Canaries in a Coal Mine IV: Alternative Explanations



Join 36.9K other subscribers

This is the fourth and last installment in this series. To tell the truth, I will be glad to be done with it because shooting rampages is an inherently depressing subject, in more ways than one. However, it is also an important one. Today I need to review the alternative explanations of the pattern I documented in the previous installments.

First, a recap. I have argued that indiscriminant mass murder (IMM) has been seriously misinterpreted. It is not just another type of crime, but a form of political violence. More precisely, suicide terrorism. It is not motiveless. But the motive is not to kill a specific person, or persons, but to strike at a social group, at an institution, or at the society as a whole. ‘Indiscriminant’ is the key word coding for the Principle of Social Substitution.

Rage is a common, perhaps universal component of the motivation leading to an IMM attack. We still don’t know what motivated Adam Lanza to commit this horrible act, but we know that he was driven by rage. The six-year old girl, who survived the massacre by playing dead, described the shooter to her mother “as a very angry man.”

The rampage shooters see themselves as moralistic punishers striking against deep injustice. In a perceptive opinion at New York Times, Adam Lankford writes, “we should think of many rampage shooters as nonideological suicide terrorists” (I would remove ‘nonideological’ because many such killers in my database were ideologically motivated). He then points out that a common factor in both rampage shooters and suicide terrorists is “a deep sense of victimization and belief that the killer’s life has been ruined by someone else, who has bullied, oppressed or persecuted him.” I would add that this ‘someone else’ does not need to be a person (a point that Lankford acknowledges elsewhere in his opinion). In fact in the case of IMM (with an emphasis on the I), it is usually a group, an organization, an institution, or the whole society that are held responsible by the killer.

I further argue that the frequency of IMM depends, in the first place, on environmental conditions. As the degree of cooperation in the American society declined over the last four decades, and the degree of intrasocietal competition rose, increasingly large numbers of susceptible individuals were victimized, bullied, and oppressed, and a certain (very small) fraction of them chose to become mass murderers to avenge such injustice.

The fundamental forces underlying this environmental change have been two structural-demographic trends – popular immiseration and elite overproduction (see my article on the Conversation). The first was correlated with the deterioration of working conditions, the second with the growing social pressures on the campus, which is why the two most common settings for shooting rampages are the workplace and the schoolyard. Overall, as the level of cooperation within the society decreased, social competition, political polarization, the dog-eat-dog economic climate, and the general level of nastiness increased (see my blog on this topic).

As I am quite sure that some will willfully misread what I am saying, I want to add that while an environmental change may explain why the frequency of IMM increases, it does in no way excuse them. Mass murder is a horribly inappropriate response to even those situations when injustice is real. Mass murderers are insane, evil, or both.

Now that I have summarized my explanation, let us consider alternatives.


I don’t think that anybody seriously contends that the reason why IMMs are increasing is due to changing patterns of gun ownership. The evidence is overwhelmingly against this explanation. During the last couple of decades a series of laws have been passed that impose more and more restrictions on the purchase of powerful firearms. Partly as a result of these laws, but probably having to do more with cultural change, the proportion of Americans who own guns have been steadily declining. Here is a graph from a recent New York Times article:


As we can see, gun ownership rates have been declining across the political spectrum, although naturally the Republicans lagged behind the Democrats. Since the trend in the firearms runs counter to the trend in the frequency of IMM, it does not help to explain why the latter have been increasing.

This doesn’t mean that the proposed measures of gun control will be ineffective in reducing IMMs. It seems quite possible to me that if we were to ban all firearms completely, then there will be fewer IMMs. It won’t change the underlying environmental change that explains the motivations of potential rampagers, but it will surely reduce their ability to inflict damage on the society. In the absence of guns they will have to turn other means of inflicting casualties – constructing bombs, using knives, axes, bows and arrows, cars, and planes. But all those methods of inflicting mass homicide have problems – they are more easily detected by the police, or they inflict lower casualties. So in the absence of guns it stands to reason that a higher proportion of killers contemplating IMM will be arrested before they commit murder, or if they go on rampage, kill fewer people, or even choose to just kill themselves (perhaps in the form of ‘suicide by-cop’) instead of committing murder-suicide. Overall, while these measures will not address the conditions causing the increase in IMM, they will surely reduce the number of innocent people killed.

Mental disease

This is one of the most common explanations for the increased incidence of IMM. Here’s an article by Clayton Cramer that lays out the logic in a very clear way (thanks to the reader lrb for supplying me with this link).

