,
this process of reducing the search domain (and
the entropy) can occur anywhere, but mainly oc-
curs in steps 1, 2, and 3 (See, for example, [6].)
The decision-making process depicted in Figure
2 assumes that some states can be skipped or that
very little (or no) time is spent in a given state.
Figure 2. Information-Flow Model
The model assumes that decision makers use
some variant of the following procedure.
Gather information needed to make a
decision.
Group information according to the al-
ternative courses of action it supports.
Evaluates and prioritize alternatives.
Make the decision.
Prepare a decision product, which could
be a PowerPoint slide, a report, an email,
a meeting, a phone call, or some other
means to communicate the decision.
Communicate the decision to others in
the organization,
Receive feedback from others regarding
the outcome of the decision.
Either re-enter the evaluation state or
finish the decisions process. (In Figure 2,
the start and end states are not shown.)
With this model, one can consider the impact
on a decision maker who gets the right informa-
tion at the right time. The most immediate and
obvious benefit of an improved time-management
is that the decision maker can spend less time in
State #1, gathering information. This is particu-
larly true if one assumes that the decision maker
has support personnel in State #1 who gather,
process, and present information to the decision
maker in an efficient manner. However, a person
who is awaiting the decision also is a decision
maker who needs the decision as input. From this
perspective, time spent preparing a decision prod-
uct (State #4) and communicating the decision
(State #5) represents a delay in the information
flow because the decision already has been made
in State #3, and other decision makers down-
stream are still in a state of uncertainty.
Given the assumption described above, the model
suggests that the impact on a decision maker from
increased information velocity is that the decision
maker can spend less time in States #1, #4 and #5,
the states colored red in the diagram, relative to
States #2, #3, and #6 in the diagram.
Unless considerable data fusion has occurred in
State #1, the decision maker will spend a
greater proportion of the decision-making time
in green states. The decision maker can move
through the decision cycle faster and/or with a
higher quality output. In this case, the decision
maker and the organization overall can handle
an increased number of decisions or workload.
Speed and quality are important in making de-
cisions. High quality decisions that are based
on subordinate decisions as input depend heavily
on honest, unbiased, unfiltered assessments. Simi-
larly, improvements in useful, high-quality infor-
mation flow depend on a policy that fosters hon-
est, unbiased, unfiltered assessments. The states
that account for this part of the process need to be
included in the decision-making state model to
ensure that it is included in the metric. Figure 3
depicts an expansion of State #3 (Make Deci-
sion/Assessment) from Figure 2.
Figure 3. Information-flow model expanded decision substates
In Figure 3, the following subprocess is de-
scribed in terms of substates of State #3 from
Figure 2. The decision maker might use the fol-
lowing procedure (in this order).
1. Assess the alternatives according to one’s
own personal objective assessment based
on education and training (Substate #1).
2. Perform subjective assessment based on
personality and experience (Substate #2).
3. Make an internal, unfiltered, and externally
unbiased decision (Substate #3). Ideally,
the decision maker would exit these sub-
states with a direct, honest, unbiased deci-
sion; however, most decision makers will
continue to the next substate.
4. Weigh the above assessment against the
possible acceptability of this decision with
the decision-maker's supervisor, col-
leagues, and upper management. Consider
the answer that they want, the easiest an-
swer, and the answer that poses the least
extra work or political risk (Substate #4).
5. Make an “external” decision filtered by ex-
ternal influences less honest, more biased,
and likely to reflect the desired and ex-
pected answer without any independent
decision process. At this point, a decision-
maker exits the substates with both the
honest, direct internal assessment and the
filtered external answer. Depending on the
setting and the employee's organizational
bravery, others will receive a recommen-
dation that includes either one or both de-
cisions.
What does improved time management mean in
this context? In Figure 3, substates #1, #2, and #3
are colored green to represent states that are de-
sired to derive useful information. Substates #4
and #5 are colored red because they reflect a
“varnished” or influenced opinion that results in
less “useful” information from an organizational
perspective. As other decision makers down-
stream in the information flow await upstream
decisions, time spent tailoring a decision for out-
side influences delays the process and results in
an opinion less honest and more biased. Thus,
improved time management enables decision
makers to spend less time in Substates #4 and #5
as a proportion of the total time spent in State #3.
5. Elements of a Course of Action
More often than not, a decision will result in
some type of Course of Action (COA). This sec-
tion describes the components of a COA.. Gen-
erally, decisions pertain to future COAs based on
current information, current options and pro-
jected trends. Components of a COA also are
components of a decision process because to ar-
rive at a decision requires a procedure, or COA.
For the sake of argument, we consider the ele-
ments in a COA in general, but the same ele-
ments apply equally to a decision. Elements of a
COA or a decision could include the following:
What to do? Actions that must be per-
formed as a result of the decision usually
are conditioned on the receipt of data.
COAs generally consist of many sub-tasks,
some of which are sequential and some can
or must be performed in parallel. Some
COAs have critical actions that determine
the outcome of the COA. If one or more
critical actions cannot be performed, it may
be necessary to select an alternate COA.
Where to start? This could be a collection of
geographic locations or it could be what
document to examine to begin analysis.
Where to end? The decision maker may not
be able to specify exactly where to end be-
cause by the time the COA is complete,
many modifications may have been made in
the COA that could affect the end location.
When to start taking action? This could be
unconditional or conditioned on circum-
stances that may be unknown when the
COA is formed. When to start could be a
time, such as a date-time group. If the COA
has many subtasks, when to start may be
different for different subtasks. When to
begin one subtask may depend on the out-
come of an earlier subtask.
When to end the COA? This could be a
specified time, such as a date-time group,
but usually includes exit criteria that depend
on when each subtask is complete. The
COA cannot end until each subtask is com-
plete or not necessary to execute. When to
end likely is conditioned on the success of
the subtasks and how well they proceed.
In what direction will information regarding
the COA proceed? This relates to the part of
information velocity that pertains to direc-
tion of information flow. Who should know
about the outcome of tasks and subtasks in a
COA will determine who gets these reports.
The commander and subordinates alike will
need to know this.
Who will perform the COA? The larger the
group involved, the more complex the esti-
mation of the conditional entropy because
each group member involved in the COA
could have a different level of uncertainty
associated with specific tasks that pertain to
that individual.
6. Uncertainty and Information Flow
Previous work aimed ultimately at reducing un-
certainty in command centers has focused on
reducing search domains in database integration
[6] and the delivery of available, relevant, timely,
and needed (and therefore, useful,) information
from the battle space to a user [7]. The reduction
of a search domain in a database integration
problem reduces uncertainty by limiting the
search for the solution to the problem where it is
most likely to be found [6]. Similarly, an im-
provement in the availability of information that
can help solve a problem related to a user’s task-
ing increases information flow to the extent that
it reduces the user’s uncertainty about the solu-
tion of the problem under consideration. In any
case, users at all levels must make decisions. For
example, an engineer recommends using a num-
ber that was the solution of an equation, or a
commander orders a COA in battle.
In this section we derive an equation for v(info)
to show how it is related to power and task trac-
tability, T
R
Y
R, by invoking simple physics and in-
fodynamics [5]. Specifically, information flow
depends on the values of variable h, (which re-
lates to the decision or solution) and variable l,
(which relates to the available data) in a given
process. Information flow, p, is defined in equa-
tion (1) as the difference between the uncertainty
before the process started and the uncertainty
after the process is finished [8].
(1) p = H(h|l
R
1
R) - H(h|lR
2
R)
In equation (1), H(h|l
R
1
R) is the conditional entropy
of variable h before the process started given the
variable l
R
1
R, also before the process started. In
contrast, H(h|l
R
2
R) is the conditional entropy of h
before the process started given the variable l
R
2
R,R
Rwhich is the value of l after the process is fin-
ished. Conditional entropy can be conceptualized
as the number of alternatives that are viable
given the information at hand.
For example, as depicted in Figure 2, informa-
tion is gathered in state 1. Its transfer to a deci-
sion maker constitutes the transition to state 2.
Here, H(h|l
R
1
R) applies to the decision maker be-
fore new information is received. State #2 is
where the transition from l
R
1
R to lR
2
R takes place as
information is evaluated for its content and abil-
ity to reduce uncertainty. H(h|l
R
2
R), is related to the
alternatives available to the decision maker after
the gathered information is evaluated in State #2.
Information flows to the decision maker to the
extent that the uncertainty that previously existed
becomes reduced as a result of receiving the new
information. When State #3 is entered, the effect
of the information flow already has taken place
and the reduced uncertainty facilitates whatever
decision is made.
To understand equation (1) in a command-and-
control context, let the variable, h, be the COA,
and variable l, the information from the battle-
space that supports the selection of a particular
COA. Conditional entropy, H(h|l
R
1
R), which de-
scribes this state, could be the number of COAs
consistent with the information initially avail-
able. Thus, variable l changes from l
R
1
R toR
RlR
2
Rwhen
useful information is received in the command
center. The conditional entropy also changes to
H(h|l
R
2
R) after this information is instrumental in
reducing the uncertainty. Thus, H(h|l
R
2
R) will be
smaller than H(h|l
R
1
R) when the new information,
l
R
2
R, reduces the number of alternative COAs be-
cause COAs that are inconsistent with the new
information, l
R
2
R, will be ruled out.
Another way to conceptualize this process is as
follows. Before the receipt of useful information,
l
R
2
R, many COAs, h, are available. Entropy H(h|lR
1
R)
is high and the uncertainty about what to do,
therefore, is high. Information flow, p, increases
in a command center when it results in a reduc-
tion of the number of viable COAs. This is be-
cause l
R
2
R forces the elimination from considera-
tion of COAs that will not work in light of the
new information. This reduced selection in
COAs is quantified by a reduction in the condi-
tional entropy. Thus, what to do becomes clearer
as a result of information flow.
