Ethics, Philosophy

Precedent Utilitarianism: A Primer

Preamble

When I first heard about deontology, I was intrigued. Here was an ethical system that could break you, if you weren’t careful. I was young and hadn’t really systematized my morality yet, but I dearly wanted to. I’d just learned about the stages of moral development and I felt a keen need to be at Kohlberg VI.

Time passed and I forgot that systematizing was a goal of mine. While I aimed for consistency across my moral principles, I did this largely blindly, lacking a single meta-principle to guide me.

Last year, I read Eichmann in Jerusalem, A Report on the Banality of Evil, the (in)famous book by Hannah Arendt. The only ethics mentioned in the book is Kantian and Arendt herself is hard to pigeonhole into any one system. But reading the book set my mind afire. By the time I finished it, I knew what kind ethical system I wanted to drive me. I just didn’t have a name for it.

Arendt had shown the weaknesses in deontology, shown how someone who didn’t think, who just followed the right as their society defined it could, with no irony, claim to be a Kantian while committing the most unimaginable crimes. At the same time, Arendt’s response to the judges, her justification for Eichmann’s death felt wrong to me. I never disagreed with Arendt more than when she said: “certain procedures… important in [their] own right can never be permitted to overrule justice, the law’s chief concern.”

I filled up the whole last page of Eichmann in Jerusalem with a cramped response to Arendt. I felt like her conception of justice was little better that vengeance and that justice couldn’t exist without the procedures she just disparaged.

Eichmann in Jerusalem left me with nagging questions and an empty space I yearned to fill. It would be a while before I had my answers.

First and Second Order Utilitarianism

The summer after reading Eichmann in Jerusalem, I flirted with utilitarianism. I wasn’t entirely satisfied with it. It’s not that I mind debating torture vs. dust specks or trying to select a value function. My problems were partially caused by the fact that I’m a romantic and utilitarianism is cold and utilitarian. But it’s also that I continue to worry about systems and precedents. For me, too many discussions about utilitarianism stick to the object level. I wanted to talk about the ripple effects of every decision and found often there was no room to.

One day, I found myself looking for high value books to read. One option was Utilitarianism: For and Against, a book I didn’t read until long after this post post was published. Luckily, even before I read it, it led me to the concept of precedent utilitarianism. Finally, I had a name to put to the nagging voice inside of my head. I read a quick summary of precedent utilitarianism and knew that I had the ethical system I was looking for.

I’ve previously written an overview of several types of utilitarianism. At the end, I mention that they’re all what I call “first-order utilitarianism”.

Precedent Utilitarianism is a form of second-order utilitarianism. It doesn’t just look at first-order consequences of an action. It looks at the precedents an action sets.

Precedents

I wrote an essay about justice that focused on precedents. In it, I make the claim that “precedent is what changes actions from unprecedented to normal”. This may sound facile or even tautological. But there is a deeper point I’m driving at. For every action now considered normal, there was someone who was the first to do it. In Eichmann in Jerusalem, Hannah Arendt mentions something similar in passing. She believes that that the recurrence of any crime is more likely than its invention.

Many actions are done once, then never again. Or only a few times, by a few isolated groups. Others get repeated and copied until they become the new normal. The Manson family murders did not lead to a sudden outbreak of murderous cults. But the actions of Marius and Sulla led almost directly to the Triumvirate and the unravelling of Roman democracy.

What makes Manson’s actions different than Sulla’s? It isn’t just that murder is more horrific than dictatorship. A cursory glance at the history of the last half-century of ethnic cleansings lends some credence to Arendt’s belief that after the Nazis systematized genocide many others would follow in their footsteps.

Why some crimes and not others? I think the answer to this question lies in part with the influence or charisma of the person setting the precedent. Hitler committed his grievous crimes at the helm of a country. Sulla was surrounded by patricians who wished that it had been them who seized Rome. Charles Manson has been influential in certain underground scenes. But he never led a country or commanded more than thirty people.

