ARRC JOURNAL
a few staff officers. Assessments were
conducted once during each battle
rhythm cycle prior to the commander’s
situational update brief. The results
were neither comprehensive nor timely.
The staff lost the initiative by presenting
issues to the commander as they arose
because timely and relevant information
had not been presented earlier. This
deficiency was attributed to the lack of
integration between risk management
and decision-making processes. What
was later discovered, however, was that
the headquarters needed a new risk
assessment process that incorporated
relevant information related to risks as it
became known across the headquarters.
Identifying potential solutions
Bottom line up front, there is no cure-
all solution, and never will be, to
producing accurate, precise and timely
forecasts on the likelihood and impact
of future events. The root causes to
this impossibility include uncertainty,
individual bias and subjectivity. At the
ARRC we decided to mitigate these
impediments by combining alternative
analysis approaches and probability
models like Bayes’ Theorem to close the
gap between the unknown and known as
new information is discovered.
An alternative approach to
group sourcing
One of the main issues with the previous
traditional approach was relying on the
judgement of only a few staff officers
to make organisational assessments.
A departure from the norm was needed
and thus a pan-organisational risk
assessment group was formed to assess
risk likelihood by following a process
designed to solicit individual likelihood
estimates from group members from
across the staff. This risk assessment
group ensures an in-depth and relevant
understanding of the operational
environment,
while
simultaneously
serving as risk detectors because of their
disposition within the headquarters.
In practise, risks are identified by the
risk group, either through pre-condition
decomposition or discovery of new
information, and the information is
shared amongst the group. Groupthink
now becomes an obstacle, as does
positional influence. Alternative analysis
approaches, such as ‘brain writing’,
mitigate these concerns by having each
group member generate their estimate
separately, away from the group, to
reduce the influence of other members’
position or personality. Each member
evaluates the risk and provides their
Figure 1 – Reassessment Process
risk likelihood estimate not knowing how
other members have responded.
Comparing responses and
asking the right questions
Normalising
data
is
difficult
to
accomplish under normal conditions, but
it is near impossible without common
evaluation metrics. Therefore, risk
likelihood estimates must be reported
as a percentage so that a systematic
approach can be applied to provide
insightful feedback. This resulted in the
creation of the ‘ARRC Risk Likelihood
Yardstick’. Risk analysts are asked to
independently develop a number that
represents their subjective assessment
of the effect a specific event has on a
defined risk’s likelihood, using categories
that correspond to percentage probability
values.
Predictably, the range of outcomes
varies greatly from individual to
individual, especially when considering
their perspective of the problem within
the headquarters. Alternative analysis
techniques are used to mitigate these
biases by asking the risk member to
Figure 2 – Risk Likelihood Yard Stick
34
ALLIED RAPID REACTION CORPS
consider multiple alternative futures,
such as one in which the risk develops
and one in which it does not. In each
of these futures they look back into the
past and assess the probability that the
event they are considering occurred.
These questions are often overlooked
in risk analysis because the primary
focus of risk analysis is on causes of
risk. Asking other questions breaks our
natural inclination to foresee the worst
case and forces the analyst to consider:
What else could this mean? This process
could capture the context and reasoning
underpinning an individual conditional
probability estimate or provide a relevant
and detailed analysis of the identified
risks. These individual assessments are
used to produce a single risk likelihood
update in relation to an observed event.