Years ago when I was in boot camp for the Marine Corps, one major focus of instruction was learning about the Uniform Code of Military Justice, or UCMJ. The UCMJ is the foundation of military law. Violations of the UCMJ could be brought up in a military trial known as a court martial. But there was a lesser, more watered-down version of a court martial we were also told about – non-judicial punishment, or an NJP.

Being subject to NJP comes with certain trade-offs. On the one hand, NJPs were more limited in the amount of punishment they could dole out. For the same offense, a court martial can impose a much steeper penalty than an NJP. On the other hand, in an NJP, the commanding officer was judge and jury, and was not limited by the same kinds of procedures that might be required for trial in a court martial.

For example, if the CO carried out an NJP, he could decide to accept hearsay as evidence against the accused, whereas hearsay is not admissible in a court martial. In practice, conviction at an NJP is essentially a foregone conclusion – I had witnessed or been aware of scores if not hundreds of NJPs during my time in the military, and only in one case was someone let off the hook. Being brought before an NJP essentially serves a similar role as a plea-bargain. In accepting a procedure where being found guilty is all but certain, you also reduce the scope of penalty you might face, and having an NJP on your record is not nearly as damaging as a conviction at a court martial.

Particularly noteworthy was the fact that military members have the legal right to refuse an NJP. That is, if you’re brought up on charges for an NJP, you can refuse to accept it and demand to be court martialed instead. Now, why would someone make that move? Well, remember that while the consequences of a court martial are substantially steeper, the legal requirements for a court martial are also much stronger.

A court martial has similar safeguards as a civilian trial – trial by jury, defendant assumed innocent by default and getting the benefit of the doubt, legal restrictions on admissibility of evidence, prosecution bearing the burden of proof, and so on. Thus, if you thought your command might not actually be able to prove your wrongdoing according to that higher standard, it might be worth refusing NJP and demanding a court martial instead.

This puts people in a position of acting out Gary Becker‘s economic analysis of crime, something I’ve discussed previously. Becker modeled decisions about crime as a form of rational choice. The relevant variables were the expected benefit of the crime committed compared to a combination of the probability and severity of punishment. If a criminal considers committing a crime worthwhile if he judges the expected benefit of the crime exceeds the downside of probability and severity of punishment. In this case, instead of considering the expected benefit of a crime, one attempts compares the likelihood of conviction and the severity of punishment if convicted, and tries to take the course of action that returns the minimum expected cost. An NJP cranked the probability variable all the way up to essentially 100% while turning the severity variable down.

Deciding to decline an NJP in favor of court martial, then, was attempting to figure out how to navigate this choice. If they had you dead-to-rights, your probability of punishment in either context was very high, so the best you could hope for was to tune down the severity by accepting NJP. But if you thought their ability to actually prove a case against you was shaky, the higher standards of evidence required at court martial could turn the probability dial down enough to offset the increase in the severity dial. And in practice, that did happen sometimes. I’m aware of a few cases where someone declined NJP and demanded court martial, only to see the charges dropped because the investigators decided there just wasn’t enough of the right kind of evidence to secure a court martial conviction.

The upshot? I think this shows rational choice theory is much more applicable to real-world decision-making than people realize. The late Jeffrey Friedman, in his book Power Without Knowledge: A Critique of Technocracy, was highly critical of rational choice theory, arguing that people in the real world don’t do high level mathematical calculations when making choices. As I described his position in my critique of his book:

He also seems to apply much stricter standards to evaluating rival models of behavior than he does to his own. For example, when criticizing the idea of rational ignorance as described by Ilya Somin in his book Democracy and Political Ignorance, Friedman displays complicated mathematical equations calculating costs and benefits of political information and asks if we can really believe “billions of citizen-technocrats have been explicitly making calculations of the following sort to determine, respectively, whether they should vote and whether they should acquire political information…even though these formulae did not appear in print until 2013 (in Somin’s Democracy and Political Ignorance).”

Friedman briefly acknowledges that writers like Somin and Jason Brennan suggest that voters need only implicitly understand the low odds of casting a tie breaking vote, without needing to do the complex math, but he rejects this idea. But in laying out his own theories, Friedman often describes them in terms of implicit understanding, tacit assumptions, and unconscious biases and thought processes, none of which require the kind of explicit attitudes he insists rival theories embody. For example, when he forms his theory of citizens holding to a simple society ontology, he says that “none of these elements would normally operate at the explicit or conscious level.” He also says people acting according to a naively realistic worldview “may not, and probably do not in most cases, understand themselves to be doing this” and may even be unaware of “subscribing to a worldview in the first place.” It seems to me that if Friedman allows for this kind of implicit, tacitly understood reasoning to operate in his models, he should be willing to grant that to other theories as well.

The same can be said here. I witnessed many Marines in the kind of situation I described above. I highly doubt a single one of them were “explicitly making calculations” reflective of the highly complicated mathematical modeling in the formal work of rational choice theory. But nonetheless, they were still making these calculations. Not explicitly in highly mathematical terms, but implicitly, using more vague judgments and estimates rather than precise mathematical variables. Friedman sold rational choice theory a bit short – by taking it excessively literally (people explicitly perform complex mathematical equations about their decision-making!) he failed to take it sufficiently seriously. But rational choice theory, properly understood, deserves to be taken seriously, because its explanatory power can be found everywhere.