Slashdot Mirror
Statistical Analysis of Terrorism
Harperdog sends in a Miller-McCune story about Aaron Clauset, a researcher whose studies on the statistics and patterns that arise from large numbers of terrorist attacks could help governments better prepare for such conflicts and reduce uncertainty about their frequency and magnitude. Quoting:
"After mapping tens of thousands of global terrorism incidents, he and his collaborators have discovered that terrorism can be described by what mathematicians call a power law. ... Using this power law relationship — called 'scale invariance' — the risk of a large attack can be estimated by studying the frequency of small attacks. It’s a calculation that turns the usual thinking about terrorism on its head. 'The conventional viewpoint has been there is "little terrorism" and "big terrorism," and little terrorism doesn't tell you anything about big terrorism,' Clauset explains. 'The power law says that's not true.' Massive acts of violence, like 9/11 or the devastating 1995 bombing of the US embassy in Nairobi, obey the same statistical rules as a small-scale IED attack that kills no one, Clauset's work suggests. 'The power law form gives you a very simple extrapolation rule for statistically connecting the two,' he says."
1st ish
interesting...first
No doubt this sort of analysis will soon be used to plan terrorist attacks.
Oh yeah, let's add another layer of technocrat masturbation to the problem. How about we just stop killing and otherwise pissing off brown-skinned people?
That might work right up until the planners of the attacks read his paper and change their plans accordingly.
...rather than reaction or "prediction".
The government is very happy letting irrationality dictate discourse. Fear keeps rational discourse out of the conversation. It is much better to have the people think that it is a good idea to duct tape themselves into their homes and suffocate. Fewer trouble makers that way.
Rhymes that keep their secrets will unfold behind the clouds.There upon the rainbow is the answer to a neverending story
The observation of scale invariance in this kind of data tells you nothing about the short term relationship between low level and high level attacks. Physicists really shouldn't be doing statistics...
Interesting article. I however have a beef with the thought that "“This gives you hope that terrorism is understandable from a scientific perspective.” " Furthermore, I have a problem with his thought that patterns of probability can be seen to develop over time, while not explicitly stating _when_ an attack will happen. To me, that's akin to stating that the San Andreas fault-system will trigger with a mounting probability over the years. Of course it will - as tension builds at some point it's inevitable that the fault will release. When world governments have bad foreign policy (which most seem to have at least some time, if not most of the time), of course you're going to create disenfranchised members of the world community - and when you arm and train them like Al Quaeda & Taliban were when the west wanted them to fight the Soviets they will turn on you. It's not a matter of if, but when. Stating that over time the likelihood of an attack increases seems to not scientific, but rather obvious as somebody (or some group) is almost guaranteed to slip through the security net in place to detect/predict such actions. My $0.02
Statistical analysis shows that the amount of terrorist incidents is actually quite small, but the governments around the world like to exaggerate how many there actually are, to deprive decent hard working people of their freedom and democracy, and pee a lot of money up a wall in the process.
Take Nobody's Word For It.
We found that we can't really reduce it to the realm of P, so we can tell you how big the attack might be, it just might take a few decades and hundreds of computers working together. We can get a grant for that right?
so what? earthquakes are also a power law distribution.
doesn't really help in prediction.
These are the same type of guys that gave us statistically accurate risk modeling for the complex derivative securities and we know how well that turned out. One must be careful with mathematical models, especially when you're modeling sentiment.
This is just like all those crap magazines you can buy to show you how to make millions in the stock market. There is always someone willing to look at a graph of past occurrences, draw a line through it, and show you the formula for what happened.
The trick being, of course, that they are all 100% worthless for predicting future trends. The only thing you know in the stock market is this: If a stock is going up, it can continue to go up. Or it might stay the same or go down. The only thing this guy will learn from his analysis is that there might be another terrorist attack. Or there might not. And it might be more, equal, or less severe than previous attacks.
Then they wouldn't be terrorist attacks. The element of surprise is the chief weapon.
"I bless every day that I continue to live, for every day is pure profit."