The above example illustrates equation (1) using
COA as variable h. However, other variables
commonly measured in the battle space also
could be selected to illustrate equation (1). One
such example can be defined as, h’, the position
of a particular hostile platform. In this case, any
information, l
R
2
R that sheds light on that position
will reduce the uncertainty in variable h’. This is
particularly easy to conceptualize and compute in
terms of the ellipse of uncertainty typically
drawn around a position measured by the trian-
gulation of sensor lines of bearing. (See, for ex-
ample, [7].) Initially, l
R
1
R is the information avail-
able in the command center about the hostile
platform, which could be anywhere in the battle
space if no one knows anything about it.
Thus H(h|l
R
1
R) is high. After the receipt of infor-
mation, l
R
2
R,R
,
Rthe hostile platform’s position is con-
fined to the ellipse of uncertainty associated with
the measurement of its position. Now H(h|l
R
2
R) is
reduced considerably compared to H(h|lR
1
R). The
value of p can be calculated in this case in a
much more definitive way than in the case where
h represents COA, which can depend on the in-
tegration and fusion of many different types of
data, including but not limited to the positions of
hostile forces.
Knowing the position of a hostile vessel may
reduce the uncertainty of a COA because it may
cause any COAs that are inconsistent with the
vessel’s position to be ruled out. The actual
overall uncertainty in COA selection depends on
the uncertainty in many variables and their inter-
actions in the battle space, including but not lim-
ited to the positions, capabilities, and strengths of
hostile and friendly forces; various sources of
uncertainty associated with the measurement of
data from sensor networks; weather observations
and forecasts; sea states; time of day; the order in
which messages are received and processed; the
availability of logistical support; and the com-
mander’s interpretation of the culture to doctrine.
By definition, v(info) is the time derivative of the
information flow, p. d stands for a small change.
(2) v (info) = dp / d t
= d H(h|l
R
1
R) / d t – d H(h|lR
2
R) /
dt
Note that v(info) is like acceleration because it
involves a change in time derivatives of entropy.
Thus, v(info) has the units of entropy, H, over
time. In thermodynamics, entropy has the units
of energy, heat (U), or work (W) over tempera-
ture (T). Equation (3) defines the thermodynamic
entropy, dS, in terms of the change in the re-
versible heat, U
R
rev
R over T [9].
(3) dS = dU
R
rev
R / T
The infodynamic [5] analog of entropy, H(h|l),
also has the units of entropy over a temperature-
like entity, T
R
Y
R, but in this case, the TR
Y
R refers to
the degree of tractability [5]. A task is highly
tractable if the completion of a task requires sim-
ple, logical, straightforward combinations of in-
dependent data. The dependence of one data
element on another tends to make tasks less trac-
table because more complex data-fusion algo-
rithms are involved in task completion. A task
will be highly tractable when entropy and uncer-
tainty are low and the number of alternative deci-
sions is relatively low. Another way to conceptu-
alize T
R
Y
R is expressed in equation (4), where in
most cases, T
R
Y
R will be proportional to the recip-
rocal of the sum over algorithms of the number
of bits in each algorithm that is necessary to
process and fuse raw data. (At least equation (4)
should be a good estimate of the relative com-
plexity of the data-fusion processes.)
(4) T
R
Y
R = C / n R
bits
C is a constant that depends on the type of task.
T
R
Y
Ralso could be expressed as the reciprocal of
the sum of the variables and their interactions in
a manner analogous to the Virial expansion in
gases. (See, for example, [9] and [5].) Here, we
assume that T
R
Y
Rdoes not depend directly on time.
The infodynamic analog of (3) is equation (5).
(5) d H(h|l) = d W / T
R
Y
In information management, especially as it ap-
plies to organizations, T
R
Y
R is like T because when
H(h|l) is high, T
R
Y
R is low. The tractability of a
decision task decreases in inverse proportion to
the number of possible alternative solutions i.e.
the uncertainty. A high information flow, p, re-
duces the uncertainty and reduces the amount of
work necessary to make a decision. Highly in-
tractable tasks require a larger information flow
to solve them. As p increases, the tractability,
T
R
Y
R, of the decision task increases. TR
Y
R, actually is
not a constant and is conditioned on the data and
their fusion algorithms
Newton’s law defines physical force below.
(6) F = J d
P
2
PX/ dtP
2
where J is mass, X is distance, and t is time. J
represents only the useful information that is
relevant to a particular task. Only this useful in-
formation will be instrumental in reducing uncer-
tainty. The second time derivative of the dis-
tance, X, is d
P
2
PX/dt, acceleration. Work, W, is
given by equation (7).
(7) W = F X
(8) d W = F dX
The concepts expressed in equations (6) through
(8) can be adapted to apply to information flow.
Combining equations (2) through (8) yields
(9) d H(h|l) = (J/T
R
Y
R) (d P
2
PX/ dtP
2
P) dX
Dividing equation (9) by dt, as in equation (2),
and proceeding to the limit yields equation (10).
(10) dH(h|l) / dt = (J / T
R
Y
R) (d P
2
PX/dtP
2
P) dX/dt
Now set X
R
d
R= dX/dt and substitute.
(11) (d
P
2
PX/dtP
2
P) (dX/dt) = ½ d/dt (dX/dt) P
2
= ½ d/dXR
d
R (XR
d
R)P
2
P(dXR
d
R/dt)
Combining equations (2), (10), and (11) yields
equation (12).
(12) v(info) = d/dt
{ ( ½ J XR
d
RP
2
P)R
1
R/ TR
Y1
R
- ( ½ J X
R
d
RP
2
P)R
2
R/ TR
Y2
R }
Where the (½ J XR
d
RP
2
P) terms are energies. As is
evident from equation (12), v(info) has the units
of power per degree of tractability. The power
term, G , is given by equation (13).
(13) G
= d/dt (½ J XR
d
RP
2
P)R
Thus, an increase in either v(info) or TR
Y
R, results
in more power, G. Thus, the aphorism, “informa-
tion is power” is almost accurate. Actually in-
formation velocity is proportional to the power,
as is evident in equations (12) and (13). This is
especially true in a command-and-control con-
text, where the flow of information, or the lack
thereof, influences power projection and could
make the difference between victory and defeat.
The velocity at which information flows in an
organization will determine who and how many
have power, how much power they have, when
they have it, and for how long. This reflects the
fact that information is important for individuals
as well as for an organization as a whole. How-
ever, this power will be important and can be
used only if the information is received in a
timely manner.
Equation (12), suggests the analogy between in-
formation dynamics and fluid dynamics. X
P
2
P is
like the cross sectional area of the virtual “pipe”
through which the information flows. J is the
useful information moving through that “pipe.”
Equations (12) and (13) are best tested in M&S
experiments where variables are well defined and
uncertainties can be calculated and controlled.
7. Assumptions and Approximations Re-
garding Information Flow
M&S not withstanding, exact, general, and direct
metrics for overall information flow, p, and
hence v(info) in a command center have yet to be
developed and are not likely to be obtained in the
foreseeable future using current technology.
Therefore, additional assumptions are necessary.
For example, to simplify equations (1) and (2),
assume that H(h|l
R
1
R) is large, unknown, intracta-
bly complex, and unknowable before the deci-
sion deadline. (N.B. It is not necessary to assume
that H(h|l
R
1
R) remains constant, just large com-
pared to H(h|l
R
2
R). In this case, the task reduces to
minimizing H(h|l
R
2
R)).
Actually, l
R
1
R, and lR
2
R each represents the informa-
tion associated with a data set or collection of n
variables that consists of the potentially relevant
data elements, l
R
i
RthroughR
RlR
i+n
R and, ideally, their
pedigree metadata m
R
i
R [10]. Each metadata ele-
ment, m
R
i
R, describes or summarizes the metadata
associated with a particular data element. Each
metadata element, m
R
i
R, which can be conceptual-
ized as a weighting factor, expresses the uncer-
tainty of l
R
i
R. Metadata element, mR
i
R, also can in-
clude an estimate of the relevance of each l
R
i
R to
the scenario associated with the decision task, as
part of the metadata.
H(h|l
R
2
R) is a multivariate function, the order of
which is the number of data elements in l
R
2
R. To
find the minimum of H(h|l
R
2
R), the gradient, a vec-
tor of partial first derivatives, is set to zero as in
equation (14). When minimizing H(h|l
R
2
R), the mR
i
R
would be treated as constants.
(14) H(h|l
R
i
R) = 
Moreover, to preclude maxima or saddle points,
the partial second derivatives of H(h|l
R
2
R) must
form a positive definite matrix [11], which re-
duces to inequality (15) in the case of a single
variable, l
R
i
R.
(15)
P
2
PH(h|lR
i
R) > 
Algorithms were developed to determine minima
of multivariate functions [12], [13], but their ap-
plication is beyond the scope of this paper.
Two- three- and…n-way data fusion contributes
to data set, l
R
2
R,R
Ras is described in equation (16).
(16) l
R
2
R = { {lR
i 2
R mR
i 2
R}, { lR
a2
R mR
a2
R, lR
b 2
R mR
b2
R},
{ l
R
a2
R mR
a2
R, lR
b 2
R mR
b2
R, lR
c 2
R mR
c2
R},…
{ l
R
a2
R mR
a2
RlR
n 2
R mR
n2
R}}
For example, the first set in equation (16), {lR
i2
R
m
R
i2
R}, represents the contribution of each single
information source multiplied by its metadata-
weighting factor, m
R
i
R. The metadata elements, mR
i
R,
which could be expressed a real number between
0 and 1, summarize the reliability and relevance
of the information, l
R
i2
R, to the decision task. Thus,
only the relevant data elements will contribute to
H(h|l
R
2
R)R
Rbecause any data element that is either
irrelevant or unreliable will be ignored as m
R
i
R ~ 0.
The second and third data sets that are grouped
in {brackets} in equation (16) represent the con-
tributions to the relevant data set, l
R
2
R, that result
from the data fusion of two and three data ele-
ments. Note that three-way data fusion generally
will not yield the same result as a collection of
results from pair-wise data fusion. However, the
contribution to the information flow of data fu-
sion involving multiple data elements could be
very important and may be more significant than
that of the single information sources and their
respective pedigree metadata, {l
R
i2
RmR
i2
R}.