So what we currently know about precedents is: they can be set by any action and are more likely to be set by people who command a significant following. Oh and one final thing. In common law jurisdictions, every single judicial ruling sets a legal precedent, which is enforceable on all lower courts within the same jurisdiction. This is the most literal manifestation of a precedent, an action that is inscribed in law as allowed or disallowed, all because someone asked a judge to rule on it.

Precedent Utilitarianism

With the information we just gathered about precedents, we can create a second-order utilitarianism that incorporates them.

In theory, it’s pretty easy. You take whatever value function you prefer to use. You take the proposed action. You feed it into the value function to determine the utility of the action. This is just like first-order utilitarianism.

But in precedent utilitarianism, you then you think about how likely the action is to create a precedent and how many people the precedent could effect. If you’re not famous and you don’t expect your action to be well publicized, then you only need to worry about precedents set among your immediate acquaintances. If you’re the Prime Minister or President of an important country your audience will be considerably larger. And if you intend to defend your actions in the court of law in a common law jurisdiction, you must worry about the specific legal precedents you’ll potentially set. Legal precedents allow actions undertaken by a single person to be at least as momentous as those undertaken by a head of state. Just look at Oakes or Roe.

Once you know the how likely and how large, you need to think about who will use the precedent and how. If you think it is ethical for your preferred politician to cover up wrong-doing because you think there is a lot of utility in her being elected, remind yourself that if she gets away with it (for a while), then she’s set a precedent that may also be used by the politicians you despise.

Given all the people affected by the precedent, their chances of using it for various things, and the potential utility or disutility of these things, you can calculate an updated net utility for the action.

Simple, right?

You may have noticed the problem. Utility function calculations beyond simple QALY evaluations are really hard. Adding in a bunch of hypothetical actions from a bunch of hypothetical people just makes it harder. And if the calculations are already impossible, it doesn’t do you much good to have an even harder set of calculations that you’re supposed to somehow pull off.

Heuristics

Precedent utilitarians (or utilitarians in general) would point out that the correct solution to calculations that literally take forever isn’t to spend forever doing them. There’s an opportunity cost to spending all your time thinking and none of it doing and this cost is considerable. The common solution is to do the best action you can see after a reasonable period of reflection and estimation of utility.

What represents a reasonable amount of time to spend on reflection and a reasonable resolution for the estimation depends on how important the decision is. Decisions about which restaurant to go to should be very quick and simple and largely guided by factors other than morality (for example, your local public health agency’s evaluations, or more reasonably, what kind of food you want to eat). Decisions like “where should I donate ten percent of my income” require a fair amount of reflection. But decisions like: “should we go to war with that dictator” require far more. The more potential there is to influence lives, the more it makes sense to sink resources into determining the optimal actions.

When it doesn’t make sense to spend dozens of hours on contemplation, there are a few simple heuristics that the precedent utilitarian can use.

First: is the action likely lead to an improvement in utility from a first-order utilitarian perspective? If the answer here is no and you don’t have an plausible mechanism for the action setting a precedent that will redeem the negative utility incurred in the first-order analysis, then you should trust the first order analysis and avoid the action.

Second: How potentially harmful is the action if generalized? If your worst enemy did the same thing, would it reduce the utility of the world? If you’re attempting to ban a certain sort of speech, for example, the general class of thing you’re doing is “banning speech”. I think we can all agree that the people we disagree with could ban speech in such a way that it would reduce the utility of the world. But if we’re making it illegal to assault someone, there are few ways that our foes can take “don’t hurt people who don’t want to be hurt” and make it reduce the utility of the world.

In general, the goal here is to consider ways that others acting along the same general principle could help or harm the world.

Third: Consider how strong a precedent you’re setting and how likely it is that others can also advocate along the same general principle now that you’ve made it easier. Remember also that special pleading (“no, you can only act along this principle in the ways we say you can”) and hypocrisy (getting angry at others who are doing the same thing you did, just from a different set of axioms and beliefs about the world) are very off-putting and can turn people against you.

The second heuristic deals with how your precedent can be used against you. The third heuristic with how likely this is to happen.

