| |
Ethos Issue 7, Jan 2010
Opinion: Looking for Trouble
Gary Klein
TOO MUCH INFORMATION GATHERING
CAN OBSCURE ANALYSIS
Government agencies may attempt to
pick up the weak signals that indicate
that a crisis may be brewing, but the
signals only make sense in context,
in relation to the rest of the situation.
Thus, each of the forces that led to the
financial crisis that hit the US in the
fall of 2007 were fairly visible (excepting
the deceptions practised by Bear
Stearns and others), but there was little
government enthusiasm for reining in
the low-inflation growth that seemed to
be generating such prosperity, until it
was too late.
Agencies may develop systematic
methods and technologies for scanning
more and more data, but a point is soon
reached where more data results in more
ambiguity. The Japanese attack on Pearl
Harbor was fairly obvious, in hindsight,
but the plethora of signals made it easier
to anticipate a Japanese attack on Russia,
or a bold move to the south to ensure a
supply of oil. Besides, decision makers
are very good at explaining away data
that they do not like. Analysts may be
encouraged to "connect the dots", but
the dots are only clear in hindsight;
the skill of analysts is in judging what
counts as a dot in the first place.4
Agencies that feel themselves to
be under pressure often become risk-averse.
They may adopt a zero-deficit
mindset, adding all kinds of practices to
increase accuracy and reduce mistakes,
but reducing mistakes is not the same
thing as fostering insights. If we try
to reduce mistakes by having people
document all of their assumptions and
areas of uncertainty, we may instil a
passive attitude that gets in the way of
actively searching for potential crises.
We see some evidence of this in the
intelligence community, where new
analysts are assured that if they just
follow proper tradecraft (for example,
document assumptions and areas of
uncertainty), no one will blame them if
they make a mistake. If we try to reduce
mistakes by directing people to consider
more hypotheses, we may interfere with
their ability to use their intuitions. If we
persuade people to look for historical
trends, and force them to use statistics to
justify their judgments, we risk missing
the first-of-a-kind crises. Thus, during
the Cuban missile crisis in October
1962, the advocates of tradecraft assured President Kennedy that the USSR would
not try to put ballistic missiles in Cuba
because their entire history showed an
avoidance of risk. Afterwards, when
it was clear that this assessment was
incorrect, the explanation was that
Khrushchev had acted irrationally, not
that their tradecraft had failed.
HOW TO LOOK FOR TROUBLE
Therefore, in addition to the usual
steps, we can take the following
actions: Encourage people to look
into the anomalies they notice rather
than passively keeping an open mind.
Maintain a variety of perspectives and
backgrounds. Help people to see
situations from different perspectives.
Rotate in people who have fresh eyes
and haven’t been part of the prevailing
wisdom. These suggestions are reasonably
obvious, along with the advice to increase
openness and adaptation.
However, organisations incur a cost
when they become less predictable
and run into coordination difficulties.
Predictability is essential for
organisational coordination. Perhaps
what is needed is a split-level
organisational style in which the
leader is responsible for consistent
performance and mature oversight,
but can draw on subordinates who
are speculative, reacting or even
over-reacting to potential trends and
threats. If the leader shows this type
of variability, the organisation may
suffer. Better to depend on subordinates
who are free to speculate. This advice
comes with a catch—subordinates who
do their job of freely speculating will
often be wrong, which will likely reduce
their credibility. It is hard to regain
credibility. Therefore, when leaders find
themselves dismissing comments from
the speculators, it may be time to rotate
in some new warning officers.
One way to see if an organisation is
harvesting the worries of people trying
to pick up weak signals is to look at
the products they generate. Imagine a
continuum from statements that are
very likely to be true, to statements
that might well be wrong. And imagine a boundary line, separating the region
of true statements from the region of
possible errors.
I |
|
