Game Theory on the “Second Civil War”

A Strategic Path to Avoidance Flows from Analyzing Outcomes (also we trash on Vox a bit)

This article was originally posted on Medium July 6th, 2018, before the midterm violence, and before outbreak of “mostly peaceful” (partly violent) protests through 2020 which have spilled continuously into 2021. Important 2021 footnote: we already achieved Mr. Mines’ definition of “Civil War” last year, and are continuing to do so in Portland (so far) this year. And if the Chauvin trial goes sideways, expect more.

I think a lot about what happens if the culture war goes hot. It occasionally keeps me up at night. And the thing that frustrates me the most is how easy it would be to avoid, strategically speaking, if the two sides just spent a little time projecting the outcomes.

When I first published “The Surprisingly Solid Mathematics of the Tin Foil Hat Gun Prepper” on April 20th of this year, a brief jaunt into the application of basic disaster mathematics to a possible violent revolution in the United States, I caught no end of shit for even entertaining the possibility. Go check the comments section if you’re bored.

My, how two months of news cycle freakoutery can shift the narrative. Everybody’s on this now.

Media Roundup

My estimates at the time, based purely on a historical frequency analysis, set us at a 37% chance over a given lifetime that a US citizen will get caught up in a war to overthrow their government. When the conservative pundit Kurt Schlichter got a hold of my article, he restyled it into a culture war hit piece, which is fine if distasteful because that’s his job, forcing me to respond, begging for reason. I’m sure his article got more traffic, but hey, that’s cool. Since then, the concept is literally everywhere, across media sources both Blue and Red. One of the best ones in my opinion is by Robin Wright at the New Yorker, which I think we can all agree isn’t exactly a bastion of conservative thought. They have much deeper resources than I to tackle the question, and interviewed risk analysis professionals who are plugged into the topic. Their article:

As one might expect from the New Yorker, Ms. Wright talks a lot about Trump in there, and cites the increasingly questionable opinions of the Southern Poverty Law Center, but the best bits come from her conversation with Keith Mines: (bolded emphasis mine)

Mines has spent his career — in the U.S. Army Special Forces, the United Nations, and now the State Department — navigating civil wars in other countries, including Afghanistan, Colombia, El Salvador, Iraq, Somalia, and Sudan. He returned to Washington after sixteen years to find conditions that he had seen nurture conflict abroad now visible at home. It haunts him. In March, Mines was one of several national-security experts whom Foreign Policy asked to evaluate the risks of a second civil war — with percentages. Mines concluded that the United States faces a sixty-per-cent chance of civil war over the next ten to fifteen years. Other experts’ predictions ranged from five per cent to ninety-five per cent. The sobering consensus was thirty-five per cent.


Based on his experience in civil wars on three continents, Mines cited five conditions that support his prediction: entrenched national polarization, with no obvious meeting place for resolution; increasingly divisive press coverage and information flows; weakened institutions, notably Congress and the judiciary; a sellout or abandonment of responsibility by political leadership; and the legitimization of violence as the “in” way to either conduct discourse or solve disputes.

I claimed a 37% lifetime chance based on raw historical data, and caught flack, but the consensus among those particular risk analysis professionals is around 35% in the next decade. Let’s unpack that incredible figure for a second, with a little mathematics. I derived my 37% lifetime estimate based on an annual chance of 0.59% per year. If their 35% chance per decade estimate is accurate, using the same math (as done before, but backwards), they’d be looking at a 4.3% chance per year. Mr. Mines thinks it’s bigger.


According to the consensus among that particular set of experts, the current media and social climate has made a violent revolution against the government 7.3 times more likely than “normal,” on an average annual basis.

Are those experts for real? I don’t know. But that’s what mathematics says they’re saying. Given that, it shouldn’t be surprising at all that some people have started talking about it.

And when Vox tells you “don’t believe this crap,” (their words) you can file that away with so many other unscientific ideological monstrosities they’ve pushed. Since we love to trash on Vox around here, that article is so ripe I can’t put the pen down. Let’s just grab this little tidbit:

Experts on civil wars who have considered a larger sample of conflicts could’ve told Wright that civilwars in rich democracies, excluding secessionist terror campaigns like ETA’s in Spain or the IRA’s in the UK, basically never happen.