Fourth: Add this all up. If the precedent you set is safe (very difficult to use to decrease the utility of the world), your power is secure (the precedent is unlikely to be used in ways that you think will decrease utility), you’re unimportant (the precedent isn’t going to be used by anyone else), and your public support is non-fragile (you can survive hypocrisy or special pleading) then you can decide on first-order grounds. If a few of these aren’t true but you stand to gain a lot of utility, it remains safe to decide on first-order grounds. But if none of the conditions are met then it may very well be possible that you’d stand to lose net utility from second order effects. In this case, it probably makes sense to put your plan on hold while you spend more time calculating possible outcomes.

Other Ethical Systems

It’s a safe bet that most people aren’t utilitarians. It’s also true that you will eventually have to interact with people who are both not utilitarians and have different axioms[citation needed]. In both of these cases (but especially the second), it can be hard to productively express and argue about views. Some people avoid this problem entirely by embracing the comforting lie that those who disagree with them do so out of lack of education or stupidity. Alas, this uncharitable explanation is far too often just not the case. Sometimes you’re stuck arguing with someone who has beliefs that are just as internally consistent, logical, and evidence based as yours.

When faced with someone like this, you have options. You can ignore them, making like Spain and Portugal and partitioning the world between you (and then griping to your friends on Reddit/Tumblr about how stupid the other side is). You can fight them, attempting to kill and subjugate them (this one has largely fallen out of style in many places, thank goodness), or you can find common principles that you agree on that will allow you both to live in peace and have mutually beneficial relationships.

Precedent utilitarianism is very well suited to building up systems like liberal democracy, where differing groups can draft a mutually agreeable framework that allows them to live peacefully. Precedent utilitarians naturally look for principles that everyone can agree on and tend to support strong constitutional protections around many classes of actions that don’t affect other people.

On a smaller scale, precedent utilitarianism is useful when you need to convince someone with a different set of axioms or a differing ethical system that you are a reasonable person who is worth listening to. A natural effect of precedent utilitarianism is avoiding (in most cases) special pleading (whether out of desire to not alienate support, or because you’re worried about precedents your actions can set).

Avoiding special pleading makes you look principled. Someone can respect you arguing against one of their proposed plans of action (and give your arguments much more credence) if they’ve also seen you argue against other actions (especially ones they would expect you to support given your axioms) using the same general principle.

For example, if you’re a Catholic and are arguing against having Buddhist prayers at a town hall meeting you’ll have much more credibility if you have previously opposed having Catholic prayers read at town hall meetings (perhaps because you’re worried that it sets a precedent that could lead to other prayers being read, which might lead to less utility in terms of saved Catholic souls). If instead you’d previously argued in favour of Catholic prayers but are now arguing that separation of church and state preclude prayers in meetings, thein no one will take you seriously. Worse, they will probably have assorted ill feelings towards you, making you less effective at convincing them even in unrelated matters.

Practical Examples

I want to give examples of the heuristics I discussed earlier in action. To make this essay interesting to people with a variety of axioms, I’ve picked two examples of legislative interventions proposed by different groups and argued against each intervention using the axioms (as best I understand them) of the people who I’ve observed suggesting it. First I’ll use activist left axioms. Then I’ll try and pass an ideological Turing test and pull off small government religious conservative axioms.

 Activist Left

There is a growing clamour from leftists to shut down police unions. The logic goes that police unions advocate for the good of their members at the expense of society at large and most particularly, those already disadvantaged by race, sexual orientation, gender expression, poverty, mental illness, or a combination of these factors.

These activists generally believe that without the political clout and collective bargaining ability of police unions it would be easier to require officers to wear body cameras, easier to demilitarize the police, and easier to ban discriminatory practices like carding and stop and frisk. They also believe that without union representatives it would be much easier to suspend and fire officers suspected of misusing force.

Let’s assume (for the sake of argument) that activists are correct and dismantling police unions would reduce police violence. A reduction in police violence would lead to an increase in utility for almost any value function, as long as there weren’t direct effects that led to counterbalancing increases of violent crime. Let’s assume that even if there are some negative side effects, there is ultimately an increase in utility. This lets us move on to the second step.