Cramer says, “At least half of these mass murderers (as well as many other murderers) have histories of mental illness.” The problem with this statement is that it either goes too far, or not far enough.

It does not go far enough because can you imagine yourself shooting in the face a man or a woman who is begging you for mercy? Killing your mother? Shooting a six-year old with an assault rifle 11 times? Things that indiscriminant mass murderers do go way beyond the pale, as far as normal human beings are concerned. The overwhelming majority of humans are incapable of treating other human beings in this way. Perhaps only 2 or 3 percent of human beings are ‘sociopaths’ who lack such inhibitions and to whom killing another person is no more stressful than stepping on a cockroach. The Army recruits their best snipers from this small minority.

On the other hand, Cramer goes too far because there are literally millions of people in the US who suffer from some kind of mental problem (“in the early 1980s, there were about two million chronically mentally ill people in the United States”). Nevertheless, only a tiny minority of them commit mass murder. In the retrospect, we say – aha! – Adam Lanza had been diagnosed with Asperger’s syndrome. This explains why he murdered 27 people including 20 children and then committed suicide.

It actually explains nothing. This is the first time I heard of Asperger’s syndrome, and I had to look it up. It turns out that it’s a mild form of autism, and apparently there is no connection between this condition and an increased propensity for violence. Check out the New York Times opinion by Priscilla Gilman, Don’t Blame Autism for Newtown for more details.

If it could be shown that the number of mentally ill people increased in parallel with the frequency of IMM, that would indeed be a viable alternative explanation. Many, including Cramer, point to the shift in the treatment of mentally ill during the Reagan administration. But the curve of IMM incidence doesn’t show a threshold-type behavior we would expect, if this were the primary cause.

As my fellow editor Mike Hochberg pointed out, where are the data? Can somebody show me how the numbers of chronically mentally in the US changed over the last four decades?

Furthermore, Cramer points to the observation that the homicide rate in the US declined during the 1990s, and that it coincided with the so-called “incarceration revolution.” The implication, as I understand, is that a lot of criminally insane were locked up and the result was decreasing murder rates. Fine, but the incidence of IMMs actually accelerated during the 1990s.

homicidesMore generally the divergent trends between homicides and IMMs presents a serious difficulty to any alternative explanation. If IMMs are simply another type of homicide, why did the overall murder rate declined, starting in the 1990s, while IMMs kept on trending up?

It gets even more interesting. According to Grant Duwe, who specializes on the study of mass murder, and whose work I have already referred to earlier, they actually declined after 2000:

Grant Duwe, a criminologist with the Minnesota Department of Corrections who has written a history of mass murders in America, said that while mass shootings rose between the 1960s and the 1990s, they actually dropped in the 2000s. And mass killings actually reached their peak in 1929, according to his data. He estimates that there were 32 in the 1980s, 42 in the 1990s and 26 in the first decade of the century. (from here)

Note that there is no contradiction between Duwe’s data and mine because we use different criteria and thus count different types of events. Duwe counts any murder that caused 4 or more fatalities, including criminal and family disputes, while my primary criterion is indiscriminancy, meaning the application of the social substitutability principle, and at least one fatality.

But the divergent trends between homicides and IMMs are a serious problem for any alternative explanation. It is not a problem for the structural-demographic model because it posits that homicides (crime) and political violence are distinct conceptual categories, and it is not necessary for them to move together. While they share some of the causal factors, other drivers are different. In historical data for the societies that I studied homicide rates and political violence rates do not necessarily move in parallel. Whatever reasons caused the homicide rate to decline in the 1990s, the structural-demographic processes of popular immiseration and elite overproduction have continued to trend up, and so we expect that political violence should have also gone up (which it did).

Cultural contagion

I have already discussed this alternative explanation in my response to Bryan Vila, but let’s recapitulate the main points here.

1. I agree that the media coverage plays a role in the epidemic of IMM. However, this mechanism works together with the structural-demographic explanation. It is not an alternative hypothesis. In my analogy with a forest fire, structural-demographic conditions tell us whether there is enough flammable material, while cultural contagion mechanism explains how a spark that falls on the flammable material develops into a conflagration. Both are needed.

