Informa)cs

Lecture 10 Intelligence Analysis

Introduc)on
This lecture represents the culmina)on of our
informa)on hierarchy where we now need to
make a decision
We have gathered and processed informa)on
We might be missing some key informa)on
(will we know?) and some of it might be
wrong or contradictory
Nevertheless we must now decide on a course
of ac)on

Introduc)on
The content of this lecture follows the
organisa)on of the text Intelligence Analysis
a target-centric approach by Robert M
Clark
Intelligence reducing uncertainty in )mes of
conict (compe))on)
Conict or compe))on implies an
unwillingness to co-operate and ac)ons to
hide or deceive

Introduc)on
Customer = the en)ty that will act on the
result of the analysis
Analyst = the en)ty looking at the informa)on
Target = the focus of the problem
Collectors = sources of informa)on
(some)mes commissioned by analysts)
Target Centric all actors involved in
developing the target picture

The Intelligence Problem

The customer begins with a problem
The analyst must liaise with the customer to
understand the problem and dene it with an
agreed model
The analyst can now iden)fy what steps need
to be taken, how to instruct the collectors and
what analysis will be required

The Intelligence Problem

In this way the target problem becomes an
intelligence problem where informa)on gaps
and issues can be iden)ed
Feedback to the customer may lead to the
target problem being rened or modied
Framing eect a tendency to accept
problems as they are presented

The Intelligence Problem

Five ques)ons to dene the problem:
When is the result needed?
Who is the customer?
What is the purpose?
What form of output is required?
What are the real ques)ons?

Models of the target

Intelligence analysis involves iden)fying or
formula)ng a model of the target
A model can be physical or conceptual and
allows the analyst to iden)fy what needs to be
known or monitored
Key assump)ons about the model must be
reviewed and challenged if necessary to
ensure that the model is appropriate

Models of the target

Examples:
Supply chain iden)fy suppliers and loca)ons
Materials sources and processes
Financial iden)fy key indicators
Spa)al loca)ng key resources
Timeline an)cipate events
Social ci)zen ac)ons iden)fy issues

The Analy)c Spectrum

Intelligence is concerned with conict situa)ons
and the mode of opera)on can be over a
spectrum of outcomes:
Preven)on to prevent a situa)on from
developing
Deterrence if an unfavourable situa)on
develops then further ac)on can be deterred
Defeat resolve the conict to our advantage

The Analy)c Spectrum

Intelligence analysis can also operate over a
spectrum:
Strategic long term policy informing.
Complex.
Opera)onal planning and execu)on of
opera)ons predic)ve.
Tac)cal immediate opera)ons
Customers oYen confuse these aspects

Crea)ng the target model

It is essen)al to develop a model of the target to
iden)fy what intelligence is required and how to
analyse what we have
Target or system models can be:
Structural organisa)on
Func)onal specic roles
Process produc)on of an ar)fact

Crea)ng the target model

We can use a number of generic models as a
star)ng point:
Lists arguments for and against
Curves factors follow a specic trend
Pa[ern standard modus operandi
Chronological xed or expected )me scales
Rela)onship people, facili)es, materials
Simula)on predic)ng outcomes

Intelligence Sources
This has been men)oned in a previous lecture
and so only a quick summary is presented
Two categories of source literal and non-
literal
Literal OSINT COMINT HUMINT Cyber
Non-literal SIGINT IMINT etc

Intelligence analysis generally moves from

cheap and easy sources (OSINT) to the more
expensive (HUMINT)

Evalua)ng Data
Once we have formulated a model for the target
we are trying to use data that we have collected
in the model to formulate a recommenda)on
A number of issues arise at this stage which
have been touched on earlier in a previous
lecture:
Evaluate the source is the source
competent, did they have the necessary
access and do they have any bias?