A combined uncertainty also applies to the vari-
ables that are interdependent. The interdepend-
ence between two or more variables gives rise to
and necessitates the multi-way data fusion in
equation (16) for l
R
2
R. Fusion algorithms that com-
bine data elements will have to account for this
combined uncertainty in some way. The result of
a complete fusion algorithm must include not
only a combined data result, but a combined un-
certainty derived from a consideration of the m
R
i
R.
How to compute or estimate the combined uncer-
tainty from data fusion of two or more data ele-
ments is context dependent and also is an ongo-
ing research topic.
The fusion interactions in equation (16) are
analogous to the two- three- and…n-way inter-
molecular interactions that contribute to the
thermodynamic and spectroscopic properties of
gases, as expressed in the Virial expansion. (See,
for example, [9] and [5].)
A general evaluation of H(h|l
R
2
R) is beyond the
scope of this paper. The data set, l
R
2(min)
R, that de-
scribes the minimum of multivariate function,
H(h|l
R
2
R), is not only unknown but is impossible to
derive analytically in actual, non-trivial cases for
the following reasons.
The exact form of H(h|l
R
2
R) itself is un-
known and may not be analytic.
The number of variables, n(l
R
2
R), is gener-
ally unknown and is task dependent.
Therefore, the exact order of the multi-
variate analogs of equation (14) and ine-
quality (15) cannot be determined in gen-
eral.
The distribution of each variable, l
R
i
R, is un-
known and one cannot assume it is a
Gaussian distribution.
The existence of some potentially impor-
tant variables in l
R
2
Rmay be unknown.
Time constraints preclude discovery of the
details of the items above before decision
deadlines.
However, we can ignore inequality (15) if we
assume that some information already is avail-
able in a command center and that this puts
H(h|l
R
2
R) close enough to a local minimum so that
the risk of maximizing H(h|l
R
2
R) would not be an
issue. At best, quantities such as p and task
workload can be determined only in very con-
trolled modeling-and-simulations experiment
[14] where the models and variables can be
evaluated. Even with many assumptions, an ex-
act calculation of information flow, p, and the
final conditional entropy, H(h|l
R
2
R), in actual field
situations is not an option, for reasons listed
above. Therefore, a different approach is needed
to explore the various aspects of p.
8. Metrics for Factors that Influence In-
formation Velocity
Because v(info) depends on both p and t, time
management in organizations is considered here.
Time is an important factor in decision making
as most decisions have, either explicit or de facto
deadlines. The visibility of information, the em-
powerment of people in the organization, and
communications efficiency can affect v(info).
Thus, although equations, such as (1), (12), (14)
and (16), have no exact and direct solutions, we
can observe the effects of better v(info) if they
lead to increases in decision speed and quality.
This section describes each of these factors with
suggested metrics. Unlike v(info), the metrics
described below have no units. The metrics here
depend on time estimates and arbitrary scales
from 1 to 10 for individuals to use when estimat-
ing how information and the speed of its move-
ment affect their tasking in the organization, or
conversely, how their tasks affect the disposition
of information.
These metrics will not track v(info) in all cases
and the correlation between each of these metrics
and v(info) will not always be 1.0. The metrics
are assumed to be independent of each other but
this will not always be the case in actual situa-
tions. However, the improvement over time of
these metrics may be more significant in assess-
ing the impact of improved v(info) in an organi-
zation than any single quantification of the met-
rics themselves in a given instance. Future re-
search is necessary to shed light on this subject.
8.1 Direct Measures in Time Management
Information flow is only one aspect of informa-
tion velocity. Other aspects include the delivery
of the right information in a timely manner and to
the right place or person. Timeliness in measures
of performance and measures of effectiveness has
been studied in the command and control context
[15]. The following two aspects of timeliness
were considered [15]:
Time delay between the moment when the
C
P
3
P system receives a stimulus and the
moment it can deliver a response (i.e. the
phase delay in the system), and
Tempo of operations – number of actions
per unit of time that the system executes –
a measure of how complex an environ-
ment the system can handle (i.e. band-
width).
In contrast, this section describes direct measures
of time segments in the decision process and re-
fers to the model described in section 4.
Direct measures have the greatest simplicity,
utility and appeal when designing a metric, as
compared with causal or effects-based measures.
The many advantages include accuracy, desir-
ability, reliability, speed, and independence.
First, a direct measure is more
Accurate because it does not involve any
intervening processing.
Desirable because it does not necessitate
any assumptions regarding causes or prob-
able effects.
Reliable because the measure correlates
well with the process.
Rapid and continuous since it tends to track
as closely as possible the quantity being
measured.
Independent because applying the metric
does not necessitate an understanding of the
causes and effects.
However, direct measures cannot always be used
because:
They may be too costly or impractical.
They may be too intrusive,
They may interfere with the process being
measured, especially when the human ele-
ment is a significant factor.
The measured quantity may be hidden or
otherwise inaccessible.
A direct measure of time usage associated with
information velocity based on the decision-
making model is defined in equation (17).
(17) IVMDirect = Tg / (Tg + Tr)
where Tg is the average time employees spend in
green decision states (states 2, 3, and 6) and Tr
is the average time spent in red decision states
(states 1, 4, and 5). The assumption inherent in
equation (17) is that Tg and Tr can be measured
directly to yield a fraction of time spent in green
states compared to the total time in the decision
process (green and red states combined).
This metric has the advantages described above
for direct measures. Equation (17) can be accu-
rate to the extent that time spent in various states
of the decision-making process can be estimated
because the metric is tied directly to the states.
The metric also is simple and can be computed
quickly in real time, assuming the information is
accessible. Application of the measure is inde-
pendent of the causes and effects. If a decision-
maker spends less time gathering, preparing and
sharing information during the decision-making
process, the decision can be communicated more
quickly and the information velocity has im-
proved by that relative amount.
Equation (17) can be separated from the harder
problem of understanding cause and effect. Re-
searchers pursuing this deeper understanding can
use equation (17) whereas practitioners can use it
to gain insight into practical situations.
8.2 Causal Measures
Causal measures are an option when direct
measures are intractable, inaccurate, too intru-
sive, too costly, or otherwise unavailable. Yet, at
the time of this writing, these instrumented tools
for direct measures do not exist or are not widely
employed. Moreover, standards [4] have not
been adopted for representing decisions. Direct
measures of information flow are not available.
Therefore, causal measures that could influence
information velocity are considered here. Equa-
tion (18) defines a metric that relates to v(info)
based on causal components.
Min (Vi, Vy, Ep)
(18) IVMCausal = ---------------------
Max (Hh, Pcr, Bc)
All component values in equation (18) are nor-
malized to a 10-point scale. “1” is a low score;
“10” is high score. The components are de-
scribed below.
Vi represents the visibility of information and
decisions across the enterprise that has the poten-
tial to decrease uncertainty as it pertains to task-
ing and decision making. Vi is an estimate of
how immediately accessible information is to
everyone in the organization. To qualify for con-
sideration in estimating Vi, the information
needs to be delivered in a form that is useable,
scalable and manageable. Usability includes con-
ciseness, generic descriptions, tiered structure,
and net-centric accessibility. For example, a tool
has been developed to provide this type of visi-
bility [16]. Vi can be estimated using an arbitrary
scale from 1 to 10 as follows.
1-“no visibility or web presence;”
5-“Web-based visibility through search,
blogs, wikis;”
10-Web-based 30-second situational
awareness at any level of the organiza-
tion.
Vy represents the visibility of the decision-maker
across the enterprise that increases information
flow. Vy is included in equation (18) because
important information is tied closely to the au-
thor, emphasizing pedigree metadata where
knowledge of the author affects the usability of
the information. People in the organization can
help each other because they can see the impor-
tant concerns and needs of others. The use of Vy
recognizes that the source of information is as
important as the information itself. The funda-
mental issue is whether the concerns and needs
of decision makers are visible to the organization
in a clear and efficient format.
One option is to generalize Vy to any factor that
affects the reliability of Vi, including but not
limited to the visibility of the decision maker.
For example, level and ease of web presence for
Vy also can account for the visibility of pedigree
metadata that supports the decision process by
estimating the level of uncertainty, or conversely,
the reliability of the information that contributes
to the decision maker. Components of a pedigree
metadata set can include uncertainty of sensor
data, data-fusion algorithms, and visual observa-
tions as specified by the observer. The impact of
these factors evaluated with respect to the visibil-
ity of the decision maker can be summarized us-
ing a scale of 1 to 10 that is similar to the one
used to estimate Vi. More detailed formulae to
estimate Vy can be determined in future work.
Ep represents empowerment of people in the
enterprise to increase information velocity. Ep is
an estimate of authority and responsibility shar-
ing throughout the enterprise, that includes de-
centralized control, and how much people are
allowed to share information. For example, one
would expect to estimate an elevated Ep where
the management philosophy reflects the goals
and values of a “high-performance organization.”
High scores for Ep imply that supervisors listen
to and support employees and try to implement
employee suggestions. Elevated Ep implies an
inverted pyramid organizational structure and/or
a flatter hierarchy. The best way to estimate Ep is
through user assessments and employee surveys
that probe the flexibility of the organization’s
policies. (See, for example, [17], [18], [19],
[20].) A scale of 1- (low Ep) to 10- (high Ep)
could be assigned as follows.
1- Traditional organization management
attitude in which workers are not trusted;
5- Flexible policies with respect for em-
ployees;
10- Employee-led organization, inverted
pyramid.
Hh represents the amount of human-to-human
communication that limits v(info). Hh represents
a traditional but inefficient, unscalable, and
largely unmanageable form of information shar-
ing, symbolized by meetings, telephone calls,
chat, e-mail and conversations.
Using a computer to perform activities that Hh
represents is not much of an improvement from
non-computerized methods that contribute to the
estimation of Hh because it means that new tech-
nology still is used in an old way that fails to
take advantage of the opportunity for a paradigm
shift. (This is like the idea of using the trucks to
transport horses into the battlefield, or using e-
commerce for the on-line purchase of slide
rules.) The lack of significant increases in worker
productivity despite the increase in computer
technology is indicative of the problem.