First, we see a typical Vox rhetorical tactic of bombarding the reader with links to establish a sense of intellectual consensus, knowing nobody’s going to follow the links. They do “this crap” all the time. Let’s peel it apart.

The first link is a study saying civil wars don’t often happen in “coherent democracies,” to which I intellectually retort, “duh,” and further wonder if they’ve noticed all the articles on Vox saying we no longer have a coherent democracy. The second link is a study saying the same thing as the first with different words. The third link is a study showing that racial and ethnic conflict isn’t a major civil war driver. Okay great, Blues and Reds aren’t races. The fourth study repeats the third, but also actually undermines Vox’s overall point, because it does list some of the very things Mr. Mines talks about as being civil-war predictive. (I brought similar things up in my article up top) Thanks for the links to academic studies of marginal relevance, Vox. They sure did make that paragraph pop, as long as nobody reads them. Click “Publish,” rake in the cash.

Those links aside, the worst is the “basically never happens” punchline link, which is supposed to prove their thesis. I follow it looking for another study, and instead it’s literally a link to a tweet by an assistant professor at Georgia College, that doesn’t help their thesis at all. The tweet:

“Excluding secessionist conflicts” …. What the heck? We are literally discussing secessionist conflicts here! Mr. Rudkevich excluded the American Civil War in his data pool for his probably well-informed tweet, which didn’t actually have anything to do with what Vox was talking about. Ahh, Vox, ye fickle mistress.

Game Theory

Enough with the media bashing. Let’s define terms, outline scenarios, and put them into a game theory matrix. We first adopt Mr. Mines’ definition from the New Yorker piece, despite its explicit exclusion in Mr. Rudekevich’s tweet:

“We keep saying, ‘It can’t happen here,’ but then, holy smokes, it can,” Mines told me after we talked, on Sunday, about Charlottesville. The pattern of civil strife has evolved worldwide over the past sixty years. Today, few civil wars involve pitched battles from trenches along neat geographic front lines. Many are low-intensity conflicts with episodic violence in constantly moving locales. Mines’s definition of a civil war is large-scale violence that includes a rejection of traditional political authority and requires the National Guard to deal with it.

This is a good definition. It’s also functionally useful in separating the wheat from the chaff, and helps us categorically throw out all the current media pieces talking about how we’re already in a civil war. We’re not. Civil strife is up, but in terms of raw violence we still probably haven’t cracked the 1960s benchmark. We get the occasional unhinged Red shooting Gabrielle Giffords or the occasional Bernie Blue shooting up a baseball field full of Red congressmen, but it’s still not a weekly occurrence. Yet. I mean good gracious, 12% of all U.S. Presidents have been shot, all but one fatally. An occasional politically motivated killing here is basically noise. ‘Merica.

(Author’s note: If the media ever finally releases the motives of the Vegas shooter, who accumulated over two dozen semi-automatic rifles after the Trump election and shot up a country music concert, and that motive is shown to be political, I reserve the right to change my mind. But closure on that is doubtful, given what we know.)

On face, we’re not there yet. Unless you’re poor Rand Paul, who’s been beaten, shot at, and had to field threats of an imminent hatchet attack. I’m waiting patiently for ESPN to nominate “Rand Paul Intern” as one of the new events on the X-Games.

Moving on, let’s presume we’re inside Mr. Mines’ definition and not talking about literal state secession. In my opinion, neither #Calexit nor #Texit could happen unless they both had a tacit agreement to bail simultaneously with each other, and that probably won’t happen because they aren’t on the same team. There’s never going to be a central government that pisses both off at the same time, and no other states have enough weight to bail unless it’s en masse, “Second Constitutional Convention” level stuff.

Let’s walk through the belligerents. Conservative/Liberal is too confusing, Progressive/Traditionalist doesn’t functionally work, and Republican/Democrat refers to two factions within the existing politic, so those are all garbage for our analysis. Let’s talk Reds versus Blues.