(The proper utilitarian thing to do here would be look into studies and data analysis about what the likely crime effects of such a move would be. Because the focus of this essay is precedent utilitarianism, I’m not going to go into the nitty gritty here. I’m just going to do what the proponents do and assume everything will work out OK.)

The generalized action here is: “it is acceptable to weaken collective bargaining rights or forcibly de-unionize workers”. If you are a leftist, I want you to take a moment and imagine what sort of effects there would be if your worst enemy did this kind of thing.

The first target would almost certainly be teachers’ unions, long a target of conservative ire. The possible results of the weakening or abolition of teachers’ unions reads like a grab bag of all the left’s education bogymen: performance based pay (The Rand Corporation found performance based pay to be ineffective, but see also Slate Star Codex), more charter schools (in America), less job security for teachers, and larger class sizes.

Beyond teachers’ unions, there are dozens of ways that conservatives would love to stymie organized labour. It is generally accepted among leftists that unions are good for everyone, even people who aren’t unionized. Therefore, a great decrease in union membership or weakening of union power would lead to a loss of utility from a conventional leftist point of view.

If we could get rid of police unions without significantly risking other unions, then the analysis would probably come up positive (given our other assumptions). Unfortunately, it would probably take laws (and successive legal victories) to force police unions to disband or strip them of collective bargaining rights. There is no way to argue that laws and court cases don’t set precedents. Laws passed by previous governments give future government permission to legislate in the same space. And successful court cases (especially in common law jurisdictions) set the legal precedents that were discussed earlier. Courts cannot support disbanding a police union without setting the general precedent that unions may be forcibly disbanded.

In addition to creating one of the strongest possible precedents, abolishing police unions but demanding that no other unions be affected is a strong case of special pleading. In this specific case, there is even more potential for harm than in most, as a majority of Americans are confident in the police.

Looking at all of this, we’re looking at a potential for an increase in utility (assuming that there isn’t a protest or other work action from police that leads to rising crime rates), while setting a precedent we acknowledge is both strong and dangerous.

From a precedent utilitarian point of view, it seems unlikely that abolishing police unions will actually lead to any increase in utility. Instead, precedent utilitarians might focus on the outcomes they wish to see (increased use of body cameras, better use of force policies, more restrictions on discriminatory policing, funds for hiring more police officers from diverse backgrounds and diverse communities) and try to legislate them individually.

Of these, body cameras and hiring police officers from more diverse backgrounds (which can be spun to constituents as simply as “hiring more police officers”) seem the most likely to be easy to pass with broad support and probably represent the easiest starting place for a quick utility gain.

Small Government Religious Conservative

Several conservative controlled legislatures in America have begun to pass (or attempt to pass) so called “religious freedom restoration acts” or similar bills. These bills are largely designed to pre-empt local non-discrimination ordinances that conservatives feel place individuals into a conflict between their deeply held private beliefs and the law.

As above, I’m going to skip questions about the correctness of these beliefs or that they represent a gain in utility. Just as some people feel that forcing police to de-unionize will lead to a better world, some people feel that these bills will lead to a better world. Instead of disagreeing with these beliefs on the object level, I want to show that they are inconsistent with other conservative axioms and would fall under the class of beliefs that precedent utilitarianism suggests should be reject even if they’re based on correct axioms.

The generalized action here is: “it is acceptable for distant legislators to force lower levels of government to legislate as they would.” If you’re a conservative, I want you to take a moment and imagine what sort of effects there would be if your worst enemy did this kind of thing.

The first target would almost certainly be rural communities in otherwise liberal states, which tend to have much different laws around gun ownership and property taxes than the larger metropolises which make up the majority of the voter base.

Beyond guns and taxes, there are dozens of regulations that central liberal governments would love to impose upon rural conservatives. Look at what’s going on in Alberta for just one example. And would any conservative trust a liberal state government to protect the coal or fracking jobs on which so many rural communities survive? Living in a city, it’s far too easy to forget that these things have to come out of the ground somewhere.