The first step to improve Hh is to become aware
of how each factor that contributes to Hh can be
reduced without reducing communications effi-
ciency. For example, the Top-Ten Best Practices
[21] address effective meetings. To avoid wast-
ing time, the moderator must strive to keep the
meeting focused on the topic and to adjourn the
meeting when the goal is accomplished. Hh will
be smaller if everyone is not required to attend
meetings that have little potential to improve
v(info). This is a challenge in government or-
ganizations and corporations that may value the
control of employees and the conformity of their
behavior more than they value production effi-
ciency. In this sense, Hh may be related to Ep
because common factors may influence both.
Clearly, top management must be committed to
value time and efficiency to observe any substan-
tial improvements in Hh.
Measures for Hh include objective measures of
the number and length of e-mails, meetings, and
phone calls, as well as the subjective estimates of
user assessments. An arbitrary scale of 1 (low
Hh) to 10 (high Hh) can be assigned as follows.
1- < 20 minutes/day spent on Hh;
5- < 2 hours/day;
10- > 4 hours/day.
Pcr represents the level of pressure and personal
and cultural risk influencing the decision-maker.
Although Pcr is an estimate based on an individ-
ual’s personal reaction, when carefully collected
statistically across an organization it becomes a
more objective factor summarizing the level of
pressure, etc. that people experience in specific
organizational environments. Pcr is an estimate
of organizational pressures to become “yes” peo-
ple who are forced to dilute the truth to protect
individuals from personal risk rather than to fur-
ther the objectives for productivity of the organi-
zation as a whole.
An elevated Pcr is an estimate of the level of in-
hibitions that impede honest, direct opinions
thereby reducing the usefulness of information
provided. Elevated Pcr also indicates an organi-
zation with structures and policies that waste de-
cision-making time while people strive to gener-
ate safe and acceptable answers. Examples of
factors that contribute to elevated Pcr include
risk and cost to the individual; risk and cost to
the teams; the pressure to be a “team” player; the
pressure to stay on schedule; the pressure for
promotion or other rewards. Measures include
objective analysis of the traditional organiza-
tional structures and policies as well as subjec-
tive user assessments. A scale for the evaluation
of Pcr could be constructed as follows (1 low Pcr
to 10 high Pcr).
1- External factors rarely affect decisions;
5- External factors often affect decisions;
10- External factors always affect deci-
sions.
Bc represents the level of barriers to rapid, con-
cise, honest communication. Bc is an estimate of
factors that impose costs in time, discourage par-
ticipation in decision-making processes, and cre-
ate a climate of inefficiency, which can lead to
inactivity. Often the requirements for formatting
and submitting information such as proposals,
ideas, patent applications, reports, and publica-
tions are based on the legitimate requirements of
the receivers of the information to reduce errors
and streamline the review process, rather than to
encourage and promote widespread efficient par-
ticipation.
Bc, therefore, is always estimated from the point
of view of the organizations and individuals pro-
viding information rather than receiving it. A
high-performance organization should promote
the efficient and user-friendly submission and
sharing of ideas by breaking down these barriers
while simultaneously preserving the efficiency
and accuracy of the review process. Efficient
submission and review need not be totally mutu-
ally exclusive. For example, to reduce Bc, well-
designed web-submission processes need to pro-
vide these advantages:
Provide clear, simple, and intuitive in-
structions that explain what to do,
Facilitate initial submission of abstracts
and idea summaries rather than impose
up-front requirements for large amounts
of information that are time consuming to
accumulate and useless to review for an
idea that may be off topic or simply not
needed at the time,
Avoid undue delays and ambiguities in
feedback regarding ideas.
Measures of Bc include objective measures of
the length and subjective estimates of the formal-
ity of policies for submitting information. Other
factors that should be taken into account when
estimating Bc are the number of people partici-
pating in submission opportunities, as well as the
total number of submissions because an efficient
submission process may encourage productive
employees to submit more than one idea. C can
be estimated on a scale of 1 (low Bc) to 10 (high
Bc) as follows.
1- The initial submission of an idea re-
quires a very low investment in time and
is not expected be more than a paragraph.
Because of its brevity, any format is fine
and the idea can be submitted at any time;
5- Fewer than 3 textual pages (e.g. using
Microsoft Word), or 6 briefing slides (e.g.
using Microsoft PowerPoint) with mini-
mal formatting, specific due dates;
10- More than 10 pages expected initially,
detailed formatting requirements that are
ambiguous and hard to implement in a
short period of time, several submission
deadlines that occur in quick succession to
impose a work schedule that is too com-
pressed.
The philosophy of equation (18) is that the
strength of an organization’s v(info), which can
be conceptualized as a chain, is only as strong as
its weakest link. The numerator of equation (18)
represents positive factors that can improve in-
formation velocity. The numerator of equation
(18) is determined by selecting the lowest score
among Vi, Vy, and Ep. This method is designed
to ensure that only the weakest link in this three-
member chain should be used to represent posi-
tive factors in IVMcausal. In contrast, the de-
nominator of equation (18) represents a three-
member chain of negative factors that tend to
limit information velocity. In this case, the larg-
est value from among Hh, Pcr, and Bc is selected
as the weak link in the denominator of equation
(18) to represent the limiting factor, which will
overshadow the effects of the other two factors.
In short, previous research and personal experi-
ence in various projects suggest that increased
visibility of decisions and more visible, empow-
ered decision-makers improve information veloc-
ity, whereas increased human-to-human commu-
nication (meetings, teleconferences, etc.), in-
creases in perceived risk, and barriers to commu-
nication tend to degrade information velocity.
Of course, many other causal measures of a
lesser degree may apply, but the measures cap-
tured in equation (18) are expected to correlate
with the concept of information velocity illus-
trated in Figure 1. Future research described be-
low will explore the performance of the compo-
nents in equation (18) through modeling and
simulation and a practical experiment to validate
the causal metric.
8.3 Effects Measures
Another approach for measuring important fac-
tors that influence information velocity indirectly
is to measure the effects of elevated information
velocity. Improved information velocity by defi-
nition will reduce uncertainty in a timely manner
for the specific decision maker who needs it to
accomplish a current, time-critical task. Im-
proved information velocity also will improve
some state of the organization. Therefore, meas-
ures of the organization’s state can reflect on the
effects of information velocity.
One difficulty with direct time and causal meas-
ures is that a task activity that demands that an
employee’s time or, alternately, a causal chain
may not have a clear beginning or a clear end.
One way around this difficulty is to measure a
number of quantities along the causal chain. The
long-term objective of this research area in gen-
eral and this study in particular is improved deci-
sion making. Therefore, we identify factors in
various categories to improve v(info) with the
overall objective to increase the efficiency and
quality of decisions. The effects of improved
V(info) can be estimated using equation (19) as a
metric.
(19) IVMEffects = < N
R
D
R> <QR
D
R> <SR
D
R>
where <N
R
D
R> is the average number of decisions
per unit time, <Q
R
D
R> is the average estimated
quality of these decisions, and <S
R
D
R> is the aver-
age satisfaction of the individual with the rate of
uncertainty reduction. In short, improved infor-
mation velocity should increase the rate of deci-
sion making, raise the average quality of the de-
cisions, and increase the flow process. These ef-
fects measures are difficult to quantify and cer-
tainly require either instrumentation of the deci-
sion-making process and/or subjective assess-
ment by individual participants. The average
number of decisions per time unit is best ac-
quired through instrumentation of decision-
making tools. Some assumptions to achieve this
will be necessary. For example, it will be neces-
sary to determine where one decision ends and
another begins. This will not be trivial when de-
cisions are interdependent.
The quantification of the quality of decisions is a
difficult task. Previous research suggests various
pitfalls, including the typical pitfall of tying as-
sessments too tightly to results based on 20/20
hindsight. The satisfaction, S
R
D
R, of individuals
with the uncertainty reduction in the decision-
making process is important because, according
to equation (1), it should correlate highly with
information flow.
The individual Q
R
D
R and SR
D
R that contribute to the
averages in equation (19) can be estimated on the
scale of 1 to 10, whereas N
R
D
Ris really a measure
of decision speed.
9. General Metric
Ideally, this initial work would lend itself to a
simple metric or rough ad hoc measure, in addi-
tion to a full set of metric components for con-
tinued research. This general metric would pro-
vide a quick “back-of-the-envelope” calculation
that anyone could perform to estimate the per-
formance of the organization and to assess the
value of certain policies or techniques. This met-
ric could be a few questions to pose to an indi-
vidual or subgroup of an organization to assess
the state of information velocity quickly, regu-
larly and efficiently. For this purpose, the follow-
ing survey questions are offered [15], [22].
1) What percentage of your day do you spend
in meetings, reading and writing e-mail,
talking on the telephone, in teleconfer-
ences, and in other forms of conversation
and communication with others?
2) What percentage of your day do you spend
preparing products intended for sharing in-
formation? For example, what percentage
of your day do you spend preparing brief-
ing slides, reports, agendas, minutes, and
completing forms and logs?
3) What percentage of what people are doing
in your organization is important and rele-
vant to you and your assigned tasks?
4) What percentage of what others across
your organization decide is important
throughout the day is visible and easily
understandable to you?
5) What percentage of what you decide is im-
portant is visible and appreciated across
your organization or enterprise on a daily
basis?
The percentages derived from these questions
could be combined as follows to give a score
ranging from 0 to 100.
(20) IVM% = { (Q3% + Q4% + Q5%) / 3
- Q1% - Q2% + 100 } / 2.
How IVM% differs between one individual and
the next, and how it changes for a single individ-
ual over time is more significant than any iso-
lated value of IVM%. Rank orderings of IVM%
are more important than absolute values of
IVM%. However, “100” appears in equation (20)
to ensure non-negative values for IVM%. Equa-
tion (20) weights evenly improved efficiency of
sharing information (by moving away from tradi-
tional human-to-human communication methods
and products) with visibility, awareness, and
relevance of what people in the organization are
doing. An assumption associated with equation
(20) is that if the important work of the organiza-
tion is visible, understandable, and relevant, the
right information is getting to the right person at
the right time. The second line of equation (20)
implies that human-to-human communication
methods are inefficient and unscalable.
