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



Join 36.9K other subscribers

Hello Peter,

I’ve read the blog post and think it’s an interesting and well-presented idea. But before the evolutionary-ecologist/criminologist/historian in me could accept this as plausible, I’d want to account for three sets of causal forces. Here they are, in rough:

I. Theoretical construct: Opportunity/Routine Activity: Any type of criminal act will tend to increase with an increase in the probability that a motivated offender will converge in space and time with a suitable target — when there is no person or physical “guardian” to prevent a crime (e.g., cops, neighbors, physical barriers). Limits: This theory does not deal with the causes of motivation, just environmental and ecological variation.

Causal factor examples:

  • Change in population density
  • Change in routine activities
  • Change in physical structures
  • Change in number or attractiveness of potential targets

Hypothesis: As access to target populations in which a “killing rampage” can occur increases….

II. Theoretical construct: Rational Choice/Deterrence: People act in a way that they perceive as promoting their own self interest; behaviors defined as criminal often can be—or at least seem to be over a given time frame—a good way to get what one wants. Limits: Many people who commit crimes are not “rational” in that they don’t share the same frame of reference as the majority (think deranged killer, suicide bomber); or underestimate the odds/consequences of getting caught (think schmuck who embezzles 10% of his/her salary, any one who robs a bank).

Causal factor examples:

  • Social/cultural/biological/developmental factors that affect impulsiveness—the extent to which people discount potential time to punishment and favor immediate gratification
  • The stake a person has in conventional life
  • Impaired ability to calculate risks/benefits
  • Impaired ability to control urges
  • Changes in peceived probability of receiving consequences (e.g., if you’re dead, they can’t hurt you)
  • Changes in perceived time-to-receipt of consequences (see impulsiveness)
  • Changes in magnitude/perceived undesirability of consequences

Hypotheses: [follow from causal factors]

III. Theoretical construct: Transmissibility of Culture (i.e., contagion): All else equal, the relative frequency of ideas/behavioral strategies is more likely to increase when a larger proportion of people are exposed to them, or when the magnitude of exposures increases. Limits: Susceptibility, attractiveness, …

Causal factor examples:

  • Technologies make information about low base-rate phenomena more available
  • Information media place more emphasis on, say, dramatic but rare threats
  • 24/7 news cycle magnifies perceived importance and prevalence of rare threats

Hypotheses: [follow from causal factors]

It seems to me that these theories highlight several weaknesses in the leading-indicator argument being made in Canaries in a Coal Mine:

1. There have been massive changes in the structure and density of American society since I was born in 1947. These have had major impacts on crime in general. For example, U.S. Population has increased 231% from 1947-2006 (suggest changing data in your figure from frequency to rate). Average population density also has increased dramatically as modes and means of production and employment shifted—more people in cities increases frequency of interaction, target density, etc.

2. Dramatic increases in connectivity and connectedness of people tends to make people much more aware of mass killings, much sooner.

3. 24/7 news cycle makes recency of threat greater, constant repetition of announcements magnifies threat perception, dramatic presentation increases perception of threat magnitude, etc.

4. Historically, the increase in number of mass murder events that began in the 1980s coincides with dramatic changes in how the mentally ill were dealt with in the United States.  After decades simmering on the back burner, policies for dealing with the mentally ill rapidly shifted from institutionalization to community treatment.  This seemed enlightened, but had unintended consequences. When the forces of liberal deinstitutionalization and conservative Reaganomics converged, the result was a sudden decline in treatment delivered to seriously mentally ill people.  See Learning From History: Deinstitutionalization of People with Mental Illness As Precursor to Long-Term Care Reform (2007)

The two articles I’ve attached [to this e-mail] lay out the elements of an evolutionary ecological theory of criminal behavior (and provide cites). When published in the ’90s, they were heresy in the social sciences. But I think all of the major connections I hypothesized have been supported in recent years.

Kind regards,

Bryan Vila, Ph.D.
Professor of Criminal Justice and Criminology
Washington State University – Spokane

Notify of
Most Voted
Newest Oldest
Inline Feedbacks
View all comments
Peter Turchin


Thank you for these thoughts and critique. The conceptual framework you have developed makes a lot of sense to me (but I am not a criminologist).

I respond to your main points below. However, keep in mind that indiscriminate mass murder (IMM) is a type of political violence (suicide terrorism, if you accept my argument). This means that some elements of your framework will be inapplicable.

I. I don’t think there was any substantial increase in the opportunity to commit IMM. Yes, the US population has grown, but it’s not necessarily concentrated more than 50 years ago. In fact, as a result of the movement to suburbia, it may be less concentrated now. Additionally, note that the graph of IMM incidence scaled by population continues to increase, indeed accelerates, during the 1990s and 2000s. I am not aware of any dramatic changes in the opportunity over the last two decades. Finally, a very substantial proportion of IMMs takes place in rural areas or small towns, such as Newtown.

