Trolley problem

Lies, Damn Lies and Statistics →Go uptown Index: Click ᐅ to expand: Statistics Systems theory Tell me more Sign up for our newsletter — or just get in touch: for ½ a weekly 🍺 you get to consult JC. Ask about it here.

Le mieux est l'ennemi du bien.

—Voltaire

A runaway trolley is hurtling down a track toward five workers who will certainly be killed if it continues on its course. You are standing next to a point switch. You can divert the trolley onto a side track, saving the five workers, but only at the cost of a sixth person working on the spur line. Should you pull the switch to divert the trolley, actively choosing to save five lives, but at the certain cost of one?

Philosophy students love the trolley problem. Though introduced in the sixties as a thought experiment about some Roman Catholic moral principle, these days it gets great play when we think about real runaway trams, not metaphorical ones — driverless cars. This intuition pump now functions in the public mind as some kind of live moral dilemma, to be solved (and how could it be solved? It is designed to be insoluble!) before driverless cars can be more widely mandated.

What if a child runs out in front of a robo-car, inside the total stopping distance for the car? What should the robot do? Should it swerve to avoid the child at the expense of oncoming vehicles? Or carry on, possibly killing the child?

How should we configure this potentially horrific technology? Can we — should we — programme your vehicle in advance to manage the kinds of moral quandary the trolley problem presents?

Look: there are plenty of logistical problems and risk allocation issues to be solved before autonomous vehicles can go mainstream for sure, but this highly artificial hypothetical is not one of them. This is a thought experiment dreamed up for a philosophy journal. It is not meant to address real-world problems.

Parallel worlds of science and philosophy

By design, Philosophy professors live in a world entirely disconnected from the actual problems of real human beings. It is useful by comparison to compare the world philosophers live in with the one scientists live in.

Scientists live in this simple, stable world filled with hundreds of benign little nomological machines they have designed, and thanks to which everything reliably works: thanks to the pendulum equation, t = 2π√(l/g), we can predict the period of a swinging object. thanks to Boyle’s law we can calculate the pressure and volume of a quantity of gas.

Philosophers inhabit, more or less, a dark inversion of this world: it is also filled with ingenious nomological machines, only they are malicious: they exist — philosophers invent them — to bugger everything up at the merest opportunity. In the philosophers’ world, nothing works.

For each of the scientists’ happy heuristics that illustrate how the world works, the philosophers have their own intricately-crafted intellectual contraptions — here, a Brain in a Vat, there, a Chinese Room, over yonder, a Parallel Universe — designed to illustrate how, really, it doesn’t. A world where nothing works is an excellent one for philosophers because it means they can sit around arguing why not and they never have to get on and do anything.

It is a world of imaginary problems Like a, well, parallel universe. Philosophers therefore occupy themselves with rarefied, nomological problems that suit that ascetic vibe.

Negative-Lindy effect

If any of these philosophical conundrums actually happened in real life, real humans would quickly solve them. For something to have lasting value as a philosophy experiment, therefore, it must have some kind of negative “Lindy effect”.

The Negative Lindy Effect: A good test of the quality and resilience of a philosophical thought experiment is how long it can survive in the literature without ever presenting as a genuine problem in real life to actual people.

The real Lindy effect: things that have stood the test of time are more likely to continue doing so, while new things are more fragile and likely to become obsolete. The philosophical effect is the opposite. Plato’s cave: brilliant.

The Trolley Problem is just such an excellent philosophical thought experiment. What kind of real-world event has such a deterministic outcome, over such a short time horizon that is just long enough for a subject to take action that is partly but not fully remedial, that can be acted on with total moral clarity, and yet there is not enough time to contrive an alternative different solution, or warn anyone, or avert the outcome some other way or, if you can’t, at least communicate to the putative victims, in a scenario in which there is no overriding moral agency from someone else?

Artificial certainty

The trolley problem assumes that the subject has perfect, certain knowledge of both the present state and the future outcomes on both tracks — that the train is coming, it cannot be stopped, that the only possible variable is points operation and death on one spur or another is certain — but no-one else in the scenario has any knowledge, either of the current status or the future potential.

The subject is absolutely certain the affected victims will not survive. How? Can she see them? If so, why can’t she communicate with them? If she can see them, can they not see her? And the on-rushing trolley?

And what are they doing on the train tracks in the first place? Are they tied down there? If so, why are we obsessing about this plainly second-order moral dilemma of the points operator? Should we not be interrogating whoever tied them down? If they are not tied down, then what the hell are they doing? Can they not pay attention to their own environment? Can they not optimise their own well-being? Why can’t they get off the tracks? Who hangs out on train tracks? If they are rail workers, have they not had any HSE training?

So, while the subject of the thought experiment is omniscient, necessarily everyone else is entirely ignorant of every aspect of the scenario going down, and therefore unable to do anything to prevent it.

Artificial timeframe

In real life, there will usually be many “intervention points” before a crisis. Where there are not, nor will there be time for abstract moral calculation.

The Trolley Problem artificially compresses moral deliberation into an instant to close off opportunities for potential problem-solving and communication that would spoil it. It is a hypothetical: it requires the subject to pre-solve a fantastical moral dilemma in the abstract, before it happens, without being allowed to change the system to prevent the dilemma arising in the first place. This is to impose a simple system on a complex problem. The real world is not so bone-headed as to foresee problems and not solve for them in system design.