Evalua)ng Data
Evaluate the communica)ons channel could
the length or quality of the channel
compromise the integrity of the data. Could
the channel be intercepted or manipulated
Evaluate the creden)als of the evidence itself
is the data reliable and relevant. The data
might be good but cant be used in our
current model why?
Evidence can be convergent or divergent

Collec)on strategies
The use of a target model can iden)fy what
informa)on is required and then the way in
which that informa)on is acquired has to be
addressed
Step 1 correlate the target model to collec)on
sources to iden)fy gaps in knowledge
Step 2 review the gaps iden)ed and
priori)se. Collaborate with collectors to rene

Collec)on strategies
Step 3 dene collec)on strategies. Iden)fy
op)mum INT source and compare reliability
against priority.
Step 4 plan for future collec)on strategies and
acknowledge denial and decep)on aspects

Although shown as a linear sequence the
process is usually itera)ve

Denial and decep)on

Most targets of intelligence prac)ce some form
of denial or decep)on and this must be
constantly in mind when analysis of data is
being performed
Denial = withholding or hiding informa)on can
be rela)vely easy depending on the
circumstances
Decep)on = providing false informa)on can be
very dicult but incredibly eec)ve

Denial and decep)on

Defend against D&D by denial of your own
intelligence capabili)es HUMINT especially
vulnerable
Complex decep)on is performed with high
value assets percep)on management
Collabora)ve processes involving customers,
analysts and collectors are more robust
The opposite of D&D is signaling an
unocial no)ce of inten)on Falklands War

Analy)c Methodologies
A large number of methods have been
developed over the years by the intelligence
services to analyse data so as to avoid common
errors and tendencies:
Reaching conclusions too quickly
Group think
Ignoring data that challenges the conclusion
Analysing the wrong target etc.

Analy)c Methodologies
There are 3 broad categories of method:
Structured argumenta)on a formal process
of combining evidence graphically or
numerically e.g. Wigmore and Bayesian
Alterna)ve target create the counter-
argument to the consensus view and analyse
Predic)ve see below

Predic)on
To be useful intelligence analysis must be
predic)ve but this is not always possible even
a targets analysts could not predict everything.
Analysis involves predic)ng the future state of
the target by:
Extrapola)on unchanging factors
Projec)on changing factors
Forecas)ng new and changing factors

Predic)on
Typically the analyst will create several scenarios
for the future state of the target based on
dierent assump)ons:
Dene the problem create a relevant
scenario
Iden)fy factors appropriate to the scenario
Rank the factors
Create and rank the scenarios - iterate

Predic)ve techniques
As might be expected, a number of techniques
are available for crea)ng scenarios to predict
the future state of a target
Some will be rela)vely obvious if based on
numerical data which show trends or cycles.
Others will need to be used for more qualita)ve
informa)on

Predic)ve techniques
Extrapola)on assumes that exis)ng forces
will not change. Reliable over the short term
but seldom accurate beyond that. OYen used
for linear or cyclic trends and involve
correla)on and regression techniques
Projec)on assumes that some or all forces
may change and so probabili)es must be
assigned to these outcomes to determine
their inuence inuence trees.

Predic)ve techniques
Forecas)ng assumes that not only will
exis)ng factors change but that new ones will
arise. Factors can include economic, social and
technical and once again probabili)es can be
assigned. Forecas)ng makes the most
extensive use of scenarios to compare
outcomes or determine sensi)vity.

Organisa)onal analysis
It can be incredibly useful to determine the
structure of an organisa)on to iden)fy people,
func)onal units and processes. This can help
iden)fy realis)c boundaries of scenarios or the
an)cipated response of an organisa)on to a
given set of factors.
The inten)on is usually to predict the future
behaviour given a number of scenarios

Organisa)onal analysis
Ra)onal the logical response resul)ng from
the organisa)on ac)ng similar to others
Emo)onal perhaps a key individual
determines the response based upon personal
experiences or expecta)ons
Cultural the organisa)on reacts in what
appears to be an illogical way but the
response is typical within the given culture

Technology and Systems

In both the military and commercial domains it
is essen)al to analyse the impact of the
development or adop)on of a technology.
We could interested in the future performance
of a technology, the use or transfer of a
technology or consider the eect of a technical
breakthrough.
The analysis of technology factors can follow a
progression:

Technology and Systems

Most technologies follow a classic S curve
iden)fy where the technology is at the
present to extrapolate.
Iden)fy innova)ve developments that
represent a disrup)on look outside the
organisa)on for these.
Simula)ons can predict future performance
incorpora)ng a new technology

Conclusions
Intelligence analysis is becoming increasingly
sophis)cated to deal with the vast amounts of
data available
Nevertheless it is s)ll something of an art simply
because of D&D and incomplete data
A surprising amount of material is now
published on the techniques of Intelligence
Analysis