2. Cultural mechanisms are particularly important in determining how forms of political protest and political violence are ritualized. Which forms become part of the cultural repertoire of any particular social group.  So when a Tibetan monk wants to protest oppression and injustice he goes to a public place and sets himself on fire. When an Arabic youth wants to do the same she straps on a suicide belt, goes to a public place, and blows up herself and whoever is nearby. When an American wants to do the same, he arms himself to the teeth, goes to a public place, and murders as many people as he can shoot until he is killed by the police.

Does economic reasoning explain acts of madness?

Finally, I return to this objection to my original Freakonomics post (see it in the first installment): “Also, [Turchin] tried to explain acts of madness using economic reasoning, which, again, is not that convincing to me.”

By this point it should be clear that I certainly do not explain IMM incidence by any kind of rational choice mechanism, which includes economic reasons. Murder-suicides are quintessentially irrational acts. As I explained earlier in this post, economic conditions, however, can change the environment in ways that may elicit more, or fewer such acts.

It is ironic that my critic posted his disagreement in the Fall of 2008, before it was widely appreciated how serious the Great Recession was going to be. This severe economic downturn, then, can serve as a natural experiment testing the hypothesis that worsening economic conditions should result in an increased incidence of IMMs.

Here’s what the data say:


As we can see, the incidence of IMM tends to jump during years when male unemployment grows. In 2009, which was probably the worst year of the Great Recession, there were 21 rampages – more than during the entire two decades of the 1950s and 1960s (16 rampages).

Notify of
Most Voted
Newest Oldest
Inline Feedbacks
View all comments
Gene Anderson

Fascinating and mostly convincing…. However, while gun ownership is down, regulation of assault rifles has ended and there are many more of them out there. As to mental illness, at least some of the shooters are clearly schizophrenic; apparently poor Adam Lanza was (I don’t believe the “Asperger’s” thing except in so far as it’s a catchall; it’s way overdiagnosed). I think there is a lot of copycat crime here–school shootings are all the rage among the enraged.

Peter Turchin

As I said in the post, I think that both copycat dynamic (what I called cultural contagion) and the underlying fundamental shifts are in play. This is probably why the graph of IMM incidence is so ‘spiky’ – we get outrbeaks of rampage shootings, then things subside, then another outbreak. But the reason we have one outbreak after another is not due to cultural contagion, it is due to the underlying conditions that change on the long time scale.

Michael Hochberg

Peter, a more parsimonious explanation (that I mention in my reply to your last blog) is probabilistic, with no defining cause. That is, the causal reasons for any given rampage may fit the different mechanisms you elucidate, but they do not really explain “why” (that is, predict), because there are millions of people that can potentially commit an IMM, but only an infinitesimal fraction ever do so. So, limiting firearms and increasing mental healthcare — if done in selective ways as per the causal mechanisms I mention in my last reply — will OBVIOUSLY reduce the expected incidence of IMMs, but will not effect the stochastic dynamics. That is, little or no prediction. Indeed, your last graph above does not suggest a marked pattern, or even a clear trend.

This is the kind of data we need:

We also need temporal data on severity. Is it epidemic? Random? Underdispersed? This may also give clues as to whether the putative “causes” have any predictive value. It could very well be that one large event is associated (either precedes or follows) many more minor events. This is actually NOT inconsistent with under-dispersion of IMMs (that is, the population is sensitized to the tragedy of an IMM each time one happens, and time must past before this wears off), or that IMMs are reactions to nebulous social, political or media “climates”.

Peter Turchin

Mike, what do you mean by ‘probabilistic’? It depends on what aspect of data you look. Each shooting rampage is a highly stochastic, and probabaly completely unpredictable event. But in the aggregate there is a definite trend. Look at the other graph that takes a longer view. There is a definite (and stastically significant) inicreasing trend.

It’s like incidence of suicides. Each suicide is an act of free human will, and is essentially unpredictable. But in the aggregate the rate of suicide changes gradually and in predictable ways.

By the way, in my JPR paper I analyzed the severity of political violence incidents in my database (severity = number of people killed). The distribution is fat-tailed, and a power law provides a good fit to the data. Check it out in the article.

Michael Hochberg

Probabilistic is what is the probability of an event of severity x happening in a future, unit time frame (e.g., per week).

Peter Turchin

Well, in that case, my prediction is that the probability of IMM will not decline, and may well increase. From an already very high level. Just earlier today there was another shooting rampage. I don’t yet have enough info to determine whether it was an IMM. But if yes, it will be the third this month. I haven’t yet collated the data for 2012, but we are approaching the record levels of 2009.