II. The Rational Choice perspective is particularly unhelpful when dealing with acts of suicide terrorism, which are ultimately irrational acts (‘irrational’ here is used in the narrow sense of the Rational Choice Theory). Just about the only factor I can think of that would be relevant in this category is “the stake a person has in conventional life,” but that actually supports my argument.

The theoretical construct that is a much better fit to the situation is ‘moralistic punishment,’ which is, as we all know, a completely irrational act (again, within the framework of rational Choice Theory). Many indiscriminate mass murderers see themselves deeply wronged by the firm, the school, or the society as a whole, and they strike back in revenge. Their rage against these groups or institutions is an almost ubiquitous motive, at least in the cases where the perpetrators survived to tell why they did it (or left suicide notes). While moralistic punishment plays a very important role in maintaining cooperation in healthy societies, in this case it is, of course, completely destructive. This is the dark side of moralistic punishment.

III. This category engenders one of the most common alternative explanations. I actually agree that there is probably a feedback loop connecting extensive media coverage and increased frequency of IMM. Certainly, IMM has entered the American cultural repertoire. Populations in different countries tend to use culturally specific forms of political protest and violence. For example, self-immolation is not part of American cultural repertoire, but shooting rampage is. Shooting rampages were very rare prior to 1965 (although I have at least one clear case going back to 1900).

But while this factor may be operating it is not really an alternative hypothesis to the one I argued for. When discussing outbreaks of political violence, I found the forest fire analogy useful. As time goes along, forests tend to accumulate deadwood – fallen trunks, branches, etc – that can serve as fuel for a forest fire. Sooner or later there is a precipitating event (a careless match, a lightning strike) that starts the fire. Initially fire expands – the more wood is burning, the more new wood gets ignited. This is the process that is analogous to how a cultural fad develops.

However, I am interested in structural conditions that make a fire possible (the amount of flammable material, so to speak). So the two explanations are complementary, they are not alternatives.

A couple of other points. Cultural fads tend to develop – and burn out – on much faster time scales than the dynamics depicted in the figure in the previous blog. You cannot really have a cultural epidemic developing very slowly over the period of 50 years. So there must be some structural conditions underlying it.

Second, in addition to positive feedbacks that cause fires to burn at an accelerating pace, there are also negative feedbacks that eventually bring them down. An example is the backlash against extremist American groups brought on by the horrible tragedy of Oklahoma bombing. A number of such organizations have folded in the late 1990s, and the overall number decreased (but then started increasing again after 2000).

1. I have already addressed this point above in I. And I have changed the data to a rate, scaled by population. As you can see, the overall trend (10-fold increase) is quite significant. Again, a lot (perhaps a majority – I need to do calculations) of IMMs happen outside the big cities.

2. This is the question of awareness. But as I showed in my previous post, while awareness may have increased, the underlying frequency increased much faster.

3. Again, this deals with the perception of the threat. Although, I might add, when you have a 20-fold increase (in crude numbers), perception had better also increase – this is the reality of living in today’s America.

4. This an important alternative hypothesis. I don’t think it works, but I will need to develop my reasoning at length (in the next blog).

SEF Editor

The discussions over the past few days have been very informative, looking at causal factors leading to what our society regards as irrational, unacceptable, and unthinkable acts. In my opinion, what has not been exposed in sufficient detail is the fact that despite > 200 million firearms in the USA, events such as Sandy Hook are extremely rare. How can we predict when and where such acts could occur in the future? I argue that we need both statistical and causative approaches.

First, we need to know what the severity distribution of shootings is. Is it “fat tailed”?

Second, when examined through time, do these events show any form of epidemic behavior (e.g., Clauset and Woodward. 2012. Estimating the historical and future probabilities of large terrorist events. arXiv:1209.0089v1)? Or rather, do they show random distribution or even under-dispersion?

Third, as highlighted in the discussions at the Social Evolution Forum, what are the predictive abilities of causative factors, such as media effects, fire arm control, health care policies, etc.?

Identifying both the statistical patterns and causative factors is in my view the most productive route to a “seismic approach” to evaluating risks and developing adaptive approaches to reducing the use of fire-arms, and eliminating future Sandy Hooks.

Michael Hochberg

Peter Turchin

Mike, I think that we cannot predict where and who will be the next perpetrator, but we can predict the overall dynamics. Unfortunately, structural-demographic model suggest that the incidence will continue to trend up.

In my next point I will present answers to some questions (but not all) that you raise.

  1. Home
  2. /
  3. Cliodynamica
  4. /
  5. Guest Posts
  6. /
  7. Bryan Vila: A Criminologist...

© Peter Turchin 2023 All rights reserved

Privacy Policy