Back when I was on the University of Idaho football team I had the most interesting end to a conditioning workout imaginable. We’d just finished mandatory offensive line summer conditioning — about 16 of us 6’2″-6’8″ 280+ pound gorillas sweaty with our shirts off. Returning to the locker room in the Kibbie Dome we heard hip-hop music thumping inside the dome and saw flashing police lights. Like moths to a flame, we wondered in to see what was happening. The Dome floor had several police cars, a large military enforcement vehicle, blockades, and a line of about 50 police officers in full riot squad gear like those in the picture above. As we were standing there, an older guy yelled, “Hey! You guys want to throw shit at cops??” — moths to a flame.
It was a practice for the Quad Cities riot squadron. Two of the senior organizers were trying to behave like a disobedient riot and had evidently forgot to invite anyone else. The line parted and about 10 of us crossed to the “bad side”. Knowing that this could get ugly, but was going to be an experience, I made sure the two senior organizers were throwing stuff too. Surely the riot squadron wouldn’t injure one of their senior commanders?
We started grabbing tennis balls, garbage cans, barriers, anything at hand became a resource to throw at the line. It was fun! And they were trained! My old trick in snowball fights is to underhand one high up into the air and then nail somebody in the face as they watch the first fly up — not this group. The line, following commands, made steady, even marches down one side of the floor to corner us. We improvised blockages of various forms to disrupt the line. Eventually, one of the senior organizers formed a ring circle with us. About 8 of us sat down in a circle with our backs to each other and locked our arms and legs. The approaching line halted — some murmuring discussions — then heard the tinkling aluminum tear gas cans as they fell in the middle of us and the area around us. Not expecting this, I looked up at the senior organizer, he looked back and said, “we only had so many fakes, one of those is likely real!!!!” Instantaneous disbursement.
Sensing the game was coming to an end, one of my biggest friends decided to test the fortitude of the line. He got about a 10 yard run. As he approached the eyes went up towards his. The “target” of the line, sensing the need to act, took his shield and checked my friend across the head and chest, flat-backing him on the Kibbie Dome floor. My friend was ok, but that ended it. I left with a renewed confidence in the training of our regional riot squad. Later I found out that they were well trained, even in our small communities of less than 40,000, for a good reason. Just three years prior, about 500 WSU students had rioted after a ban on campus drinking, injuring 23 officers and landing it on the list of top 25 college campus riots.
Looking at a list of human stampedes (often associated with riots) you’ll see that the frequency, at least of documenting these events, is increasing. A stampede, crush, or riot strikes a chord with many of us, as it’s not something external, but ourselves that cause the damage. One of my friends, Kshitij Jerath, experienced a serious riot first hand during his youth in India and was traumatized enough to shape a life goal to analyze the swarm dynamics of riots. Kshitij is going to make serious contributions in this area — he actually analyzes event horizons in swarms. I can’t analyze how the event will actually play out like Kshitij’s phase-field modeling can. All I’m doing here is establishing when the properties of a group reach a point where a riot could spontaneously occur. It’s a phase change problem and our theory of Social Thermodynamics is becoming able to solve. This analysis will likely help provide properties of the social system to work with a phase-field analysis that could actually model point locations of a riot.
When I looked through the list of human stampedes one event in particular stood out: the Nigerian Immigration Recruitment Tragedies of 2014. Nigeria has a very high unemployment rate near 25%. The Nigeria Immigration Service opened 4000 jobs and held crowd interviews. Hundreds of thousands (the wiki estimates 6.5 million) people showed up to be interviewed in sweltering heat. Crowds became frustrated and unruly, and riots/stampedes occurred in at least six of the locations leading to at least 16 deaths. What is interesting about this case is that riots occurred in isolated and separate locations that were under similar conditions. This is a strong indicator that riots are not a random problem, but a statistical problem that naturally happens when the conditions are right.
Spontaneous Disassembly — The engineering definition of an explosion
Riots and stampedes are a phase change problem. Social Thermodynamics analyzes phase change in social systems using the Gibbs Energy:
g = u + Pv -Ts
where u is the internal energy or values you’re bringing to the problem, P is pressure or stress, v is the inverse of density, T is temperature or resources, and s is entropy or empathy. If the change (g2-g1) in Gibb’s energy is negative (the value for g is less after something happens then before) phase change will spontaneously occur. In other words, if g2 is less than g1 a riot could happen at any moment and just needs a catalyst/initiator. In general, phase change happens if the u + Pv terms reduce, and the Ts term increases.
The riot/stampede problem has very similar boundary conditions as our Social Thermodynamics: Explaining the Bubonic Plague and Renaissance problem: you have a high density of people in a condensed phase and a change in conditions (plague or panic) causes people to move out of the system. In the case of the plague this change happened relatively slowly over the course of several years such that a standard phase change from liquid to vapor and associated thermal wave could account for the relatively modest changes of temperature and pressure that would occur. In the case of a riot/stampede the change in conditions happens very rapidly on the scale of minutes to hours to halve or completely dissolve the density, which is much more in line with the engineering definition of an explosion: spontaneous and rapid disassembly.
There are two types of explosive phase changes that are important to differentiate: detonation and deflagration. A detonation is a pressure driven shock wave that causes the reaction to occur and moves through the explosive material close to the speed of sound. A deflagration is a thermal driven combustion front moving through a flammable material over a much larger area than a detonation. Detonations go bang. Deflagrations go whoomf.