For example, an overloaded, discouraged, un-
empowered employee might provide input to
equation (21) as follows: (1) 60%; (2) 20%; (3)
20%; (4) 10%; (5) 10%. In this case,
(21) IVM% = { (20 + 10 + 10)*0.33 - 60
- 20 + 100} / 2 = (13.2 - 80 + 100)/2 = 17%.
This low score reflects the situation of an em-
ployee who cannot share important information
and who is spending almost all the employee's
time on inefficient communication mechanisms.
In contrast, consider a high-level executive who
believes the organization is operating in concert
with the executive's philosophy and instructions.
However, the executive still spends most of the
day using a portable e-mail device and attending
meetings. Such an executive might provide the
following input to equation (21): (1) 80% (2)
10% (3) 100% (4) 90% (5) 80%. In this case,
(22) IVM% = 50%.
This score suggests that the executive is sharing
information at half strength. The executive gets
useful information, but is still using inefficient
and unscalable human-to-human communication
and lacks organization-wide visibility.
Consider a third example of an employee at any
level who benefits from a new organizational
structure and tool that enables the employees, the
employees' decisions, and the employees' deci-
sion-making processes to be visible and easily
understood across the organization. Assume the
organization is synchronized though the use of
the new structure and tool, such that the em-
ployee's vision is shared and appreciated by the
organization and vice versa. Such an employee
might provide the following input to equation
(21): (1) 10% (2) 5% (3) 90% (4) 80% (5) 80%
(23) IVM% = 84%.
This high score reflects good information flow,
since the decisions of the organization and the
employee are visible, concise, and relevant to
both the organization and the employee, and both
benefit from the use of improved, more scalable
communication methods and structures. Ques-
tions 1 though 5 together with equation (20) pro-
vide somewhat arbitrary percentages based on
rough estimates that may be difficult to measure.
However, individual employees and organiza-
tions can use the general metric method de-
scribed above as a point of departure for im-
provement. Here again, the trend of the organiza-
tion as it changes over time, according to equa-
tion (20), may be more significant than any sin-
gle assessment of an organization taken in isola-
tion. The results can be averaged across a statis-
tically significant sample of employees to assess
information velocity. For organizations where
information velocity is high, the right informa-
tion goes to the right person at the right time.
10. Limitations of the Methodology
As with all approximate methods that rely on
estimates and assumptions to make the process
of assessment tractable, the metrics described in
this paper will be more valid in some instances
than in others. This section describes the condi-
tions under which the method will be most useful
and valid. IVMDirect, as defined in equation
(17), is not a direct measure of p or v(info). It is a
metric for time management in decision making
the validity of which depends on the following
assumptions.
Better time management increases v(info).
States and substates described in Figures 1
and 2 are independent of each other.
Time spent in green states increases
v(info).
Time spent in red states does not increase
v(info).
Decision makers have the information
they need in green states.
Decision makers can define start and end
times for entry and exit of various states.
Sometimes the states described in Figures 1 and
2 can be interdependent in which case, Tg and Tr
might not be independent. Increased time spent
in state 1 refining the data and presenting them in
a form that decision makers can understand may
reduce significantly the time spent in states 2 and
3. However, the model implies that the time
spent in state 1 is inefficient whereas only states
1 and 2 will get credit for the efficiency.
Decision makers don’t always have the informa-
tion they need in green states. They may have
some of the information they need. They might
not be able to wait until they have all the infor-
mation they need because decision deadlines can
occur before sufficient information is received.
Decisions made under these conditions can re-
flect considerable uncertainty. Thus, the time that
the decision maker spends on a particular task
does not always correlate to preferences but may
be regulated by the decision deadline. A decision
maker may spend less time gathering, preparing
and sharing information because less time is
available before the decision deadline and not
because the decision will be better if made before
additional facts and alternatives emerge. Simula-
tions and surveys to test and evaluate IVMDirect
will need to be designed very carefully to con-
sider the effects of decision deadlines.
The model described in Figures 1 and 2 does not
explicitly account for the time that support per-
sonnel spend performing data fusion and data
integration and preparing information for the
decision maker. This would occur presumably
somewhere between states 1 and 2. Time spent
preparing the data for the decision maker is not
time wasted and may be included in more detail
in a future refinement of the model.
Another limitation of the time-measurement ap-
proach is that it may not address adequately the
relationship between information quality and the
timeliness-quality tradeoff. The time spent in
certain states depends on the quality of informa-
tion that enters these states. More time may be
needed to refine, process, and understand the
meaning of a noisy or incomplete data set. If
more time is spent on a task the quality may in-
crease. Quality, which is not (yet) part of the
IVMDirect, can be the cause of spending more or
less time with a data set, a report, etc.
However, IVMEffects depends on estimates of
decision quality and user satisfaction with the
quality of information, Q
R
D
R and SR
D
R respectively,
in equation (19). Thus an approach that combines
the metrics of IVMDirect and IVMEffects may
be more accurate than either metric used in isola-
tion. The challenge with IVMEffects lies in the
subjective nature of the estimates of N
R
D
R, QR
D
R, and
S
R
D
R. NR
D
R may be difficult to estimate because it
depends on a value judgment of where one deci-
sion ends and another begins. It also depends on
being able to select a time interval that reflects
when one decision is finished and another be-
gins. For example, if a decision is detailed with
many steps, is each step considered a separate
decision or are they counted as one?
IVMDirect also does not account for the uncer-
tainty before and after the information was
passed. It is possible to have a high measure of
IVMDirect, i.e. low time in green states and high
time in red states, and still not transmit any in-
formation that reduces uncertainty. This observa-
tion also underscores the need for a combined
metric that includes the contributions of IVMDi-
rect, IVMCausal, and IVMEffects in one equa-
tion or algorithm.
11. Ongoing and Future Research
As is the case with any metric, the metrics de-
scribed here need to be demonstrated and vali-
dated in studies, surveys, experiments, and ob-
servations. Although this preliminary research
has suggested some metrics to estimate factors
that affect v(info), unanswered questions remain.
For example, how should the metric components
be normalized, weighted, and combined for an
optimal, single value? Do the causal measures
actually measure factors that correlate well with
v(info) in simulations and in practice? How do
the various metrics interact and correlate with
each other? Can we design and build tools to
measure observable variables in the decision-
making process? What emergent behavior may
appear in simulations that model p and v(info)
under expected conditions?
To address these questions, current and future
research will explore the behavior and perform-
ance of the metrics in a modeling-and-simulation
environment. The work described above lends
itself to modeling with Markov processes, poten-
tially hidden-Markov processes, Bayesian adap-
tation, and agent-based techniques. Preliminary
work in the area of agent-based modeling sug-
gests some interesting behavior [23]. The results
of an agent-based model suggest that a tradi-
tional organizational hierarchy and promotion
policy will suppress and even stagnate poten-
tially useful information [23].
Future research will explore components that
contribute to improved v(info) in a modeling and
simulation environment to determine whether the
theoretical analysis matches the expected per-
formance in typical use cases. Then the metric
will be used in a realistic organizational setting
by developing a baseline, employing a v(info)
improvement technique, and using the metric to
estimate the resulting improvement in v(info).
The directional component of the v(info) (See
Figure 1.) will be addressed in future work. This
relates to directing the flow of information to the
right decision maker.
Future studies will include modifications to
IVMDirect to include a quality factor and esti-
mates of uncertainty, as well as time estimates.
The relationship between IVMDirect, IVM-
Causal and IVMEffects needs to be explored to
use the quality factors in IVMEffects to compen-
sate for the lack of quality factors in IVMDirect.
The resulting theory also is needed to relate
IVMDirect, IVMCausal and IVMEffects to pro-
vide a more unified approach to factors that can
affect v(info) and their interdependences. For
example, section 10 above expresses the need for
a metric that combines IVMDirect, IVMCausal,
and IVMEffects in one equation or algorithm.
This future research also will be directed toward
testing the assumptions that supported the devel-
opment of the metrics described in this paper.
Whereas experience suggests that the following
assumptions are good in most cases, they still
need to be tested.
Improving organizational v(info) improves
decisions by timely reducing uncertainty.
Direct human communication, such as
meetings, telephone calls, and e-mails, lim-
its the velocity of useful information
Reducing this and other inefficient forms of
communication will increase v(info).
Honest, efficient and visible decisions in-
crease v(info) and are derived from em-
powered employees without much cultural
bias or other undue pressures.
Formality, deadlines, and format require-
ments can reduce information velocity from
the sender’s perspective, even if they sup-
port efficient evaluation.
Decision products need to be transformed
from both the formal and time-consuming
briefing slides and reports, and the informal
unmanaged personal conversations, meet-
ings, telephone calls, into a managed, con-
cise, dynamic, lightly structured, generic
format that is useful for sharing information
rapidly across the organization.
12. Conclusion
This paper introduces the concept of information
velocity, which combines the notion of informa-
tion flow and direction with time dependence. A
theoretical development is described to explain
the relationship between entropy, uncertainty
reduction, information flow, information veloc-
ity, and the time dependence that is so important
in making agile decisions. The relationship of
entropy in thermodynamics to entropy in infody-
namics is discussed, as well as the analogy be-
tween information flow and the flow of a liquid
or a gas through pipes. An equation was derived
to show the relationship between information
velocity and power.
A decision-making model focused on time man-
agement is described. This model breaks the de-
cision process down into various states that a
decision maker will experience in pursuit of use-
ful information that supports uncertainty reduc-
tion.
Direct, causal and effects-based measures for
estimating factors that affect information veloc-
ity in organizations in general were introduced.
The advantages and limitations of the metrics are
discussed. In these metrics, time management
and information visibility are central, contribut-
ing themes due to their importance in decision
making. These metrics can be applied to military
command centers to assess information velocity
and agility in the decision-making process. The
goal of this research is to develop and refine a
method for estimating factors that influence in-
formation velocity across an organization or en-
terprise to help determine whether the right in-
formation is getting to the right person at the
right time. This will support creative, agile deci-
sion-making.