... misleading, after all many so-called terrorists are merely frustrated people who have not had their voices heard or who have been abandoned by lawless and reckless rulers or who've had their countries unlawfully invaded.
I wonder if these studies check the conditions that these "terrorists" arise out of.
If the parent was generalizing, her would have said something like Muslims are the terrorists.
Let's say this - if we stopped meddling in Middle Eastern affairs, we'd see a huge reduction in terrorism. Because, in the last 15 years most of the terrorist attacks have been made by Muslims (mostly Arab) pissed off at the US for supporting Israel, being in the Middle East and basically throwing our weight around like we own the Goddamn planet.
Yes, we'll still have to occasional Tim McVeigh or abortion clinic bombing, yeah, yeah, yeah - heard it all before.
Does the quoted author mean the 1998 Nairobi embassy bombing?
Or is he just so meta he doesn't even need to get the date right?
An old prof told me that everything is a straight line in the log-log paper. You can literally draw any conclusion you want once you choose the axes to be logarithmic.
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
Americans seem to ignore the most important stat about terrorism that there is, you are almost infinitely more likely to be killed by an SUV than you are by a terrorist. And yet Americans are uber paranoid about terrorism and yet go apeshit for their shitty ass, ugly, poorly performing, insanely dangerous SUVs. Wake the fuck up people.
Monstar L
into other country's affairs making people mad enough to become so called "terrorists"? Did they they also include all the people such as Osama that the CIA trained then decided to exterminate when they had serve their usefulness? Oh Im pretty sure this will be taken seriously because it has a lot of graphs and math in it. What bunch of American bullshit.
This information is useless in terms of predicting a large attack in the short term.
If you flip a coin 100 times each time it has been heads, it doesn't mean the next time its going to be tails.
In the same way, just because lots of small attacks have happened without a large attack doesn't mean a large attack is about to happen.
Guess this methodology is just bullshit then.
I question whether road-side bomb attacks against soldiers in Iraq and Afghanistan should be considered terrorism (I am, perhaps wrongly, assuming that most such attacks are against soldiers rather than civilians). If soldiers are being 'terrorised' by the threat of facing bombs, they probably aren't very good soldiers.
or rather, as soon as it starts to work, and stops terrorist attacks...
well then it stops working, doesn't it?
You reply suggests that you misunderstood.
The power law doesn't suggest where and when an attack happens, so it can't stop a single one from happening. Statistics doesn't predict that just like statistical climate laws won't predict whether it rains tomorrow or not.
The power law only says how many attacks will probably happen in the next period of time in a certain large area - within a certain degree of freedom.
And with that infinite wisdom, politicians are able to take appropriate measures. That's the whole point of it.
Today, politicians scream the loudest so that all voters can hear they take the strongest measures against terrorism of all. That may not be necessary when terrorism can be regarded with the same statistics as traffic deaths, plane crashes, diseases and other causes of death.
Maybe in the future politicians will say that it is indeed a little pointless to allocate 20% of the annual governmental budget to prevent 3000 terror-deaths, while the same money could save 100,000 in hospitals if it were to be spent on medicine rather than anti-terror measures. (But maybe that's just my wishful thinking).
Besides, I don't think you can't stop terrorism. You can only motivate people not to be a terrorist in the first place.
Once people cross a line, and decide they want to hurt our society, they will. Somehow.
"'The power law form gives you a very simple extrapolation rule for statistically connecting the two,' he says" ..as long as all terrorists are perfectly spherical and act in a complete vacuum.
Yes, actually, it did; the largest acts of terrorism are perpetrated much less frequently than the smaller ones; for every major bombing, there have been many smaller ones, for every major war, there have been several smaller ones. For every smaller wart, there have been many large acts of non-war violence, etc. down to for every shooting, there have been many aggressive confrontations.
If the power law were not true, most conflicts would escalate; we see they don't, and small events are more likely than large ones.
Terrorism is also threatening to blow people up just because you can. It's also threatening economic sanctions or embargoes if certain ultimatums are not satisfied. By this measure, the United States government is the largest and best funded terrorist organization in the world.