Detonation: When human stampedes are likely to occur
In the case of a stampede, the detonation analogy is much more relevant. You have a crowd of tightly packed people (similar to a solid state as they have limited ability to move), these people are under pressure, and then suddenly a gun shot or scream cries out that sends the information through the crowd that somebody is getting hurt or killed. The values of the people in the crowd immediately drop into a survival state and the need to get back to your own comfortable space (gaseous or liquid state) is paramount but highly restricted by the rate people can move away. At the speed of sound this wave moves through the crowd which starts to run or move as fast as it can back to a comfortable density. In the case of a detonation the immediate change in phase from solid to liquid or gas takes a considerable release of energy that gets converted to temperature and pressure. In the social case, the pressure/stress is real, but the question is where the temperature/resources comes from? I thought back to my riot experience in the Kibbie dome, when you’re functioning in that survival mindset ANYTHING can become a resource (we were throwing the Kibbie Dome’s garbage cans!) In the case of a stampede, any tree, light-pole, truck to climb on can become the key resource to deal with the value problem of not being trampled. So the conditions where a stampede could happen at any moment:
- huge drop in values from u1 to u2, associated with people trying to suddenly survive at all costs, often initiated by a gun shot or catalyst.
- high packing density that has limited ability to initially change from v1 to v2, associated with the people gathered in a tight area, but eventually increases dramatically as the crowd disperses.
- huge drop in pressure/stress as people spread out to safety P1 to P2.
- low empathy for fellow crowd members going into the situation s1, resulting in a much higher empathy for the people who were hurt or killed s2.
- low resources T1 going in, but everything/anything becoming a resource on the way out T2.
Remember that all of these properties are related by an equation of state, if you change the pressure the density and temperature/resources will change, I haven’t developed that surface yet. It actually will make this analysis much easier as it will no longer be arbitrary how the properties change together from state 1 to 2.
Deflagration: When riots occur and how to diffuse them
Riots are more in line with a deflagration and are subject to the availability of fuel and how confined of a space the reaction is occurring within. Hence the time scale of a deflagration/riot should be longer than a stampede and is subject to the continued fueling of the rioters. The trends in the properties are generally similar with some small differences, particularly in the value term. Riots are still low empathy/resource crowds condensing together with a specific value/purpose in mind. However the need to disperse for survival is ancillary to the value of causing change and having your desires be expressed at all costs. Often, the values are socio/economic inequality (low Ts trying to become higher Ts).
Diffusing a riot situation like this is all about facilitating the phase change process in a slow and controlled fashion. Basically increasing the values (nobody needs to get hurt, we’ll work through this), decreasing pressure and density (let’s all take a deep breath and have a seat), and increasing resources and empathy (we have a meal cart coming over, let’s all sit down and talk through this). It’s the very same tactics advocated for hostage negotiation by Chris Voss in his book “Never Split the Difference: Negotiating like your life depends on it.”
The value and empathy levels going into the riot/stampede are key to whether it ends badly or not. I’ve participated in friendly social statement marches, like the Women’s March and the March for Science this year. We as a society know what a phase change looks like and how we create the conditions for it to happen. We’re just not aware that we are actually following the very same thermodynamic phase-change processes all around us in nature! I alluded to this in my very first post on Social Thermodynamics — whether you go through a disruptive phase change or not is directly related to the values, empathy, and resources you bring to the problem. It’s important to remember that we’re using Spiral v-Memes as our non-arbitrary value taxonomy analogous to energy levels/modes.
What doesn’t work
The command and control approach shown by the riot squad members at the top is the non-empathetic authoritarian/legalistic way to control stampedes and riots. We can see from this Social Thermodynamics analysis that approach is more likely to exacerbate the situation than to help. At the most it simply could control the situation long enough to spread information/messages to try to build empathy/values. It’s generally not a good idea to try to control/cap off an uncontrolled reaction.
What might be even more of an issue is the leading research in this area. Malcolm Gladwell recently advocated for Mark Granovetter’s “Threshold” models for riots and change in his revisionist history podcast. Granovetter is a Professor of Sociology at Stanford that published the seminal article in this area in 1978, “Threshold Models for Collective Behavior” which Google says has been cited 4436 times! (While it may seem like I go after Stanford researchers frequently, I assure you I have no problems or connections with the institution, it’s just a coincidence.) The first two sentences of his abstract read:
“Models of collective behavior are developed for situations where actors have two alternatives and the costs and/or benefits of each depend on how many other actors choose which alternative. The key concept is that of “threshold”: the number or proportion of other who must make one decision before a given actor does so; this is the point where net benefits begin to exceed net costs for that particular actor.”
I find it challenging to believe that when getting en-trained in a human stampede most are likely to consider whether to run or not. The Spiral v-Meme value taxonomy predicts that binary decision processes play a big role in authoritarian environments, which are likely sources for riots — you’re trying to decide whether to follow the regime or oppose and the regime is trying to decide if you’re the scape-goats or the golden children. This is likely why Granovetter chose riots to apply his model to and why it has received so many citations. At any rate, we only have limited free will to decide within the bounds of thermodynamic law.
Whether we participate in a riot or not is a statistical problem subject to the values we are employing based on the boundary conditions. Our brains are wired to understand the thermodynamic processes that shape the world around us and we inherently know how to drive phase change, whether we’re aware of it or not. Because Social Thermodynamics has an embedded value taxonomy, it applies regardless of the values you’re approaching a phase change problem with. We just now have the ability to predict when things transition from a peaceful march to a riot or stampede.