Ethos Issue 7, Jan 2010

Opinion: Looking for Trouble
Gary Klein

It is not just that weak signals are hard to detect and understand. They are also hard to communicate and difficult to take seriously, warns naturalistic decision making guru Gary Klein.

INTRODUCTION
Every government has a responsibility to identify potential crises in order to avoid them or ameliorate their effects. The sooner a government can identify a problem, the easier the job of avoiding or containing it because fewer resources will be needed, and more time will be available to mobilise those resources.

Therefore, government agencies may attempt to pick up the weak signals that indicate that a crisis may be brewing. They may develop systematic methods and technologies for scanning more and more data, and for connecting the dots. They may strive for higher levels of accuracy, seeking to root out errors that might lead to mistakes of omission (failing to spot weak signals) or commission (crying wolf). A metaphor would be the use of space-based telescopes to spot asteroids that might be heading towards our planet, in the hope that with enough time we might find a way to deflect the orbit of such asteroids away from us.

While I think that all of these steps are valuable, I have some concerns with each of them. I worry that the way an agency pursues each of these steps can increase its vulnerability, if it is not careful. The steps I have described above emphasise the gathering of information over analysis. They emphasise technology and procedures over expertise. They are best suited for recurring crises, rather than a first-of-a-kind crisis.

Returning to the metaphor of a killer asteroid, the threat of asteroids is not new. It is a recurrent crisis. Our planet has been shaken by asteroids in the past. They are cited as a likely reason for the extinction of dinosaurs, and for a massive explosion in Siberia in 1908. Asteroid-hunting is a poor metaphor for the challenge of detecting first-of-a-kind crises such as the airborne suicide attacks of 9/11, Pearl Harbor, the collapse of Enron, and the current global economic meltdown. First-of-a-kind crises correspond to the black swans that Taleb has discussed.¹ Thus, in hindsight, the weak signals were there in plain view for 9/11, for Pearl Harbor, for the collapse of Enron, for the US economic crisis in the fall of 2007.

Most people missed the weak signals because, as Karl Weick has pointed out in his book Sensemaking in Organizations,² we understand the significance of the weak signals only in hindsight. Weak signals do not announce themselves, particularly for first-of-a-kind crises. In hindsight, we can see the signs of the US housing bubble several years ago. We can shake our heads at the granting of home loans to people with limited resources. But there have been bubbles before this. What made this bubble so devastating was that it was associated with lax regulations, insufficient risk analyses by the Wall Street investment banks, infusion of capital from thriving Asian economies, a perversion of the loan insurance practices for mortgage-backed securities so that they accelerated risk rather than dampened it, loss of transparency for financial instruments, and so forth. Each of these signals was commented on at the time, but it was the combination of these factors that led to the crisis.

In The March of Folly, Barbara Tuchman described a number of world events that seem inevitable in hindsight, but were invisible to the decision makers at the time.³ Gathering more data would not help decision makers who cannot understand what the data mean, and computational models work best by extrapolating from previous events and historical trends. Computational models assume continuity with the past and are even less sensitive to first-of-a-kind events than people are. Using these models is like driving a car by looking in one’s rear-view mirror. The sophisticated computational models used by Wall Street were no match for the realities of the fall of 2007.

Not only are weak signals hard to spot, but when keen observers do pick them up they usually have trouble convincing others to pay attention. In each of the cases I listed above—Pearl Harbor, 9/11, the economic meltdown—a few people identified the weak signals and tried to warn others. In each case, the warning was ignored. It is not just that weak signals are hard to detect and hard to understand. They are hard to communicate because they are so unusual and unlikely that they do not fit the mindset of the decision makers.

Page 1 I 2 I 3