There are a variety ways we express it: an private diplomatic threat, a publicly implied threat, an threat of economic sanctions through the UN (while we ignore UN resolutions against us), military "exercises", CIA coups, and of course, the outright invasions and public threats of invasion.
America is like the local mafia that you have to do business with, or else you could end up like that guy down the street.
According to this 2005 Nature News article about Clauset and his research, the observation that social interactions (deadly feuds) follow a power law distribution dates back at least half a century. Along a similar vein, Neil Johnson (of University of Miami) and research collaborators recently produced a decent model for this kind of distribution (see their paper in Nature from last year).
Engineers also speak PDE, only in a different dialect.
It's like showing a relation between murder and crime. During some period of time there is more likely to be a murder if there is more "other" crimes. A better statistic would be to show the relationship between the size of a country's military and the number of wars it instigates.
... I remembered reading about a study like this years ago. Turns out, I did. http://physicsworld.com/cws/article/news/21465
Before you design for reuse, make sure to design it for use.
Statistic should not be used to try and predict phenomena which depend on variables or data that are neither quantifiable nor reliable.
SB
It's old. The more humans I meet, the more I like my cats. At least they are honest.
when you are mismatched n technology and they are pissing you off. Anything you do back n retaliation is terrorism. Stealing from a vendor OMG hes scared now = terrorism, the whole phrase is as bad as the term HACKER...LOOK em both up.
This math should have been kept top secret so the terrorists couldn't use it to plan future attacks where we, using this math, would not expect them to. :-/ Now it's already obsolete.
I'm afraid you are wrong. Compare it to the number of car accident deaths, for example. (and yes I know the source isn't great, but you are welcome to google to your hearts content to see if you can show me how terrorism is more dangerous than driving or riding in a car.)
Guns don't kill people; Physics kills people! - John Lithgow as Dick Solomon on Third Rock From The Sun
The difficult work in predicting terrorism is getting human intelligence - people on the ground, double agents, infiltrating the terror groups, and getting civilians to tell you the suspicious activity they saw. That is hard, dirty, dangerous work.
Mathematical equations are certainly helpful, and technology is very helpful (see drones), but don't think these things are going to replace HUMINT.
Just because you can get the data to map to a statistical function does not mean the the function is predictive. There are lies, damn lies and this guy.
Facts take all of the premium out of arm waving - T. Reynolds
If terrorism follows the power law, then that actually just confirms that terrorism is as random as sunspots and global earthquakes.
It's random. Nothing to see here.
You do realise that these are trivial observations and that power laws are everywhere, right? Some people are going as far as to claim that we should stop calling the normal curve by that name, because the power laws seem to be the norm in the universe. And no. Drawing up a nice distribution graph does not help in predicting anything. This work is just more of the same old publish-or-perish. By the way citations to journal articles obey a power law, as does the wealth of people, firms, and nations. The variance in the stock market prices? Power law. To go from the fucking shiny graph to the idea that I CAN PREDICT THE FUTURE!!! is nothing but modern astrology.
It's a question of psychology, not physics, statistics, or politics. The latter three will not convince a rational person to kill, or even harm another person.
For justice, we must go to Don Corleone
> Actually, it's very simple: Blowing up innocent people, just because you can, is terrorism.
No--terrorism requires some component of "terror." Blowing up innocent people often qualifies, but not always. For example, blowing up innocent people may be genocide, with an intention of eliminating--rather than terrifying--a population. Or it may be an untempered reaction to being a twenty-year old who's just seen his friends killed--a twenty year old with automatic weapons, who lashes out too easily at a race he dehumanizes because the enemy largely consists of people of that race.
Also, by your definition, terrorism in asymmetric warfare would not count as terrorism--because there, people aren't blowing up innocent people just because they can. They're doing it because they can *And* because it gives them something--a way to fight back against an overwhelming force, or a way to maintain control over their own people, or satisfaction of a perverted concept of honor.
It's very simple and true that blowing up innocent people is wrong--we leave aside the moral dillemna involved in blowing up an innocent person to kill Hitler or Stalin. But that doesn't mean that all blowing up of innocent people is terrorism, nor that all terrorists blow up innocent people.
