The Moral Case for Pushing Utilitarians Onto Trolley Tracks
In his paper "Should We Sacrifice the Utilitarians First?", Saul Smilansky argues for a "Platinum Rule"--treat others as their moral principles would have everyone treated.
Just about every introductory ethics class I’ve ever taught has included a day where I illustrate the Trolley Problem with some crude drawings of stick figures and train tracks on the white board, and I ask for a show of hands. Who would save the five people on the first track by diverting the train onto the secondary track where there’s only one person in the initial example? Who would push the large man onto the one and only track in the second version?
Consequentialists think moral rightness or wrongness ultimately derives from good and bad consequences. Deontologists think categories of actions are right or wrong in themselves. On the face of it, we’d expect pure consequentialists would raise their hands both times while pure deontologists would keep their hands in their laps throughout the whole exercise. But, from the time I started teaching these classes (near the end of George W. Bush’s second term) to the one I taught last semester, I’ve never had a class without a large majority of diverters-to-the-second-track-who-wouldn’t-push-the-guy.1
There are, of course, clever deontological explanations for why it’s ok to divert the track and clever consequentialist ways of justifying not pushing the guy. But one way of reading most people’s reactions is just to assume that most people are neither pure consequentialists nor pure deontologists but pluralists—we think consequentialist considerations matter but we don’t always think they’re decisive.
This suggests the possibility that even many people who wouldn’t push the guy in the “classic” second trolley case might be reluctantly persuaded to sacrifice an innocent person in a remixed version where the consequences of inaction are so extreme they simply wash away the deontological considerations. Assume that we’re in such a situation, but that there’s some wiggle room about who we should sacrifice. In his paper “Should We Sacrifice the Utilitarians First?”, University of Haifa philosophy professor Saul Smilansky makes the novel suggestion that we should take into account the moral views of potential victims.2
Kantians, he thinks, have less of a claim than the rest of us to be saved through the sacrifice of others. And utilitarians have less of a claim against being sacrificed.
Note, by the way, that “utilitarianism” is technically a specific kind of consequentialism, but the specific features that make it different from other possible consequentialisms aren’t important here. I’ll use “consequentialism” and “utilitarianism” interchangeably in what follows, the way some Southerners call every soda “Coke” (even Pepsi).
I won’t quite do that for Kantianism and the broader term “deontology” since the specific features of Kantianism are going to matter at least a little bit. Feel free to skip ahead to the next section if you reread Kant’s Groundwork of the Metaphysics of Morals every morning while you eat your breakfast cereal and you know all this stuff by heart, but:
Kant thinks there’s just one moral rule (“the Categorical Imperative”) and that everyone is morally obligated to follow it all the time, even if it leads to bad consequences. Confusingly, he gives several different and not-always-obviously-equivalent formulations of this “one” moral principle, but for now let’s just stick with the first two:
The Formula of Universal Law: Always act according to that maxim that you could will to be a universal law.
The Formula of Humanity: Always treat humanity, whether in your own person or that of another, as an end in itself and not merely as a means to an end.
The italics are there to avoid the standard misinterpretations. We’ll get back to both of these.
Smilansky calls his view “Designer Ethics” (DE), which is fun but slightly misleading. It makes it sound like he’s offering an alternative overall theory to standard views like utilitarianism and Kantianism. He’s not. Smilansky is a pluralist. Early in the paper, he clarifies that, “DE is one resource among others, in our moral-reflective arsenal. The claim made here is that this resource has been unduly neglected.” And anyway, treating it as a comprehensive moral doctrine wouldn’t really work, since it’s only applicable when the people the Designer Ethicist is interacting with themselves advocate comprehensive (non-DE!) moral doctrines. Otherwise, he says, DE would be a “mirror with nothing to reflect.”
While he doesn’t quite connect these dots himself, this can help us see our way past one of the pragmatic concerns about DE he acknowledges upfront, e.g.:
How will we know, in daily cases, what is one’s normative position? Surely, for example, her view might not always be the view she professes.
And:
How will we treat groups of people holding different positions (say, if among the five in our Trolley case some are utilitarians and some deontologists)? Or, if the five are utilitarians and the one a deontologist, then this apparently generates an inconsistent judgment.
These might be big problems for trying to navigate moral dilemmas using nothing but DE. But if we’re pluralistically adopting DE as one kind of consideration among many, neither of these present any special problems.
Starting with the first worry, we might not be sacrificing someone “first” because we falsely believe that (unlike other possible sacrifices) she object on Kantian grounds to sacrificing anyone, when she’s really a closet utilitarian. But this doesn’t seem obviously worse than taking into account the five children someone claims to have at home when they’re secretlessly childless. In both cases, the Designer Ethicist could say, we acted wrongly to the extent that we acted on bad information, but if this possibility doesn’t give us a strong reason to not take children-waiting-at-home into consideration, it doesn’t seem to give us a strong reason not to take the moral views of potential sacrifices into consideration. And anyway, given pluralism, neither of these worries make it particularly hard to apply DE. Given significant uncertainty about the views of the guy on the bridge, or a mixture of consequentialist and deontologists on the track, there might not be any clear way to apply DE. But that just means we can fall back on the rest of the resources in our moral-reflective arsenal.