2BAcknowledgements
The authors thank the Office of Naval Research for
their support of this work, and Drs. I. R. Goodman
and J.R. Smith for helpful discussions. This paper is
the work of U.S. Government employees performed
in the course of their employment and no copyright
subsists therein.
3BReferences
[1] N. Webster, Webster's New Universal Unabridged
Dictionary, Deluxe Second Edition, (Webster's
New Twentieth Century Dictionary), J. L.
McKechnie, editor, Simon and Schuster, New
York, N.Y., p. 587, 1983.
[2] Sir William Thomson Lord Kelvin,
HTUhttp://www.zapatopi.net/kelvin/quotes/UT
[3] J.E. Russo and P.J.H. Schoemaker, Winning Deci-
sions, New York: Doubleday, 2002.
[4] J. Waters, R. Patel, J. Eitelberg, and M. Ceruti, “A
Proposed Common Decision-Exchange Protocol
for Representing, Managing, and Sharing Organ-
izational Decisions,” Proceedings of the Intl.
Conference on Naturalistic Decision Making
(NDM’09), London, UK, June 2009.
[5] M.G. Ceruti and S.H. Rubin, “Infodynamics: Ana-
logical Analysis of States of Matter and Informa-
tion,” Information Sciences, 177(4), pp. 969-987,
14 Feb. 2007.
[6] M.G. Ceruti, B.M. Thuraisingham, and M.N.
Kamel, “Restricting Search Domains to Refine
Data Analysis in Semantic-Conflict Identifica-
tion,” Proc. of the 17
P
th
P AFCEA Federal Database
Colloquium and Exposition 2000, pp. 211-218,
San Diego CA, 6-8 Sep. 2000.
[7] D. Swanson, T.L. Wright, M.G. Ceruti, D.R. Wil-
cox, and S.C. McGirr, “Design, Development,
and Reuse of Software Agents in the Knowledge
Management for Distributed-Tracking (KMDT)
Program,” Proc. of the 2
P
nd
P Intl. Workshop on On-
tology, Conceptualization, and Epistemology for
Software & Systems Engineering (ONTOSE 2007)
27-28 June 2007, Milan Italy.
[8] http://en.wikipedia.org/wiki/Information_flow_
(information_theory), 29 Oct. 2008.
[9] G.W. Castellan, Physical Chemistry, Addison-
Wesley Publishing Co., Reading MA, 1964.
[10] M.G. Ceruti A. Ashfelter, R.Brooks, G.Chen, S.
Das, G. Raven, M. Sudit, and E. Wright, “Pedi-
gree Information for Enhanced Situation and
Threat Assessment,” Paper 43, Proc. of the 9
P
th
P
IEEE Intl. Conference on Information Fusion
(FUSION 2006), Florence Italy, 10-13 July 2006.
[11] http://en.wikipedia.org/wiki/Positive-
definite_matrix, 4 Dec. 2008.
[12] E.R. Lancaster, “A Method for Finding a Mini-
mum of a Multivariate Function with Applica-
tions to the Reduction of Missile and Satellite
Data,” Proc. of the
T14th Annual National Confer-
ence Meeting of the ACM
T, Cambridge, MA, pp.
70-1 - 70-4, 1959.
[13] D.F. Shanno and K.H. Phua, “ALGORITHM 500
Minimization of Unconstrained Multivariate
Functions (E4), ACM Trans. on Mathematical
Software, 3(1), pp. 87-94, Mar. 1976.
[14] B.G. Knapp, “C3I Information Flow Models:
Integrating and Interfaces,” Proc. of the Work-
shop on Engineering of Computer-Based Systems
(ECBS’97), pp. 136, 1997.
http://www2.computer.org/portal/web/cdsl/doi/10
.1109/ECBS.1997.581838.
[15] P.H. Cothier and A.H. Levis, “Timeliness and
Measures of Effectiveness in Command and Con-
trol,” Information Technology for Command and
Control, S.J. Andriole and S.M. Halpin, Eds.,
IEEE Press, New York N.Y., 1991.
[16] J. Waters, J. Eitelberg, D. Kellmeyer, and M.
Stelmach, “Knowledge Web for the SOC: Inter-
operability through the Use of Open Standards,”
Space and Naval Warfare Systems San Diego Bi-
ennial Review, TD 3202, pp. 54-59, Sep. 2006.
[17] M. Biazi and J. Walinga, “Employee Empower-
ment Study,” 20 May 2007.
http://axiommedialab.com/portfolio/employee_em
powerment_Research_Methodology.pdf
[18] A. Mueller, “Employee Empowerment,” St.
Louis Business Journal, 15 Apr. 2005.
[19] Navran Associates, “Employee Empowerment
Evaluation Kit,” pp. 1-4, 2004
http://www.navran.com/downloads/Employee-
Empowerment-Evaluation-Survey.pdf
0B[20] D. Seeley, “Industry Insights: Create Culture that
Empowers Your Employees,” Business Courier
of Cincinnati, 6 Jan. 2006.
1B[21] SSC, “About Continuous Process Improvement
(CPI) and the SSC Pacific Balanced Scorecard,”
16 Sep. 2008. http://sepo.spawar.navy.mil/
Process_Improvement.html#TTBP
[22] J. Baron, Thinking and Deciding, New York:
Cambridge University Press, 2000.
[23] J. Waters, R. Patel, J. Eitelberg, and M. Ceruti,
“Investigation of Information Flow in Hierarchi-
cal Organizations Using Agent-Based Modeling,”
Proceedings of the ACM/SCS Agent-Directed
Simulation Symposium (ADS’09), San Diego,
CA, Mar. 2009.
SSC San Diego … on Point and at the Center of C4ISRSSC PACIFIC … on Point and at the Center of C4ISR
Information-Velocity Metric
for the Flow of Information
through an Organization:
Application to Decision Support
Information
Information
-
-
Velocity Metric
Velocity Metric
for the Flow of Information
for the Flow of Information
through an Organization:
through an Organization:
Application to Decision Support
Application to Decision Support
Jeff Waters, Ritesh
Patel, James Eitelberg,
Gunnar
Ramstrum
and Marion Ceruti, Ph.D.
14th
ICCRTS, Washington, D.C.
15-17 June, 2009
SSC PACIFIC…on Point and at the Center of C4ISR
22
Presentation Topic Outline
Presentation Topic Outline
Information flow
Information velocity, v(info)
Relationship between information and power
Reducing uncertainty in decision making
Can we measure v(info)? Yes and No
Information-flow model for decisions support
Can we measure factors the influence v(info)? Yes
-
Direct measures
-
Causal measures
-
Effects measures
SSC PACIFIC…on Point and at the Center of C4ISR
33
Information Flow
Information Flow
p = Information flow, summarized as the
difference between in conditional entropy, H(h|l)
, of variable, h, before the process started given
the variable l
1
and after the process finished,
given the variable l
2
.
p = H(h|l
1
) - H(h|l
2
)
H(h|l
1
) is high because many alternative COAs
are consistent with sparse data.
H(h|l
2
) is low because the few alternative COAs
are consistent with the new data set.
p corresponds to the reduction in uncertainty that
results from the receipt of new data.
p depends on the specific task.
p has no explicit time dependence.
SSC PACIFIC…on Point and at the Center of C4ISR
44
Information Velocity
Information Velocity
v (info) is defined as the speed and direction of
information flow, p.
First time derivative of the information flow.
Explicitly a vector quantity.
v (info) = dp / dt
= d H(h|l
1
) / dt – d H(h|l
2
) / dt
For example h = COA, l
2
= a new data set.
Depends on the task tractability, T
y
, & the
power
of information to reduce uncertainty.
An important topic of research aimed at
reducing uncertainty in time-constrained
decision making scenarios as fast as possible.
SSC PACIFIC…on Point and at the Center of C4ISR
55
Taxonomy of Information Velocity
Taxonomy of Information Velocity
Information Flow:
Magnitude
Information Velocity
Velocity
Direction:
Information should
flow in the right
direction to the right
person or network
site where the
information is
required for tasking.
Quantity: How
much information
can move?
Time management: How
fast can information move?
Defined by entropy &
uncertainty reduction.
No direct, general metric
or calculation available.
Modeling & Simulation
Bayesian Networks
Indirect methods:
Current paper
Future research
Assumptions
Approximations
SSC PACIFIC…on Point and at the Center of C4ISR
66
How Is Information Related to Power?
How Is Information Related to Power?
2
nd
law of thermodynamics, dS = dU
rev
/ T
S = entropy, U = reversable heat, T = temperature
Infodynamic analog:
H(h|l) =
W/ T
i
W = work, T
Y
= task tractability, assumed constant at given entropy
W = F dX=J (d
2
X/ dt
2
) dX
X = distance, J = information (analog of mass).
dH(h|l) / dt = (J / T
i
) (d
2
X/dt
2
) dX/dt
Combining these equations with p yields
v(info) = d/dt{(½ J X
d
2
)
1
/T
Y1
-( ½J X
d
2
)
2
/T
Y2
}
where X
d
= d X/dt
Energy = ½J X
d
2
Power = d/dt ( ½ J X
d
2
)
Take away from the derivation:
The rate at which
information travels is proportionate to power.
SSC PACIFIC…on Point and at the Center of C4ISR
77
Can We Measure Information Velocity?
Can We Measure Information Velocity?
Yes. In modeling-and-simulation experiments.
Where we can define and control all the variables.
No. Even with many assumptions, v(info) is too
hard to evaluate in practice because:
The nature of data interactions is not always known.
Pedigree metadata elements may not be available.
The data sets may be incomplete.
Data and pedigree metadata are time dependent.
Data distributions may not be Gaussian.
The form of H(h|l
2
) may be unknown.
Time constraints preclude detailed enough data
analyses in command centers.
SSC PACIFIC…on Point and at the Center of C4ISR
88
Can We Measure Factors that Influence
Can We Measure Factors that Influence
Information Velocity?
Information Velocity?
