As we approach the 11th anniversary of the 9/11 terrorist attack, we can be grateful that nothing like it has happened since. But that doesn't mean that something very much like 9/11 — or even worse — couldn't happen again. In fact, new research suggests that we may be seriously underestimating the risk of another large terrorist event.
In fact, statisticians say there's a 50% chance that another 9/11-scale attack will happen again, within the next ten years.
Show me your dataset
Data is the fuel that drives statistics. Without it, statisticians cannot make proper predictions, whether it be the next hurricane, earthquake, or stock market crash. The same can be said for terrorist attacks.
And like all meteorological, geological, or economic events, these attacks tend to happen at different frequencies and at different intensities. Plus it's not enough to predict an event — what's often more valuable is the ability to predict the accompanying severity.
Such is the case with terrorism. And as a recent study by statisticians Aaron Clauset and Ryan Clauset suggests, the 9/11 attacks provided a valuable piece of information to their dataset, namely the so-called "large terrorist event." The attacks on the World Trade Center and Pentagon resulted in nearly 3,000 fatalities — nearly six times more than the next largest event. The question that Clauset and Woodard wanted to answer was, just how probable or improbable was this event? And just how likely is it to happen again?
Algorithms, outliers, and tails
To help them make their assessment, they gathered a global database of 13,274 terrorist events dating from 1968 to 2007. Then, rather than treating 9/11 as an outlier event, they created a statistical algorithm for estimating the probability of similarly large events (including global terrorism) within complex social systems. And their broad-scale and long-term probabilistic approach is similar to the one used in seismology, forestry, hydrology, and natural disaster insurance.
Specifically, their algorithm combined multiple models of the data distribution's tail and various computational techniques to account for parameter and model uncertainty.
Using their algorithm, they estimated that the historical probability of a 9/11 scale event was between 11-35%. Given this relatively high range, and given that 9/11 did in fact happen, Clauset and Woodward were satisfied that their model was generating reasonable estimations. Armed with this, and applying three different futurological scenarios, they were able to estimate the likelihood of future terrorist events.
Technology Review elaborates:
Assuming that the number of terrorist events per year remains the same as it is now, about 2000 per year, then the likelihood of another 9/11 is between 20 and 50 per cent, depending on the choice of distribution.
The 50 per cent prediction comes from the power law distribution which many experts argue is a good fit to the data. By that measure, a catastrophic attack is as likely as not.
Of course, conditions might change. The current level of terrorist incidents is heavily influenced by the number of attacks in Iraq and Afghanistan. It's possible to argue that the numbers will drop in the near future as these regions become more stable.
In that case, the likelihood of a 9/11-type attack in the next ten years drops to between 5 and 20 per cent, say Clauset and Woodward.
But as Clauset and Woodward note, the number of attacks could increase on account of any number of social, political, or economic factors — things like ongoing environmental degradation or rising food prices. Should this prove to be the case, they predict with 95% certainty that another 9/11 scale attack will occur within the next ten years.
So, what does 50% actually mean?
All this said, and on the face of it, their 50% normative assessment seems profoundly unsatisfactory. It's like a weather forecast in which a meteorologist says it might rain — or it might not. It seems like they did a hell of a lot of work simply to produce a rather wishy-washy prediction.
And while that's certainly one way of looking at it, there is a decidedly disturbing side to this figure — especially given that the 1968-2007 estimation was as high as 35% — and that 9/11 did in fact happen. As frustratingly ambiguous as the 50/50 chance appears to be, it is certainly far higher than a reasonable person would hope.
Check out the entire study here.
Images: history.army.mil, Aaron Clauset & Ryan Woodard.