Smilansky also excludes some doctrines that might be held by potential victims or beneficiaries of our actions from consideration. For example, “DE need not apply to the treatment of racists (in themselves or compared to other people), or to the choice between racist and other policy, the racist standards themselves.” While he doesn’t say much about the basis of this exclusion, the idea seems to be that there’s some baseline of moral plausibility views have to achieve before we take people’s adherence to them into account.
Elsewhere in the paper, he gestures at an analogy between DE as a view in general moral philosophy and common views in political philosophy according that take into account what Smilansky calls “the inevitable fact of reasonable moral disagreement” and insist that respecting persons involves some level of respect for their views. Think, here, of Rawls’ insistence on leaving room for people with wildly varying conceptions of the good life to pursue different individual “life plans.” Or even standard arguments for freedom of conscience and the separation of church and state.
So while the proposal might look at first glance like a fair but distinctly unfriendly demand that people who are willing to make extreme demands of others put up or shut up—an impulse that can be crassly summed as, “you first, asshole”—Smilansky is often eager to frame it the opposite way, as a matter of respect for serious and deeply-held moral positions.
[A]nother advantage of DE is that it seems most closely aligned with consent and furthest from coercion. Because utilitarians believe in sacrificing the one for the many, they have in a sense already volunteered to be sacrificed: they believe in it. The same goes for the five deontologists on the wrong side of the tracks… [Further,] this connects DE to the notion of respect for persons. This notion is notoriously slippery. Yet DE can lay claim to capturing at least certain senses of this notion, better than the alternative views. We can see this from what we have already noted: DE takes the moral views of individuals more seriously than other major normative theories. It says that the right thing to do to a person pro tanto3 tracks his or her view of what ought to be done. It takes persons to be reflective, morally serious beings, who are ready to live by (or indeed die for) their principles; wish to avoid hypocrisy, and abandon moral complaint and resentment that has no grounding in their own views. This is admittedly somewhat of an idealization of what many actual people are like, but it provides a normative guiding ideal.
One obvious concern about DE is that it, in effects, punishes people for being serious enough about morality to be in the small percentage of the population with well-thought-out moral views, and especially punishes them for being principled enough to have quite demanding views. (“Don’t save innocent people if the only way to do it is by sacrificing others” or “don’t sacrifice innocent people even to save others.”) But Smilansky can reasonably respond here by asking how morally serious you can possibly be if you complain about having the moral standards you want applied universally applied to you.
So—is he right?
Despite everything I’ve said so far, I have my doubts.
To see why, let’s start with one of Smilansky’s more memorably rhetorical flourishes:
The ancient moral injunctions of ‘Do not do onto others as you would not wish done onto yourself ’, or ‘Treat others as you would wish others to treat you’, put concerns about the liability that one acquires through one’s moral stance into morality. Later formulations of such intuitions, for example in the Kantian generalization test (Kant would not quite agree to this categorization of his view), have put them at the centre of our moral thinking. On the one hand, this helps to make DE less eccentric and more approachable. Yet this whole direction is not, in itself, an anticipation of DE. For, such pre-Kantian or quasi-Kantian positions view the thoughts about how we will be treated as a methodology for ascertaining what ought to be done to everyone. They help us to take seriously the process of moral reflection and codification but would not result in the sort of diversity that DE produces, whereby different people are treated differently (e.g. sacrificed or not), in accordance with their views.
Nevertheless there is a sense in which DE has a Kantian spirit to it. Indeed, in one way it arguably is even more ‘Kantian’ than Kant! After all, it respects people as moral agents to the point of truly treating each person as a law unto themselves. If the Golden Rule is to treat others as you would be treated, the Platinum Rule being proposed is to treat others as they would be treated.
One initially important thing to track is that saying that Kant wouldn’t “quite agree to this categorization of his view” severely understates the case. This is why we italicized the could earlier in the Formula of Universal Law. Turn it into a would and the result does look like a convoluted version of the Golden Rule (or even a kind of disguised consequentialism), but Kant sees this as a test of consistency.4 Would it be possible for everyone to act on whatever maxim you’re acting on (without undermining your reason for acting that way in the first place)? Kant gives the example of making false promises. If I falsely promise that if you give me some money today I’ll pay you back tomorrow, that fails the universalization test because in a world where everyone acted that way, no one would believe such promises and it wouldn’t work. This isn’t “how would you like it if everyone made false promises?” It’s not “no one being able to make false promises would be a terrible consequence.” It’s just evidence of the failure of a particular kind of consistency.