Yes. Focus on practical time management in
command centers using the following metrics:
Direct measures
Causal measures
Effects measures
Consider a decision-making process model of
time management in organizations in general
with applications to command centers
This model includes:
Measures of time spent on various tasks
Visibility of information
Empowerment of people
Direct, causal, effects metrics as unitless indices
SSC PACIFIC…on Point and at the Center of C4ISR
99
Information Flow Model for Decision Support
Information Flow Model for Decision Support
SSC PACIFIC…on Point and at the Center of C4ISR
1010
Information Flow Model:
Information Flow Model:
Expanded Decision
Expanded Decision
Substates
Substates
SSC PACIFIC…on Point and at the Center of C4ISR
1111
Assumptions in the Model & Metrics
Assumptions in the Model & Metrics
Metrics for time management in decision making
depend on the following assumptions.
Better time management increases v(info).
States and substates
described in Figures 1 and 2
are independent of each other.
Time spent in green states increases v(info).
Time spent in red states does not increase v(info).
Decision makers have the information they need in
green states.
Decision makers can define start and end times for
entry and exit of various states.
SSC PACIFIC…on Point and at the Center of C4ISR
1212
Direct Measures
Direct Measures
IVMDirect = Tg / (Tg + Tr)
Tg = time spent in green states
Considered productive activities
Tr = time spent in red states
Considered unproductive activities
Advantages of IVMDirect:
Time is an important factor in velocity.
Simplicity
Does not depend on causes and effects.
Disadvantages of IVMDirect:
State boundaries are not always clear.
Red states may contribute to uncertainty reduction.
All necessary info may not be available in green
states.
SSC PACIFIC…on Point and at the Center of C4ISR
1313
Causal Measures
Causal Measures
Min (Vi, Vy, Ep)
IVMCausal =
Max (Hh, Pcr, Bc)
Vi = visibility of information and decisions
Vy = visibility of decision maker and metadata
Ep = empowerment of people to increase v(info)
Hh = amount of human-to-human comms
Pcr = level of pressure, personal & cultural risk
influencing the decision maker
Bc = level of barriers to rapid, concise, honest
communication
Numerator: select smallest value of Vi, Vy, Ep
Denominator: select largest value of Hh, Pcr, Bc
SSC PACIFIC…on Point and at the Center of C4ISR
1414
Effects Measures
Effects Measures
IVMEffects = < N
D
> < Q
D
> < S
D
>
< N
D
> = Average number of decisions per unit time
<Q
D
> = Average estimated quality of decisions
< S
D
> = Average level of satisfaction of the
individual with the rate of uncertainty reduction
SSC PACIFIC…on Point and at the Center of C4ISR
1515
General IVM Metric
General IVM Metric
IVM% = { (Q3% + Q4% + Q5%) / 3
- Q1% - Q2% + 100 } / 2
Q1 What percentage of your day do you spend in meetings, reading
and writing e-mail, talking on the telephone, in teleconferences, and
in other forms of conversation and communication with others?
Q2 What percentage of your day do you spend preparing products
intended for sharing information, eg. preparing briefing slides,
reports, agendas, minutes, and completing forms and logs?
Q3 What percentage of what people are doing in your organization
is important and relevant to you and your assigned tasks?
Q4 What percentage of what others decide is important across your
organization or enterprise? What percentage of what others decide
is visible and easy for you to understand on a daily basis? (Use
a
single percentage.)
Q5 What percentage of what you decide is important is visible and
appreciated across your organization or enterprise daily?
SSC PACIFIC…on Point and at the Center of C4ISR
1616
IVM Measures Are
IVM Measures Are
Unitless
Unitless
Estimates
Estimates
Min (Vi, Vy, Ep)
IVMCausal =
Max (Hh, Pcr, Bc)
IVMDirect = Tg / (Tg + Tr)
IVMEffects = < N
D
> < Q
D
> < S
D
>
IVM% = { (Q3% + Q4% + Q5%) / 3
-Q1% -Q2% + 100 } / 2
All variables in
IVMCausal can be estimated on a
scale of 1 to 10
IVMDirect
is a ratio of times so units cancel out.
< N
D
> = integer
<Q
D
> & < S
D
> -
estimated on a scale of 1-10
IVM% depends only on percentages.
SSC PACIFIC…on Point and at the Center of C4ISR
1717
Limitations of the Methodology
Limitations of the Methodology
& How to Modify the Approach
& How to Modify the Approach
IVMDirect
does not account for the uncertainty
before and after the information was passed.
IVMDirect
could be high and still not reduce uncertainty.
IVMEffects
depends on subjective estimates of
decision quality & user satisfaction.
The ability to count decisions to calculate
IVMEffects
depends on the a subjective estimate of
where one decision ends and another begins.
An approach that combines IVMDirect, IVMCausal,
IVMEffects, and/or IVM% may be more accurate
and useful than any single metric.
Changes in metrics may prove more useful than
any single measurement.
SSC PACIFIC…on Point and at the Center of C4ISR
1818
Ongoing and Future Research
Ongoing and Future Research
Ongoing
The metrics need to be demonstrated and validated in
studies, surveys, experiments, and observations.
Explore the behavior and performance of the metrics in
a modeling-and-simulation environment
Investigate how the variables in the metrics interact
and correlate with each other.
Future
Determine ways to use the metrics in exercises and in
systems development for command and control.
Research useful ways to normalize, weight, and
combine the metrics for an optimal, single value result.
IVMDirect, IVMCausal, IVMEffects, IVM%.
Compare the result of a combined metric to the results
of individual metrics.
SSC PACIFIC…on Point and at the Center of C4ISR
1919
Conclusion
Conclusion
Concept of information velocity combines the
notion of information flow and direction with time
dependence.
Theoretical development explains the relationship
between entropy, uncertainty reduction, information
flow, information velocity, and time dependence.
A decision-making model that is focused on time
management divides the decision process into
various states that a decision maker will
experience.
Direct, causal and effects-based measures were
introduced to provide metrics to estimate factors
that affect information velocity.
Goal: Support creative, agile decision-making.
SSC PACIFIC…on Point and at the Center of C4ISR
2020
SSC PACIFIC
on Point
and at the Center of C4ISR
SSC PACIFIC…on Point and at the Center of C4ISR
2121
Backup
information
SSC PACIFIC…on Point and at the Center of C4ISR
2222
Uncertainty Reduction Depends on Data,
Uncertainty Reduction Depends on Data,
Pedigree Metadata, and Data Fusion
Pedigree Metadata, and Data Fusion
Assume H(h|l
1
) is large compared to H(h|l
2
)
Minimize
H(h|l
2
) (max uncertainty reduction)
H(h|l
i
) depends on multiple data elements, l
i
The importance of each l
i
depends on the
pedigree metadata, m
i
weighting factors.
To minimize H,
H(h|l
i
) = and
d
2
H(h|l
i
) / dl
i
2
>
Uncertainty reduction depends on 2-, 3-, and…
n-way fused data. Example: l
2
data set.
l
2
= { {l
i 2
m
i 2
}, { l
a2
m
a2
, l
b 2
m
b2
},
{ l
a2
m
a2
, l
b 2
m
b2
, l
c 2
m
c2
},…{ l
a2
m
a2
l
n 2
m
n2
}}
In general, l
2
is too difficult to evaluate during the
timeframe in which a decision must be made.
SSC PACIFIC…on Point and at the Center of C4ISR
2323
Taxonomy of Information Velocity
Taxonomy of Information Velocity
Magnitude
Information Velocity
Direction:
Information should
flow in the right
direction to the right
person or network
site where the
information is
required for tasking.
Quantity:
How
much information
can move?
Time management: How fast
can information move?
*
Magnitude
Direction:
Information should
flow in the right
direction to the right
person or network
site where the
information is
required for tasking.
Quantity:
How
much information
can move?
Time management: How fast
can information move?
Information Flow:
Defined by entropy and
uncertainty reduction.
No direct, general metric
or calculation available.
Modeling & Simulation*
Bayesian Networks*
Indirect methods:
Current paper
Future research
Current paper
Future research
Assumptions
Approximations
*
Velocity
SSC PACIFIC…on Point and at the Center of C4ISR
2424
Metrics in Time Management
Metrics in Time Management
(12) H(h|l
i
) =
... In addition, it attempted to identify other key socio-technical factors that could play a vital role in the improved management of information flow. [15] indicated that the rapid flow of information in the organization strengthens its efficiency. Therefore, a metric was proposed for the speed of useful information as it flows in an organization. ...
... Various factors, techniques, and tools can help improve information velocity. According to [15], a direct measure of information flow does not appear to be possible in the near future in light of the given the state of technology at the current time. Therefore, a focus on the timing aspect of information flow which deals with time management in organizations is critically needed. ...
... Therefore, a focus on the timing aspect of information flow which deals with time management in organizations is critically needed. [15] described an individual decision-making model that attempts to fill this gap. The researcher proposed formulas for measures and suggests a small, simplified set of questions to obtain a quick metric, and described future research. ...
Article
Full-text available
Abstract: This study aimed to develop and to validate a Strategic Information Systems Planning (SISP) model designed to improve the information flow performance by incorporating the antecedent factors into information technology components,information needs and security of information flow factors. The methodology was carried out in four phases; the theoretical study, the conceptual framework design, the survey design, the data analysis, and the model development and validation phase. Questionnaires were distributed to 350 employees at the Ministry of Higher Education and Scientific Research (MoHESR) and four (4) government universities in Yemen for data collection. Based on the correlation results, it was found that information technology components have significant relationship with information needs. Moreover, information technology components have a direct significant relationship with the security of information flow in the MoHESR and universities in Yemen and they have indirect influence through the information technology components on the security of information flow in the MoHESR and universities in Yemen. By adopting the Structural Equation Modelling (SEM) analysis using AMOS, the results indicated that all the fit indices satisfied the recommended range of value. This implies that the model developed was acceptable. The validation results also revealed that the entire model fitness was appropriate and evidence of the stability of the conceptual framework used in this study.
... Instrumentation could support the development of a metric of information flow and help us optimize our decision processes across our organization or enterprise [7]. Visibility of the decisions in their formation and evolution would enable proactive management and assistance from others [8]. ...