If the Blues start a shooting war, it’s going to be because the Reds are in charge, and the Blues perceive the Red authority to be fascist thugs destroying them with corporatist economic policy and depriving them of a standard of life they feel they deserve. And the ones who shoot first will be the ones who are the deepest into that style of rhetoric. Basically, it will be (neo?) Marxists who start shooting.

If the Reds start a shooting war, it’s going to be because the Blues are in charge, and the Reds perceive the Blue authority to be state-atheist, communist dictators, shoving universal poverty and all manner of immorality (by traditional standards) down their throats without recourse or opportunity for grievance, undermining their traditional views of America itself. Basically, it will be the Tea Fringe.

The common narrative at this point, particularly when it comes to the gun debate we’ve written a lot about around here, is “You can’t fight the government! They have tanks!” Everyone thinks they’ll send in the tanks.

Well … maybe.

But maybe not. And the “maybe not” might depend on who’s doing the shooting.

On Tanks

The two flaws in this “tanks” narrative:

(A) Sending in the tanks doesn’t necessarily work if the belligerent is armed with rifles and IEDs, (see: Afghanistan War) but more importantly,

(B) The President of the United States doesn’t physically drive the tank.

We’ve discussed (A) elsewhere. Let’s take a deep look at (B). There’s a term in the military called “Tooth to Tail Ratio,” or T3R because they love their acronyms. It’s basically a personnel ratio of killers to killer-support-staff. One of the great things about the US Army is they make lots of very interesting studies public domain, including this one by John J. McGrath (no relation to the dead US congressman) that describes the T3R evolution in the US military by different conflicts. Fascinating stuff. His study shows that modern military units used in our current brushfire wars are somewhere between 39% and 52% “teeth,” when you exclude civilian contractors. And the teeth are people. And they might have indoctrinated predilections that monkey with our game theory matrix. In fact, they should have them.

The military are deeply steeped in traditionalist indoctrination from the word go. As we discussed recently, indoctrination is operationally important to the military. They need it for morale reasons to do their jobs, which may include shooting people of different traditions to protect their own tradition. This indoctrination is most important to the “teeth” in the T3R model. They drive the tank. Not the President.

Now I have not done any vast, far reaching personality typing studies on the military personnel of the United States Armed Forces, but I’m going to go out on a limb here and say this. If you order a tank driver to drive over an American’s house, the chances he or she will disobey that order might vary significantly between a house flying a Hammer and Sickle and a house flying a Gadsden Snake. Unlike in a military dictatorship, our military is not a military pledged to a man or a woman, it’s pledged to an indoctrinated ideal.

And their superior officers must know this. They may in fact be indoctrinated along the same lines as the tank driver. And their superior officers know that too. All the way up the ladder. And the ones at the top might not be too keen on issuing orders they know will be disobeyed, regardless of their personal culture war position.

An American soldier might read this article and respond “well, dozing anyone’s house is an illegal order because they’re not an enemy combatant,” and that might be true on its face, until we dig back into our conflict definition. We are discussing a scenario where certain citizens have been labeled enemy combatants, owing to all the shooting they’re doing, and the military has this “enemies foreign and domestic” pledge to fall back on.

I’m not saying everyone in the military is a Trump voter, or has adopted a particular side in our still-mostly-peaceful-yet-amplifying culture war. The military is huge, and home I’m sure to every flavor or bent of American citizen, with a vast array of individual opinions. But it’s important for everyone conducting this admittedly strange thought experiment to realize that in civil war situations, the military becomes its own actor, particularly in democracies.

Look at the Comps

Syria and Libya are bad examples, because they were military dictatorships from the beginning, but think about what happened in Egypt five years ago. An Islamist extremist president wins a free and fair election in a developed, modern nation, and starts pushing stuff on the public in similar fashion to either of the two Blue/Red shooting scenarios above, at least from the belligerent’s point of view. The demonstrations were tremendous, and when they started to turn violent, the Egyptian army stepped in and said, “you’ve got two days to figure this out or we’re going to take over.” Which of course was an impossible demand, so they ran a coup d’etat. Fortunately for the Egyptians, they gave the government back, but not until about a thousand anti-coup and/or pro-Islamist protesters were shot. Cost of doing business in this intellectual space.