If local bills could be overruled without setting any precedents, maybe there’s a utility gain to be had. But this seems unlikely. It almost certainly will require a few court cases to sort out which level of government has which power and once powers have been taken away from local government and given to the centralized government, they cannot be taken back. Politicians almost never let go of the power that they’ve fought so hard to gain.

In addition to setting strong legal precedents, to claim that overruling municipalities in the way you like is okay while demanding no other government overrules them in ways that you don’t like is definitely special pleading. And as Americans still trust their local representatives more than their state representatives, this is likely to backfire.

Looking at all of this, we’re looking at an increase in utility from state laws overriding local non-discrimination ordinances, while also setting a strong precedent that states can override whatever local laws they don’t like; something we should acknowledge as dangerous and negative.

From a precedent utilitarian point of view, it seems unlikely that overriding local non-discrimination ordinances will lead to any increase in utility. Instead, precedent utilitarians with these axioms should focus on increasing tax breaks for religious schools or other social institutions they believe will push society in the direction they think it should go.

Back to myself: one principle of small government conservatism that I find laudable is the belief that local governments are best placed to fix problems. All too often central planners come up with ridiculous, unworkable ideas out of ignorance of the conditions on the ground. In addition to my grave concerns about the content of “religious freedom ordinances” or “bathroom bills”, I’ve been shocked to see conservatives suddenly advocating for solutions at the state level and liberals claiming that local people know best. And I’m not the only one.

Downsides

I chose one of my examples very deliberately: to emphasize one of the weaknesses of precedent utilitarianism. People who are already privileged (like me!) are going to find it easiest to demand that potential changes must be considered and interrogated for bad precedents and abandoned if there is a chance that they might lead to enough disutility in the future.

It’s easy for me to urge caution around police unions. The police aren’t busy killing people who look like me. It’s easy for me to say that unprincipled exceptions should always be avoided. Unprincipled exceptions aren’t already being made at my expense. It’s reasonable to ask: “if they’re making exceptions for us, how come we can’t make exceptions for them.”

Pointing out that we wouldn’t have these problems if everyone already followed precedent utilitarianism doesn’t count as an argument. So what if it’s true? It wouldn’t change anything. The world should be engaged with as it is, not how we wish it to be. And we have to reckon with the fact that sometimes partially adopting an idea is worse than adopting none of it (see for example most arguments that start: “well, in a perfect libertarian society…”).

But this weakness isn’t unique to precedent utilitarianism. It’s a weakness of utilitarianism or of consequentialism more generally. Most constructions of utilitarianism place no inherent value on fairness, only value on some of the effects of fairness. Instead of trafficking in an ethical coin that is intuitively understood, they deal in cold, hard utility and disutility. Life years saved or lost, pleasure and pain, preferred and dispreferred states, all aggregated over the population of the world. These are the tools utilitarians have.

Precedent utilitarianism demands a deeper examination of consequences than some other constructions of utilitarianism. But it can’t change the fact that consequences are all utilitarians care about.

I advocate for precedent utilitarianism because I think that it doesn’t suffer from the problems of libertarianism. I don’t think even stumbling, imperfect precedent utilitarianism will lead to a worse state than the current one. But I don’t have proof. I can claim some institutions (the courts, liberalism) as obvious manifestations of precedent utilitarianism.

But this leaves two avenues of disagreement. First, you can claim that these are the by-product of something else and only have a serendipitous resemblance to precedent utilitarianism. Or you can claim that these are in fact not good things. It all depends on your axioms.

And this is all circular. People like me in positions of privilege tend to have axioms that assume their experience. Meanwhile, systematically disadvantaged people tend to have axioms that assume their experience.

Here’s what I have to try and convince you, even if there’s a huge gap between our axioms. Scott and Ozy often talk about ethical systems that fail gracefully. Imagine that you thought something or someone was bad and did everything permitted by your ethical system to stop it or them. Now imagine that you were wrong. How badly have you fucked up?

Precedent utilitarianism fails gracefully. Does your ethical system?

Epistemic Status: Ethics