Juan Alfonso del Busto

Congratulations. I think you have captured all the posible explanations. I think that the economic causes explained in the first post are very interesting.

I would like to add that very little attention is given to what is usually called “spiteful behaviors” in behavioral ecology. These behaviors are not necessarily irrational. As a matter of fact they can be very rational: Natural selection only “sees” differential success, so it doesn´t really matter how you succeed, whether by being better than the average or making sure that the rest fare worse than you.

Normal people have many alternative courses of action and choose the one with the best Risk/Benefit ratio. But, what if we think we are worthless and we don´t have alternatives. What if we assess our only options are 1. to dissapear or 2. to behave spitefully against the ones that compete with us for the same resources?

Of course this doesn´t make any sense from an individualistic perspective, but this changes if we take into account the kin and/or the group. If I dissapear (commit suicide) I let more resources available for my family/group. On the other hand, if I murder all the posible enemies of my family/group before killing myself I am enhancing my family/group fitness relative to the enemy family/group. Viewed this way the saccrifice makes sense.

I think these mass murderers feel they are part of a group: the “oppressed” people like themselves. New technologies facilitate this very much. The so called “Conspiranoia” is infecting the infosphere. These people think that other mass murderers, like the Utoya island killer for instance, are kind of heroes. And they know they are not the only ones that feel that way. They want to become heroes too and, since they think they don´t have many chances, they choose this radical course of action, which is preferred to lonely peaceful suicide.

The ultimate explanation of these terrible events could be found in the evolution of spiteful behaviors by natural selection. These spiteful behaviors can be similar, if not exactly the same thing as the moralistic punishment and fits with the other explanations.

Peter Turchin

Juan, what is the difference between ‘spiteful behaviors’ and moralistic punishment? When ‘spite’ is associated with an intention to punish someone who transgressed on a social norm, it is one and the same.

Juan Alfonso del Busto

Whenever I hear of spiteful behaviours it is as “the dark side” of cooperation. It is kind of the opposite of cooperation: paying a cost to harm someone. But a lot of attention is paid to cooperation and very little to spiteful behaviors.

In my opinion the difference between spite and moralistic punishment is that spite is a competition-driven force whereas moralistic punishment is a cooperation-driven force. You need to be part of a group to deliver moralistic punishment on the members of your group, but you don´t need to be part of a group to behave spitefully. In other words moralistic punishment is delivered on defectors inside the group but spite is applied against competitors outside the group or even inside the group. We could even see war as a cascade of collective spiteful behaviors between groups.

May be these mass murderers aren´t delivering punishment inside their group but rather they are trying to reduce the fitness of a rival group at a cost to themselves but with a RELATIVE benefit for their group (whichever they think it is), because this is what spite has evolved for: enhancing differential success or fitness between units of selection even if it reduces the overall success or fitness.

Juan Alfonso del Busto

Inside a group, members who behave spitefully increase their individual fitness relative to their competitors but, by doing so, they reduce the fitness of the group… even more than the passive defectors or free-riders do. These behaviors deserve punishment that should be delivered by the group because if it is delivered by the harmed members of the group, How can you tell if it is retaliation, spite or moralistic punishment? Since we can expect that spiteful behaviors and free-ridings are going to take place inside a group there must be a way to minimize the group fitness reduction which is one of the uses of moralistic punishment, but, in order not to initiate a cascade of retaliatory acts, this punishment must be delivered collectively, which could be one of the origins of institutions.

Peter Turchin

I see your point. Spiteful behavior is then rational in the sense that it maximizes your relative personal personal fitness (by lowering that of everybody else). Moralistic punishment, on the other hand, lowers your personal fitness with respect to others in the group, while raising everybody’s fitness in the group by contributing to cooperation. This is a good example of how multi-level perspective is useful to tease apart fitness consequences of various kinds of behaviors.

Juan Alfonso del Busto

Peter, that´s exactly my point. Indeed multilevel selection can be a consequence of the complex and dynamical interaction between spite and cooperation at certain levels of selection since both can contribute to increase relative fitness in different circumstances. Thank you.


What explains the trend that all the shooters are male vs female. If raging against the “system” and society are some of the strong drivers for IMM (along with bullying, etc) why do we not see female shooters? With relatively high rape statistics in US, continuing lower pay, and high unemployment rates and/or prospects, there are perhaps more triggers to set off such a backlash at society, so why do we only see male shooters?


