Canaries in a Coal Mine

Posted on December 15, 2012 by

This morning a horrible tragedy shook Newtown, a small town in Connecticut just 70 miles from where I live. An as yet unidentified gunman (there are conflicting reports of his identity) went on shooting rampage at the Sandy Hook Elementary School, killing nearly thirty people, most of them children. We may never find out what the motive of the shooter was (in a substantial proportion of killing rampages it remains unknown). But if we do, the press will explain away this tragic incident of mass murder as an act of an evil, or deranged (or both) individual, as they have done for, literally, hundreds of previous shooting rampages.

Which the shooter, no doubt, is – evil and deranged. But what doesn’t get much attention or discussion, is why incidents of such senseless mass murder have increased so much. The question I want to ask is not why this or that shooting rampage happened, but why they are so much more frequent now than a generation ago.

How much more frequent are they? The numbers are shocking. Between 1960 and 2010 the average number of rampages has increased more than 10-fold:


I made this graph four years ago for a guest post, which I was invited to contribute to the Freakonomics blog at the New York Times.

The numbers came from my US political violence database, which was subsequently published in the Journal of Peace Research. The data do not enumerate all killing rampages – they are a sample obtained by a systematic search of online news resources, such as the New York Times. But the story these data tell – an order of magnitude increase – is very clear.

As far as I know, I am the only one who is tracking numerically what is happening to shooting rampages. (I would be very interested to find out about other research efforts on this topic.)

In my Freakonomics blog, I used the structural-demographic theory to make sense of this tragic trend:

We know that during the 1970s something changed in the American economy, and in a very a fundamental way. Between 1930 and late 1970s the real wages grew essentially monotonically (overall, they grew by a factor of 3.5). Since then the wage stagnated (white-collar workers) or even declined (blue-collar workers). These are official statistics (BLS) and the actual situation must be worse, because the real rate of inflation is probably underestimated by creative folk at government statistical agencies. In any case, the cost of big items that define middle-class way of life – house, college education, medical insurance – have increased faster than the official inflation rate.

The implications are obvious, and it is surprising that they are rarely brought up in the context of massacres. As their economic prospects deteriorate, many breadwinners find themselves under unendurable pressure to maintain the socially expected level of consumption. Under these conditions people, whose psychological problems would be borderline in a gentler economic climate of the fifties, today ‘go postal.’ So the harsher the economic conditions, the greater the numbers of those whose latent psychological problems develop into full-blown psychosis.

But this is only part of the story, I said in that blog. This hypothesis may explain why work-place rampages increased, but what about those striking schools and universities? That I promised to explain in Part II of the blog.

However, Part II was never published by Freakonomics. The first part evoked a lot of responses from the readers of Freakonomics, most highly negative (take a look here). And Freakonomics canceled Part II.

I was a bit taken aback by such violently negative comments, especially those that attacked the empirical trend that I was discussing. After all, I wasn’t writing a scientific article, but a blog. And I could hardly present the methodology underlying my data in a 600-word blog (the methodology was later published in a peer-reviewed journal – this is the JPR article to which I referred earlier).

Most gallingly, on the same day one of the editors of Freakonomics (the same who on the next day rejected the second installment of my blog) published a ridiculous blog where he proposed that the reason we have a rampant obesity epidemic in America is because we switched from outhouses to internal toilets. Unlike mine, this blog was positively received!

I think that the violent reaction to my blog was triggered precisely because I am right. I admit, the overall message is not a cheerful one. And we all know what happens to messengers bearing unwanted news…  But this is my personal, biased opinion, so I am very interested to hear what others think.

So what did the readers of Freakonomics find to criticize in my blog? Here’s one response that encapsulates most of the objections (and does so in civil language, I might add):

I think the conclusions drawn by Prof. Turchin are questionable. First and foremost, a total of 7 cases per year in a nation of 300M cannot be considered to be significant, statistically or otherwise. These cases are very conspicuous, but quite rare. The increase can be easily attributed to the growth in the US population in the last six decades, or to the more graphic media coverage in recent years (no live TV coverage in 1950, I guess). These explanations are far more plausible, in my mind, than those suggested by Prof. Turchin. Also, he tried to explain acts of madness using economic reasoning, which, again, is not that convincing to me.

I’ll deal with two of these points today, and leave the others for the next blog. Let’s address the issue of population increase first. Even if the population of the US doubled over the last 50 years, the 14-fold increase (from 0.5 rampages per year to 7) would be halved to a 7-fold increase. This is not worrying?

In fact, the population did not double, and the total numbers of rampages increased even more than I had known in 2008 (now that I have data for 2009 and 2010).

Between 1960 and 2010 the US population grew from 180 to 310 million. During the 5-year period (because numbers jump a lot between years, it is better to average them over longer periods of time) that ended in 1960 there were 3 rampages, while in the 5 years that ended in 2010 there were 54. So if you do the maths, the increase in rampages per capita over this period works out to 10.4.

Next, there is no question that in terms of total deaths shooting rampages may appear trivial. Even during the worst (the last complete) 5-year period of time, 2006-2010, the rampages recorded in my database took lives of 210 people, or 42 per year. This is just one-thousandth of annual fatalities due to vehicle accidents.

But I never argued that killing rampages are significant in themselves. Each rampage is a tragic calamity for people directly involved, the victims and their family and friends. But for most of us at the aggregate level of the whole United States, of course, they will be just horrible stories that we read about in the newspapers.

The reason we should be worried about rampages, I argue, is because they are surface indicators of highly troubling negative trends working their way through deep levels of our society.


Back in the nineteenth century, when coal mining was a very dangerous profession (even more dangerous than it is today), miners took a caged canary down into mines, and watched them with anxious eyes. If the canary suddenly dropped dead, that meant that the deadly gas, carbon monoxide, was slowly seeping into the shaft, and it was time to run like hell.

An order of magnitude increase in killing rampages is like canaries suddenly starting to drop dead all around us. They are an early indicator that something is changing for the worse. They warn of the coming of greater danger, but they are not the cause of it.

Other blogs in this series

Canaries in a Coal Mine II. “We too are asking why”

Canaries in a Coal Mine III. Is the Trend Real?

Bryan Vila: A Criminologist Comments on ‘Canaries in a Coal Mine’

Canaries in a Coal Mine IV: Alternative Explanations

About these ads
Posted in: Blogs