This is worth harping on because, if you squint at a certain angle, DE can look Golden Rule-y. But you can only make it look Categorical Imperative-y by squinting so hard you miss a crucial distinction. Or at least that’s true if you’re thinking of the Formula of Universal Law.
It’s also true that the bit about the law unto themselves captures a lot of things Kant says in the Groundwork when explaining what the Categorical Imperative is, and even that a formulation we skipped above is:
The Formula of the Kingdom of Ends: Always act as if you were through your maxims a law-making member of a kingdom of ends.
Without opening multiple non-trivial cans of worms about what exactly this means and how it be different from the Formula of Universal Law, you can basically see Smikansky’s point here. If Kantian agents are supposed to be law-making, there’s a certain sense in which there’s at least a “Kantian spirit” to treating them according to their own ideas about the nature of the moral law—though, importantly, Kant himself insists that it’s our core rational selves and not just the contingent things about our minds that are doing the law-making, this guarantees that we’ll all come to the same conclusions.
To put my own cards on the table, I’m pretty skeptical about many of Kant’s ideas about both the metaphysics and epistemology of morality. But, like many people who aren’t pure Kantians, I do think there’s something compelling about his moral picture—and that the heart of that appeal is the Formula of Humanity. If we’re focused on that, DE is considerably less ‘Kantian’ than Kant—and that’s exactly the source of my biggest reservations about it.
This crystalized for me at the end of the paper when Smilansky applies DE to political philosophy, saying that, “Those who oppose government educational subsidies should not receive them.” He hedges this in the next sentence by throwing in the word “arguably” but with or without it, this is the place where I get off the bus.
One of my deepest socialist convictions is that there are all sorts of life goods—like healthcare, or as in this example, education—that everyone in a society with a reasonable capacity to provide these things should get just for being a person. The demand for public healthcare is a richer source of dramatic, intuition-tugging examples, but I absolutely put free public college in the same basket. I find it disturbing that some people get those few years of getting to focus on reading and thinking and figuring out what they want to do with their lives and other people are denied it and have to go straight from high school into the workforce—having been, perhaps, sagely advised not to rack up loans they may not be able to repay—because of differences in family background utterly beyond their control.
If you go read some Ayn Rand and go through a libertarian phase in college, I’ll roll my eyes and hope you get over it. But I still want you to be able to have access to free public college.
Going back to Kant:
Remember that the Formula of Humanity says that persons—yourself or others—shouldn’t be reduced to “merely” means to your ends. That doesn’t mean it’s a violation of the moral law to ask your friend to help you move a couch. But it does mean that you can’t use a person as a human paperweight by pushing them in front of on oncoming trolley. That’s reducing them to merely a means. And it does seem to me that taking them seriously as a person—even if they’re a utilitarian—gives us a pretty overwhelming reason not to use them this way.
To be fair, Smilansky can technically accommodate this point. Given pluralism, it could be the case that in almost all imaginable circumstances that “pretty overwhelming reason” trumps consequentialist reasons. But in some fantastical scenario involving, perhaps, a ticking nuclear time bomb, it could itself be overwhelmed, and that in a particular version of this case there are multiple possible victims with different normative views and, voila, this is DE's time to shine.
And…maybe. But I have to say even when I consider that scenario I don’t particularly feel like patting myself on the back about how much respect I’m showing for the deeply-held moral worldview of the utilitarian I’m pushing off the bridge.
People who’s knowledge of “the Trolley Problem” derives from NBC’s The Good Place or trolley memes on social media usually don’t seem to know what the “problem” is supposed to be, but when Judith Jarvis Thomson coined that phrase, this was the exact “problem” she had in mind—that it seems right to divert to the second track, and wrong to push the guy (or, in a still-more-extreme example, to harvest five organs from a healthy patient who comes in for a check-up to save five people who need transplants) despite the fact that the tradeoff seems to be the same in all three cases—let five die or save them by killing one.
Many thanks to Matt, Victor and Ethan from the Academic Edgelords podcast who invited me to discuss this paper with them on their show and hence inspired this essay, and to Matt for reading the rough draft.
“To that extent”—in moral philosophy, pro tanto considerations are ones that have some weight even if they aren’t necessarily decisive.
If you did misinterpret it in one of those ways, at least you’re in good company. John Stuart Mill interprets it as disguised consequentialism in the introduction to his book Utilitarianism and Friedrich Nietzsche calls Kant a “catastrophic spider” for smuggling Christian ethics back into secular philosophy in The Antichrist. They both seem to have missed the would/could distinction.
Now let’s apply Smilansky’s more recent paper on the topic and we can get to work sacrificing the Christian Utilitarians first
Nice. Smilansky's claim (and your title) strike me as a wonderful example of old-fashioned philosophical trolling. Diogenes and Nietzsche would be proud.