-- IANAL, this isn't legal advice, and definitely isn't legal advice for you. Also, Squee!
Now that we have a _so called_ statistical analysis of small scale terrorism and linked it to big boom, most terrorists will deliberately avoid matching up with the statistical predictions in future. (Terrorists may be dumb, but they're not all uneducated and blind!)
There are two things that can happen. Only one can be true since each invalidates the other.
The statistics can account for the fact that terrorists are aware of the statistics and has therefore already accounted for willful attempts at avoiding being at the end of a fed-powered statistic solution. However, this would void the stats for catching terrorists who are unaware of the stats.
What fun lies in double-binds? Who came up with this idea anyway? Certainly not someone who knew what he was thinking!
Geekism is your _only_ God!
I doubt very much that there is enough terrorism-related data to have much confidence in this analysis. You need quantities lijke the number of words in a book to extract with any confidence Zipfs law or most other power laws. Even then, there are enough deviations from the straight line (on logscale, of course) to make the drawing of conclusions a very interesting job indeed.
Paai
The Nairobi US embassy bombing was in 1998: http://en.wikipedia.org/wiki/1998_U.S._embassy_bombings
I think the parent thread is pointing out that in terms of saving lives, money spent on road safety campaigns and improving driving ability might be better value than spending it on counter-terrorism in the way we do right now.
But you also make the fine point that spending money on public health and health education would also be a fine use of that money, perhaps even a better one. It's probably easier to spend money on driver education and reducing road deaths than curing the common cold but anti-poverty measures, better public health care for at-risk populations etc would also be worthwhile spending and as you note probably save many lives.
Is becomming a really serious problem coupled with the media-beloved mantra "Scientists say ..."
In general public policy, but particularly:
AGW, corrupt statistics, modeling
Finance, risk, pricing
Health, pandemics, junk pharma, vaxination fear
You have a single intellectual thread, mis-used mathematics and particularly modeling,
unwarrented veneration of authority, fostered by the MSM, and the results of BS research
focused by directed funding.
The root cause is innumeracy, but focused by too much power in too few un-accountable
hands is becomming a real threat to our political process.
.... ...
devastating 1995 bombing of the US embassy
It was in 1997.
I don't know why you think it means that. It sure as hell doesn't say that. After all, the excerpt you quoted mentions attacks that kill nobody. Just apply some common sense and you'll see that even an infinite number of those wouldn't equate to a single large attack causing hundreds of fatalities. Comprehension fail.
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
Going back to the old pirates and emperors/terrorism vs. freedom fighter argument, do the stats still stand up if you define terrorism differently or use another dataset? No complex system has been reduced to a "very simple extrapolation".
reverse the causal chain. If you see a lot of small terrorist activity, it is because terrorist activity is higher and when it's higher, you should expect more bigger attacks too.
The cause being whatever is making terrorism a more useful option. E.g. higher oppression, greater international tension, invasion or natural crisis causing displacement will all lead to more unrest which, if there is no outlet for it, will release itself as violence and hence terrorism.
And by looking at the more numerous small events you can see whether this is happening with more rapidity than you would by using the less frequent but more visible bigger terrorist attacks.
And then you can either hole up and avoid the problem or try amelioration before a big hit turns up.
So what happened with the mayor today? Suicide?
With the first link, the chain is forged.
You want to know where the next terrorist act will take place?
Just ask the Mossad.
Fer feck's saik!
Is anybody really still fooled by this bullshit? Because, as we know, governments NEVER plan in secret or attempt to manipulate the populace.
Why haven't those monsters been hauled out by their scrotums yet?
Sheesh.
-FL
The Power Law is essentially the inverse of the Bell Curve, another statistical mapping that accurately measures the occurrence of events but doesn't provide any predictive power.
The power law also applies to phase changes of elements.It accurately represents the locations of molecules. The regular old bell curve applies to the locations of molecules when the phase isn't changing. The folks at the Santa Fe Institute think that because Power Laws show up everywhere they are amazing but the SFI folks aren't so amazed at the old bell curve showing up on the perimeter of every place the power law appears.
It is descriptive, not predictive.