A Robust Feature of Terrorism

December 14, 2010 2:59 PM

He is one of a handful of U.S. and European scientists searching for universal patterns hidden in human conflicts — patterns that might one day allow them to predict long-term threats. Rather than study historical grievances, violent ideologies and social networks the way most counterterrorism researchers do, Aaron Clauset and his colleagues disregard the unique traits of terrorist groups and focus entirely on outcomes — the violence they commit. Call it the physics of terrorism.
posted by chavenet (19 comments total)

I believe it's actually called psychohistory, thank you.
posted by oddman at 3:05 PM on December 14, 2010 [8 favorites]

Ha! Best double-post ever, Mr. Seldon.
posted by vorfeed at 3:23 PM on December 14, 2010 [1 favorite]

Perhaps I'm lacking creativity today, but after reading the article I'm having trouble figuring out what practical application this approach might have, or how it would be different in practice from what is already being done. It can't help with specific locations or indications of when an attack is going to occur, so how are we not back to square one when dealing with how to prepare for a specific attack (if such preparation is even possible)?
posted by Hoopo at 3:24 PM on December 14, 2010

The article fails spell out what they mean by "terrorism", but I think the implied definition is pretty clear.
posted by fredludd at 3:40 PM on December 14, 2010

That the death tolls from terrorist attacks follow a power law seems to be pretty well shown empirically and is by no means obvious. I find the models I've read about it interesting, but I wonder about alternative explanations such as a combination of the distribution of targets and some reasonable assumptions about the impact of weapons. Total city population is clearly scale-free, but I'm not sure about population density, especially instantaneous density when you consider workplaces and markets and such. However, despite any skepticism I may have, Aaron Clauset is really one of the good guys of complex systems research. He doesn't have a tendency to make egregiously overstated claims that is all too common in the field, and I think his quotes in the linked article make it clear that he is asking questions that are within the auspices of what his expertise is in.

Physicists, primarily due to the way that phase transitions occur, are primed to view power laws as evidence that they are looking at a system where details don't matter. While his approach is definitely trying to look at where this pattern comes from, unlike other researchers quoted in this article, I strongly doubt he will throw his weight behind an absurd micromechanical model, nor oversell the results that come of it.
posted by Schismatic at 3:40 PM on December 14, 2010 [2 favorites]

The note about developed vs. developing countries is also one of those things that's not hard to understand.

Developed countries typically have a population who benefits greatly from infrastructure, have little incentive in dismantling it, and a lot of motivation to protect it. Attacks on infrastructure and random violence in the form of terrorism often see great effort to put down- that is, look at how much effort goes in when someone sets off a pipe bomb or commits arson. (Generally, not so much if you're a black church being firebombed or a mosque in the US, for example... but generally yes)

Developing countries often have a few folks who get the most benefit of infrastructure, and good number of people getting little or no benefit. When violence kicks off against the lower or middle class (but not the upper class), the people at the top just as often let it go without response. So now you have both motivation and insufficient deterrence.
posted by yeloson at 3:51 PM on December 14, 2010

Every thread like this must contain a link to Cosma Shalizi's "So You Think You Have a Power Law — Well Isn't That Special?" which acts as a necessary check to scientists' fashion-driven tendency to find power laws everywhere. He does single out for praise one physicist who thinks about power laws in a statistically principled way. It's Aaron Clauset.
posted by escabeche at 3:55 PM on December 14, 2010 [4 favorites]

Hoopo: This approach is complementary to the law enforcement/intelligence approach of stoping specific attacks. You are entirely right, it says nothing about any one bombing. However, scale free distributions, even though they have multiple causes, tend to have origins that are more particular than the standard bell curve (or normal/gaussian distribution) that underlies things like the height or weight of people in a population. Scale free distributions thus suggest that there is at least _some_ universal mechanism that might be at work, no matter the terrorist group, no matter the time. Depending on what it is, a sufficient understanding of this aspect might (lots of handwaving here, naturally) suggest ways of combating the growth of terrorist organizations in the long term.

yelosan: I would suggest that your argument speaks to the number of attacks in each place, but not the number of victims of attacks. One can still come up with just-so stories about it, but I don't think that any one empirically supported answer is nearly as obvious.
posted by Schismatic at 4:01 PM on December 14, 2010

yeloson: I would suggest that your argument speaks to the number of attacks in each place, but not the number of victims of attacks.