Based on the statistics and explanations presented here, one would expect most of the IMM to be committed by women. Women compose 50.8% of US population (2010 census), are politically aware and vote (2008 voter turnout was 65.7% of women voted vs 61.5% of men), continue to earn less than men (Bureau of Labor Statistics for 2009 show women labor force participation at 59.2% and earning less with the ratio of women’s to men’s earning at 81.2%). I would also assume the unemployment rate among women is also higher than men. Levels of violence against women are also high. All of this, using a previous example, should create a very dry forest sprinkled with lighter fluid and ready for a spark. Yet, it does not happen. The IMMs are overwhelmingly executed by men.
What is your take on this, as statistics should show that more than half of the politically or lashing out at society IMMs would be expected to be executed by women?

Peter Turchin

Actually, a small percentage of IMMs are committed by women. But you are right, the overwhelming majority are the work of men. I don’t have a very good explanation, apart from what Juan has already said.

Just keep in mind that I am interested in explaining why this type of political violence increased so dramatically. The fact that it is perpetrated by psychotic males armed with guns is one thing, but why they are much more likely to go on rampage today, compared to 40 years ago is what I aim to understand.


Most IMMs are by males for the same reason most homeless people are males. Society’s biggest losers tend to be men. Women tend to be provided by society with a better safety net than men, so they don’t get as angry at society, and don’t feel like the only outlet is violence…

Juan Alfonso del Busto

In the natural realm physical violence is mainly a male thing. There are good reasons for this: since males have a extremely high variability of reproductive success relative to female reproductive success, competition among males for access to females drives a kind of sexual selection, called intrasexual selection, that explains the arisal of sexual dimorphism in certain species. Namely males in species that are polyginous and with intrasexual competition are phisically stronger, bigger and more aggresive than females. In part we humans are in that lot and, as a matter of fact, in those societies that are polyginous, what means that low-rank males get no chance of getting a wife, violence, rapes and aggression are much more frequent than in regular monogamous societies.

We, males, have a tendency to exhibit physical power and aggresiveness both to achieve a greater status and to impress the ladies. Of course we males try to justify this violence and dress it like “heroism”,,, and often females buy it (otherwise this behavior would have never been evolved).

As a consequence males could have the instinct to behave extremely “heroic” when they feel other males have gathered all the mating opportunities and all the status positions, precisely because they don´t believe they have anything to lose.



Franz von Stuck. Battle for a Woman. Date: 1905

Nathan Dean

Another aspect to look at is the mind set of the groups targeted by the mass murderers. It would seem to me as more society becomes soft the more susceptible it will be to these IMMs. Drops in gun ownership, the avoidance of death penalties, and the more entertainment based society are all signs of a softer mindset through the people in the United States. This in turn gives more courage to the murderer to commit his act for whatever reason.

People are bombarded with news of these kinds of killings and society begins to worry and fear when the next one will occur. They see the disasters of the previous ones and if they are put into the same situation they will be frozen with helplessness. Helplessness is what may add to IMMs.

An example to bring up, would be current school drills for an intruder. Students are instructed to crouch in a corner of the room away from all the windows, with the lights off, the door locked, and everyone silent. I feel this is no different than telling students to get on their knees and face the wall ready for execution. The lights should be off and the door should be locked but I do not believe the huddling in the corner will help a thing. If the intruder were to break into the room, everyone would be in one area where he could easily fire upon and few would be able to escape because of the confusion with the cramped conditions.

Instead the drill should prepare students to fight back. The door should not just be locked but barricaded with desks and what have you as well as people ready to ambush the intruder if he were to enter. They still must stay away from the windows but in a way where they would be able to do something about an attacker. A mass murderer may think twice or quickly be subdued if a whole school was willing to fight back instead of kneeling in the corner like cowardly chickens.

Another consideration is that a few of these killers are partially driven by the power they feel over someone else’s life. They get an ego trip from committing these atrocities. However, if people show little fear to the would-be killer he may not have such a feeling of power and in turn not even commit the act.

Society must drop its assumption of helplessness and learn that with cooperation they may bring about a significant amount of help and save dozens of lives.

This all does not necessarily apply to kids so young like in Sandy Hook but a revamping of society’s mindset through middle schools, high schools, and colleges should reduce the chance of the targeting of any school and hopefully bring a drop in IMMs.

  1. Home
  2. /
  3. Cliodynamica
  4. /
  5. Regular Posts
  6. /
  7. Canaries in a Coal...

© Peter Turchin 2023 All rights reserved

Privacy Policy