If we know in advance we have an inherent “trolley problem”, we should not wait for a literally foreseen calamity to happen and only then make a difficult decision. (Legal eagles: realising there is a risk and running it anyway is the legal definition of “recklessness”: someone would be guilty of murder in this hypothetical.) We should re-design the system with fail-safes now. Not redesigning the system is the moral failure, not “pulling or failing to pull a lever”.

If the time-frame were more realistic, there would be time to problem-solve, warn victims, or contrive other outcomes or just communicate.

Artificial (lack of) social context

The scenario strips away all social and institutional context: Why have we designed the system this way? How have we managed to get to this crisis point without anyone foreseeing it?

The reason trolley problems are not generally prevalent is because, in the real world, this social context will intervene to prevent the problems arising in the first place. We build systems to anticipate problems. We encounter design flaws once, and solve them. This is the Lindy effect: bad design gets junked.

Artificial moral agency vacuum

It reduces the question to a single binary choice while stripping moral agency from the other participants, who almost certainly were better placed to foresee anticipate and control events, and therefore should be pegged with it. It asks us to ignore questions of greater moral consequence than the one at hand. Who designed such an inept system? Who assigned the workers to dangerous conditions? Why weren’t they looking out? Who lost control of the trolley?

Our subject is, by comparison, a random bystander. She is not responsible for the situation yet, for the thought experiment to work, we are asked to accept that she is somehow fully responsible for the outcome while those who, by rights, should be responsible for it somehow have no moral agency at all.

By focusing on the switch-puller’s choice, we overlook the moral failures that led to the situation. This is great for philosophical machines, but it does not advance any real world moral arguments.

False paradox

You all know how JC loves a paradox; well, this is a false one. There is no real-world scenario where the trolley problem can play out as a genuine moral conundrum. I feel a table coming on, readers.

Caption text

Player	Responsible for	Time of awareness
		Before Breakaway	After Breakaway
			time to react	no time to react
System Designer	System design safety, foreseeable failure modes	Could implement: fail-safes, emergency brakes, worker safety zones, warning systems	Could activate: emergency protocols, backup systems	Too late
System Operator	Worker safety, track access, safe rostering, risk assessment	Could ensure: safe work procedures, worker positioning, track access control Maintenance schedule	Could initiate: emergency protocols, worker evacuation, system shutdown	Too late
Track workers	Challenging unsafe conditions, being expert, keeping a look out for danger.	Raise issues. Quit!	Evacuate track	Pray
Switch Operator	Looking after his switch	Make sure the switch is working.	Call for help	Pull switch!

Of all these moral, practical and systems considerations here — there are a lot — the hapless switch operator’s position is the least vexed. If it gets to the point where the only judgment remaining is “whether to throw the switch” and no one else is already fully culpable for putting the switch operator in the position where the only person who can save the day is her — and it won’t be: it can’t possibly be — and if, Q.E.D., the switch operator has had no time to think about it, much less take any evasive action either, then it necessarily follows that the switch operator will not be responsible either.

This will just be one of those things. Shit happens. Accidents happen. The switch operator does not have enough time much less information to make the moral calculus to intervene, and realistically is not going to.

The driverless cars

Applying all this to the autonomous car proposition, there is a false equivalence. Why does the trolley problem not arise for human drivers? Because, unlike TESLAs, humans aren’t perfect rationalising machines. They don’t run on GeForce chips. They cannot flawlessly assimilate three terabytes of data coming at them in real time — was it a balloon? A fox? A dog? A child? — within a microsecond, calculate a flawless Rawlsian moral outcome, calculate the brake horsepower, vectors, momentum etc and communicate that executive decision to the machine. If humans react at all, they may blindly stomp on the anchors without a conscious thought at all. Most likely they will just flatten whatever ran out in front of them and that will be that. Humans can’t be judged because they lacked the faculties to do anything else.

So, for one thing, we are finding a moral conundrum in a fractal segment of spacetime previously inaccessible to moral discourse. It would be long gone before the moral passions arrived on the scene. But for another, a hyper-powered super-intelligent machine still wouldn’t be able to identify the obstacle, identify the alternatives (veer left? veer right?), evaluate the potential collateral damage for each outcome, and assess their relative moral consequences (is an adult worth more than a child? less? what are the criteria? what if they are roughly the same? what if you don’t know?) and correctly execute the most morally profitable manoeuvre.

There are, perhaps, safety issues with driverless cars. However, the data suggest that their accident ratio is way lower than for human drivers. By obsessing with hypothetical thought experiments we might prolong a statistically better outcome.

Le mieux est l'ennemi du bien.

Too cute

The trolley problem is too cute to be a useful nomological machine, even for philosophers. By pitting dramatically unequal moral outcomes against each other makes it easy for philosophers to get worked up about it, but at the same time creates artificial clarity where none would exist in the wild. This is the ideal negative Lindy effect. Ethical choices do not present themselves as arithmetic. Moral calculus is hard.

So let us rework the dilemma: what if there were one child on each track? You could save this one, at the expense of that one? What one were a girl, the other a boy? What if one were rich, one poor? What if one were from a marginalised community? What if they both were, from different marginalised communities? What if the required moral calculus were genuinely ambiguous?

For, surely, in the tiny set of real-world examples where there was not a prior moral outrage that created to the conundrum, these ambivalent cases would greatly outnumber those where a neat utilitarian outcome was beyond doubt.

Some might say we can draw useful metaphors from the trolley problem — that it is a useful tool for exploring moral intuitions about action versus inaction, and intended versus foreseen consequences. But there are better ways of doing that, that would resonate in real-world scenarios, and without absurd mechanical artefacts to distract from the core ethical question.