Feb 2

On Killing at Close Range, Stadium Stampedes, Genocides and Technology

If you do, as I have done, read a great deal about history of atrocities and mass killings you learn this: it turns out that it is awfully hard to kill people one by one, unless greatly aided by training and technology. It also turns out, it’s awfully easy for people to die in crowded situations.

Untrained people balk at killing. Dave Grossman’s book “On Killing" is likely the best source of the history of this. During the American Civil war, tens of thousands of soldiers "double-loaded" their muskets—i.e. they didn’t fire the bullet they had before loading another one. Similar stories of soldiers shooting in the air, pretending to shoot, simply not shooting and just ducking litter in the literature of World War I and II. In fact, it was only after the Vietnam War era that the U.S. military "solved" this problem through a combination of professionalization of the military, "distancing" weapons, and improved training. Balking at killing is not only because humans are humane, but also because it is much safer to not escalate fights. People prone to kill are also prone to die in fights—not the best evolutionary strategy. Hence, humans "quasi-fight" all the time —they posture, they punch, they crow— but have a solid aversion to the last step. (And how do you train people to overcome this aversion? Massive repetition in a realistic environment, i.e. an immersive first-person shooter video game.)

Second, the closer the person, the harder it is to kill them. Bayonets are probably the most difficult technology to employ in mass killing, and drone wars are likely the easiest. Here, technology matters a great deal. That is why this video that some of my tweeps were passing around yesterday of these “Swarm of Nano Quadrotors" scares the living lights out of me. This is the technology of mass murder, small numbers at a time—and that is the danger of our era. It turns out the technology of mass murder large numbers at a time, nuclear weapons, is hard to use because of the potential for existential catastrophe. It remains to be seen, though, if "hard to use" turns out to be enough of a safety valve. Future generations, if they exist, will have found out.

Although there are some historical instances of “low-tech” killing of large numbers, especially by untrained people, many of those reveal surprising interactions between technology and the atrocity. In the Rwandan Genocide, for example, about 50% of the victims were killed by machetes (and only about 15% by guns)—and that was enabled by the fact that machetes manufactured in China had become awfully cheap right around then. In 1993, the government of Rwanda imported about $750,000 machetes from China —enough for one new machete for about every third male in the country. The rest is heart-wrenching history. Without the cheap machetes, it might not have been possible to carry out this genocide:  a slower rate of killing does not just mean proportionally fewer people would be killed, it also means that the victims can have time to organize and defend themselves. Speed is often essential to mass-killing; the Nazis aimed to kill their helpless victims within an hour of getting off the train at Auschwitz. Those shouts of “schnell, schnell” barked my Nazi officers at movies are not just fiction. If people have time, they can regroup and resist, even if outgunned.

Photo by Jim Nachtwey of Machetes abandoned by Hutus at the Rwandan border

(Machetes abandoned the Rwandan-Zaire border: Pic by Jim Nachtwey)

Very few people, it also turns out, are natural born killers, especially at a close distance—even if they are grossly evil. SS Reichsführer Himmler, one of the architects of the Nazi genocides of World War II, reportedly fainted at the sight of women and children being shot in the Russian front. He then avoided seeing such scenes and instead tried to devise methods of killing large numbers which did not involve close-range “personal” killing by bullets—hence the use of chemical agents, mass starvation, working to death, etc. In fact, if you read some of the primary research on the Nazi officials, you learn that they encountered an awful lot of problems trying to kill large numbers of defenseless civilians and had to finally engineer a very special method, “the concentration camp”—with the gas ovens and all the ghastly history that goes with it.

Nazis did kill a large number through other, more close-range methods as well; and that is where you see the importance of political side of mass killing—dehumanization of the “other” is almost always part and parcel of “training” people to kill at close range; the virulent racism of the Nazis about not just the Jews but also the Slavs, the Roma, etc. and the massive propaganda about the Tutsi “cockroaches” in the Rwandan genocide—in which, non-incidentally, Radio Rwanda played an important inflammatory role. My awareness of the role of hate-speech and racism in pretty much every single modern mass killing outside of a strictly military context (where people get killed by trained soldiers often from afar using technologies of distancing such as air power) keeps me from becoming a free-speech absolutist on the Internet. 

Finally, on the other hand, it turns out that it is awfully easy for people to die in crowds. They don’t just happen in poor countries but they often happen in stadiums. (For example, 21 people died in 2010 in Germany during a music festival). The reason for this is that people in crowds are somewhat like molecules—if there are a large number of them traveling in one direction and if the “front” hits an object —a door, a wall, a slight narrowing of a tunnel— it doesn’t stop the “back” from advancing. Crowds are hence very dangerous if there is a reason for the speed at some point to become slower in one part of the path. (Researchers at ETH, Switzerland, have been working, for example, to redesign the Jamarat bridge which has been the scene of multiple stampedes and crush deaths during the Hajj pilgrimage—sometimes killing thousands of people at a time.) There are reports that during the awful deaths in the Port Said soccer match in Egypt, steel doors were bolted shut just as a fight broke out and people rushed to the exit.

Hence, my take-away is that we need to remain very vigilant about technologies of distancing killing —drones, these “self-guided bullets” which were announced just yesterday, those nanoquadrotors… Anything that overcomes what aversion we have left to killing —and my belief is that first-person shooter video games and constant violence in movies erodes that aversion, not in the same way for every single person but for some people and for society as a whole . And stadiums and places where large numbers of people gather need trained people who understand these dynamics and can intervene early—and stadiums need to be designed not to make sure that people without tickets can’t ever get in, but so that it is easy to get out in a moment of crisis: no narrowing tunnels, no doors which don’t break under pressure, many wide exits and entrances.

P.S. Please read this post as background to this topic; the facts of the Port Said soccer deaths are still unclear and that is not what I’m writing or making any claims about.

Can tech journalists please learn about random vs non-random sample.

Here’s one more  example of statistical illiteracy which seems to be rampant among tech journalists: 

If you think parents are keeping tabs on their kids’ Facebook profile pages and pictures, you’re absolutely right.

According to a new infographic released by market research firm Lab42, parents are keeping a watchful eye on their child via Facebook, with many checking out their pages daily (43%).

The study — which was conducted among 500 social media users – found that 92% of parents are Facebook friends with their children (of all ages) and more are turning to the site to monitor their kids’ interactions. Safety was named as the top reason for looking at their profiles (40%), followed by curiosity (15%).

Look people, it’s not that hard. If the sample is non-random, don’t use words like “most parents” blah, blah, blah in your title (which was “Most Parents Monitor Kids on Facebook — And Have Their Passwords”). I went back to the original source to check, too. There is not a word about the sampling method. Was this just friends of the journalist? Snowball?  Recruited how? Not a word. 

Not. That. Hard.

Hello, world.

Tumblr site for, Zeynep Tufekci’s blog on technology and society.