... Research shows that an analytical solution of information velocity is intractable but metrics that support the understanding of information flow can be useful [8]. An agentbased model for information flow can be used to characterize physical analogs to causal measures [6]. ...
... Information, its suppliers, and consumers are then treated as agents. The behavior of the agents and system as a whole can be discussed and infodynamic analogs of thermodynamic and other physical quantities associated with these processes could be explored [8]. The use of conceptual analogs from the physical domain implies the viability of future ontologies to characterize information flow. ...
Conference Paper
Full-text available
This paper describes the work of the W3C Decisions and Decision-making Incubator 1 , with the goal to identify requirements for a standard decision format, through a set of use cases, and to develop a first version of a potential standard format for representing decisions, fulfilling the requirements of the use cases and exploiting semantic web standards. Ongoing efforts include the identification and modelling of 'decision patterns' and development of proof-of-concept applications to validate assumptions and patterns.
... CDEP has sufficient complexity to deal with decisions taken often where it is clear how the decision is taken and what criteria are used. Typical applications given by [10] are the exchange of decision information during emergencies and also in warfare situations. These reactive situations are in contrast to much organizational and business decision making "where well-defined processes change slowly" [10]. ...
... Typical applications given by [10] are the exchange of decision information during emergencies and also in warfare situations. These reactive situations are in contrast to much organizational and business decision making "where well-defined processes change slowly" [10]. Decision making in an organizational context must consider the organizational hierarchy, organizational protocols and sources of knowledge and information within the organization. ...
Chapter
Effective decision making can help organizations to manage complexity. Here we argue that considering decisions as units of organizational knowledge and providing a means for decision storage, retrieval and reuse can facilitate effective decision making. To enable the recording and retrieval of decisions, a conceptual model is presented that can be used as a set of requirements for a data structure for decision storage. The approach conforms partly to the proposed Common Decision Exchange Protocol (CDEP) standard, but we extend it to capture decision attributes, decision making stages, decision makers and other collaborators, and the information and tools used. Capturing the linkages between decision elements enables a wide range of organizational decision making processes to be encoded for ontological reasoning. We show the derivation of a new conceptual model from cases. We use the context of decision making in Product Lifecycle Management (PLM) to motivate and illustrate the conceptual model.
... CDEP has sufficient complexity to deal with decisions taken often where it is clear how the decision is taken and what criteria are used. Typical applications given by [10] are the exchange of decision information during emergencies and also in warfare situations. These reactive situations are in contrast to much organizational and business decision making "where well-defined processes change slowly" [10]. ...
... Typical applications given by [10] are the exchange of decision information during emergencies and also in warfare situations. These reactive situations are in contrast to much organizational and business decision making "where well-defined processes change slowly" [10]. Decision making in an organizational context must consider the organizational hierarchy, organizational protocols and sources of knowledge and information within the organization. ...
Chapter
Effective decision making can help organizations to manage complexity. Here we argue that considering decisions as units of organizational knowledge and providing a means for decision storage, retrieval and reuse can facilitate effective decision making. To enable the recording and retrieval of decisions, a conceptual model is presented that can be used as a set of requirements for a data structure for decision storage. The approach conforms partly to the proposed Common Decision Exchange Protocol (CDEP) standard, but we extend it to capture decision attributes, decision making stages, decision makers and other collaborators , and the information and tools used. Capturing the linkages between decision elements enables a wide range of organizational decision making processes to be encoded for ontological reasoning. We show the derivation of a new conceptual model from cases. We use the context of decision making in Product Lifecycle Management (PLM) to motivate and illustrate the conceptual model.
... CDEP has sufficient complexity to deal with decisions taken often where it is clear how the decision is taken and what criteria are used. Typical applications given by [10] are the exchange of decision information during emergencies and also in warfare situations. These reactive situations are in contrast to much organizational and business decision making "where well-defined processes change slowly" [10]. ...
... Typical applications given by [10] are the exchange of decision information during emergencies and also in warfare situations. These reactive situations are in contrast to much organizational and business decision making "where well-defined processes change slowly" [10]. Decision making in an organizational context must consider the organizational hierarchy, organizational protocols and sources of knowledge and information within the organization. ...
Conference Paper
Full-text available
Effective decision making can help organizations to manage complexity. Here we argue that considering decisions as units of organizational knowledge and providing a means for decision storage, retrieval and reuse can facilitate effective decision making. To enable the recording and retrieval of decisions, a conceptual model is presented that can be used as a set of requirements for a data structure for decision storage. The approach conforms partly to the proposed Common Decision Exchange Protocol (CDEP) standard, but we extend it to capture decision attributes, decision making stages, decision makers and other collaborators, and the information and tools used. Capturing the linkages between decision elements enables a wide range of organizational decision making processes to be encoded for ontological reasoning. We show the derivation of a new conceptual model from cases. We use the context of decision making in Product Lifecycle Management (PLM) to motivate and illustrate the conceptual model.
Conference Paper
Full-text available
This paper describes heuristics for identifying semantic conflicts in large, heterogeneous, distributed, and federated databases in which the combinatorics of comparisons become an issue. Data can be divided along several dimensions and the search space for conflicts can be narrowed considerably. The paper considers various search criteria, such as frequency of use, importance of data and error correction, ability to fix, and data categories. Tradeoffs in reducing the search domain and the relationship of this work to other areas of computer science also are described. The paper concludes with a discussion of future directions and applications.
Conference Paper
Full-text available
Motivation – Share information; make better decisions. Research approach – Explore the use of XML code to document and transmit characteristics of decisions. Findings/Design – The paper recommends a CDEP. Research limitations/Implications – This work implies that we need an open-standard body similar to OMG in the object-oriented world. This would help to make decisions and their explanations more interoperable across a wide variety of organizations. Originality/value – The value of this work will be realized in military and other types of command and intelligence centers where decision tracking serves training and analysis purposes to preserve lessons learned. Take-away message – The authors welcome comments on this proposed decision-exchange protocol and suggestions and want to collaborate with others on this effort, which supports a global-interoperability test bed.
Conference Paper
Full-text available
Knowledge Management for Distributed Tracking is an ongoing research and development program to improve naval command, control, and decision support. The program's approach includes modeling and simulation, intelligent software agents, integrated-sensor ontology, and reusing methods from past projects. The software simulates a hypothetical scenario designed to demonstrate how knowledge-management technologies can improve situation awareness and reduce information overload. In the simulation, when additional information is desired about an unknown contact, intelligent-software agents are deployed on a secure, web-like network to find sensor data collected at other friendly platforms. The information is fused to reduce the uncertainty in the detection, localization, tracking, classification and identification of the unknown contact. The paper describes how the agents interact with the integrated sensor ontology, which facilitates distributed, heterogeneous sensor-data fusion.
Conference Paper
Full-text available
This paper describes how pedigree is used to support and enhance situation and threat assessment. It is based on the findings of the technology group of the Data Fusion Levels Two and Three Workshop sponsored by the Office of Naval Research held in Arlington, VA from 15-18 Nov. 2005. It identifies areas that need improvement in situation assessment and threat assessment, such as interoperability, automation, pedigree management, system usability, reliability, and uncertainty. The concept of pedigree must include "standard" metadata, lineage, plus a computational model of the quality of the information. The system must automatically propagate changes and update to derived products when source information or source-pedigree information changes. Several other processes must be automated: generate pedigree, identify and auto fill gaps, fuse pedigree, update pedigree, display of information quality and confidence. The paper concludes with suggestions for future research and development
Conference Paper
Full-text available
This paper describes an agent-based model for exploring the effects of traditional supervisory merit-based promotions and hierarchical organizational structure on the flow of information. Each agent simulating a worker follows the same rules for handling information and promotion: Agents differ only in the breadth of information they consider important. The model considers three types of information, "red," "yellow," and "green." One-third of the workers (red) consider only red information important. One-third of the workers (yellow) consider both red and yellow information equally important. The final third of the workers (green) consider red, yellow, and green information equally important. Red workers are more focused, restricted, and narrow-minded in the type of information they consider important, whereas green workers are more unfocused and unrestricted. Starting with an organizational structure with only one supervisor and moving toward a tiered organizational structure through promotions, the objective of the simulation was to determine how this agent-based model affects the flow of information that workers consider important. The results indicate that the more narrowly focused workers will dominate the organizational structure, thereby suppressing the flow of information they do not consider important.
Article
Full-text available
A methodology for assessing the effectiveness of command, control, and communications systems is extended to include timeliness. The assessment is based on comparing the properties of the system to the mission requirements when both are expressed as loci in a commensurate space of measures of performance. The methodology for evaluating measures of effectiveness is illustrated through application to an idealized fire support system. Copyright © 1986 by The Institute of Electrical and Electronics Engineers, Inc.
Article
One requirement common to all Monte Carlo computer simulations is an abundant and automatic supply of random numbers. For most purposes it generally suffices to draw this supply from the uniform distribution in the unit interval. The mathematical tricks ...
Article
This paper expands and consolidates the use of analogies in thermodynamics to explore concepts in the characterization of information systems. The analogy spans the range of information systems to include databases, knowledge bases and model bases. It includes but is not limited to pressure, expressiveness, temperature, tractability, degrees of order, systems of liquid–liquid equilibrium and disjunction in information-systems integration. By taking advantage of the isomorphism that exists between states of matter and states of information, we can understand new ways to characterize and measure information systems. This paper is the fourth in a series describing new aspects of “infodynamics.”.
Conference Paper
This report describes human-centered modeling efforts recently completed or ongoing so that users and developers of military C3I systems quantitatively assess changes in operational procedures and information technology brought about by today's political-military climate. One model is an information flow and task workload simulation, another is a rule-based information quality assessment model, and a third is a tool to model skill-ability demands for C3I jobs. Although each model has proven highly useful and successful independently, an intelligent interface linking all three models is proposed. This in effect establishes a C3I meta-model, where users can directly visualize and frame their C3I system questions, using a query and feedback interchange, and the interchange results are automatically translated to model terms and variable settings, and computational instructions