There are other examples too. Thailand went through a similar SNAFU in 2014, and the military has been running the show over there ever since, with no signs of a return to democracy any time soon. In fact, I’m hard pressed to think of an example of a representative democracy this century that’s shifted into civil war mode without the military taking over. In a very real way, our modern worldwide democracies operate at the leisure of their militaries.

Back to Our Scenario Analysis

One: The Reds are in power, and the Blues start shooting. Red President says, “send in the tanks,” and the tank driver says “okay.” He bulldozes the house with the hammer and sickle flag outside, people die, order is restored. Reds win, although it’s a suboptimal win, because we’d all rather just not shoot each other.

Two: The Blues are in power, and the Reds start shooting. Blue President says, “send in the tanks,” and the tank driver says, “No sir.” Or perhaps the battalion commander says “I’m not sure I want to do that, sir, my tank driver might say no.” Or perhaps the Joint Chiefs of Staff say, “you know, we’re the ones in charge of the most voluminous and effective killing machine in the history of the Milky Way Galaxy as far as we’re aware. Maybe let’s fix this on our own.” And the Reds win again.

Be clear, if the conflict is limited to isolated uncommon violence, the tank commander would still probably doze the Gadsden Flag house. But if it’s isolated, nobody calls in the Army in the first place. The scenario we contemplate is widespread violence inciting military response, and if there’s motivation for that, there’s motivation for the tank driver to be thinking about his or her side in that.

And none of this analysis takes into account the other counterfactual — that an insurgency can absolutely survive the full might of even the US military if it has rifles, IEDs, and time, on their home turf. And I think we’d all agree that one of the teams is a bit better prepared on that front than the other. The first prerequisite to winning a gunfight is having a gun.

No matter who starts the shooting, the Reds are very likely to come out on top. The win might be ugly, and entirely not-at-all-ideal in terms of US civilian casualties, which I hope we all agree are a bad thing. And it might mean an end to our current system of government. But if things turn violent, the Reds still win, in any objective scenario analysis.

(Admonition: I am not a genius Game Theorist, but I know enough to be dangerous, and to converse meaningfully about the topic. If you, the reader, would like to tackle this grid in a different way, please feel free and respond below. On we go.)


If you’re unfamiliar with Game Theory matrix notation, here’s how it works. One player’s moves are across the top, the other player’s moves are down the side, and you fill the grid out with the game outcomes, then look for saddle points. For this analysis the game rules are these: The authority acts first, the opposition acts second, and the combination of the action / reaction produces an outcome. We could probably combine these two into one graph with some rhetorical tricks, but I broke them apart purely so the results would be more clear in our current culture war context.

The first thing that both Blue and the Red partisans need to realize, is the top left and top right outcomes of either graph are not on the table, because the opposite player is not going to make an obviously losing choice. So scratch those out. Of the remaining three outcomes, we have one that leads to tremendous bloodshed, one that baits bloodshed, and one that leads to, at worst, a bunch of grumbling. Sign me up for the grumbling and non-shooting.

The Nash Equilibrium here should be obvious. The Blues should play slowly while they’re the authority, and the Reds should give the Blues a little room to roam, but not too much, when they’re the authority, and the opposition side should take what they can get. Any other choice is either an obvious loser or playing with fire.

Your personal takeaway as a reader, should be this:

  1. If you’re on the Red Team quit worrying so damn much. You hold all the cards, and by cards we mean guns, and perhaps also tanks, so chill out. Give the Blues enough room to play so things don’t turn violent, but throwing around the brinkmanship card on silly stuff isn’t doing anyone any good, and may risk the “democracy” in general.

  2. If you’re on the Blue Team, for Christ’s sake don’t incite violence! Seriously. Antifa is the worst possible thing that could ever happen to the Blues, because they’re risking everything for basically nothing. In short: (again) Chill. Out. Play your cards slowly, please.

  3. If you’re like me, and don’t have a team nor do you wish to join one, maybe consider doing a little preparing. And lord help us all, if the actors make suboptimal choices and the thing does go hot, do what the Afghans do. Say “yes sir, no sir, he went that-a-way sir,” to whoever shows up to your door carrying a rifle, be their uniform Blue or Red.


Buy the author a beer