Which, would be exactly what I was speaking to. I'm more informed on the issues of infrastructure and class- you'll have to ask someone who knows about the logistics of terrorism for the latter.
posted by yeloson at 4:16 PM on December 14, 2010

Hoopo, it's useful in risk management in somewhat the same way as studying earthquakes is. Of course you would like to actually get predictions of specific events, but this is almost impossible. The next best thing is to have a good way of measuring the severity of an episode. With earthquakes you know that a certain size seismic event will have more or less severe aftershocks, which information can help emergency planners and responders. With terrorism, if you see distinctive patterns of frequency, distribution and severity, you may not know exactly where the next attack will take place but you can make a pretty decent guess about the probability that one will occur within a certain timeframe.

Schismatic, I agree that this is an area to tread lightly and favor analysis over anticipation, given the pitfalls of sample size, reporting bias and so many other complicating factors. I do hope this work will get the support needed for serious research to continue. Although I try to avoid being carried away by mere models, I've felt for some years that ignorance of power laws' manifestation in human activity is a major obstacle to progress. Unfortunately, interest in such things also seems to be subject to a power law distribution...
posted by anigbrowl at 4:19 PM on December 14, 2010

anigbrowl: At least if interest is a power law, the tail is heavy. God forbid an exponential decay on such things...
posted by Schismatic at 4:24 PM on December 14, 2010

Grokking emergent patterns in geopolitical conflict sounds (somewhat) like a specific application of Psychohistory, a fictional science in Asimov's "Foundation" series.
posted by zwaro at 4:33 PM on December 14, 2010

you may not know exactly where the next attack will take place but you can make a pretty decent guess about the probability that one will occur within a certain timeframe.

Fair enough, but this seems to only provide the sort of information that Bush might use to move his wheel-o-terror from threat level orange to yellow or something. Put out a "everyone be extra vigilant today" message sort of thing. Only now they won't have to say they have specific intelligence, they can have SCIENCE! I suppose I'm apprehensive because we certainly don't need to give the people who make the decisions on how to combat terror more tools that can be abused unless they can be shown to provide a real, immediate benefit that can save lives.
posted by Hoopo at 5:02 PM on December 14, 2010

He is one of a handful of U.S. and European scientists searching for universal patterns hidden in human conflicts government grant funding decisions — patterns that might one day allow them to predict long-term threats trends in government science funding.
posted by ennui.bz at 7:05 PM on December 14, 2010 [1 favorite]

Deeply fascinating, thanks for sharing. Might drop a few emails based on this one to a few people in the know. :)
posted by the cydonian at 8:36 PM on December 14, 2010

He is one of a handful of U.S. and European scientists searching for ... trends in government science funding.


Fig. 1
posted by Behemoth at 8:42 PM on December 14, 2010

Developed countries typically have a population who benefits greatly from infrastructure, have little incentive in dismantling it, and a lot of motivation to protect it.

As a counterpoint to this I would like to say:

The Republican Party
posted by dibblda at 11:15 PM on December 14, 2010

People have been trying to understand the causes of conflict for quite some time, and the only real consensus is that 'it's complicated'. Sure there are common factors, greed, grievance, underlying social-economic structures etc., but to try to come up with a single universal explanation seems very far-fetched. But then, I don't understand formulae (these or any others). Maybe they're very complicated formulae.
posted by YouRebelScum at 4:26 AM on December 15, 2010

Well, with individuals the source of the conflict is usually wounded egos; managing egos has got to be at the heart of conflict resolution. I'm not sure how this would extend to nations, however. (North Korea really needs to get laid..?)
posted by LordSludge at 6:15 AM on December 15, 2010