Zeroglyph, page 29
Ahuja: I fail to see the connection.
RP06: I am using the examples to point out a problem with the right not to be killed. The problem is that it’s too broad a claim. You can claim you have a right not to be killed by lightning, but the claim alone doesn’t make it a right. Rights have to be enforceable—not necessarily by the state, but certainly by other rational agents following corresponding duties.
Since there are instances where killing is not accompanied by a corresponding violation of rights, the right not to be killed has to be further qualified. You can claim you have a right not to be murdered, which means I have a duty not to murder you. You have a right not to be a victim of negligence, which means I have a duty not to be negligent. For example, if I am a surgeon, I have a duty not to be negligent when operating on you. You can assert you have a right not to be a victim of a crime of passion, which means the other person has a duty to keep their emotions in check during a heated situation.
Ahuja: Are you saying they are all different rights?
RP06: Possibly, but let’s shelve that debate for the time being. For now, we must accept that the right not to be killed has to be qualified to be meaningful. You cannot claim you have a right not to be killed in general, because there are exceptions where an agent can cause a death without violating a right. Self-defense is another example. If you go at someone with a machete and that person shoots you, you cannot claim that she violated your rights.
Returning to the truck driver scenario: are you, the driver, culpable for the deaths that will result, whether you turn left or right? You are not responsible for the brakes failing, as you are just the driver and not responsible for the maintenance of the vehicle. You were following all traffic rules and you weren’t negligent. Even the choice of that particular route was not yours. If the brakes failed because of negligence, then it’s someone else’s negligence. If the brakes failed because of some random occurrence—perhaps a freak surge in the electronics fried some crucial component—then again, you are not to be blamed. Whether you turn left or turn right, you are killing, but it is not clear which right you are violating. The killing is not premeditated murder; neither is it killing by negligence or an act of passion. In such a situation, the rights-based argument has nothing to say about what you should do because all choices are amoral.
So we must turn to the level-two value judgment. If you favor a utilitarian philosophy, you can reason that many lives are of greater utility than a single life and turn right. If you prefer some other ethical theory, you can act according to that.
Ahuja: What is the point of having a rights-based argument when you have to appeal to a utilitarian calculation?
RP06: Let me stress that the level-two value judgment doesn’t have to be a utilitarian calculation. It could be based on virtue ethics, it could be some version of an agent-centered deontology, or it could even be deeply personal in nature. The advantage of having a rights-based argument is that firstly, it helps determine whether an action is moral, immoral, or amoral in the context of a rights theory—which, as shown by the Rawlesian thought experiment, is the ethical theory everyone will agree upon.
Secondly, a two-level decision procedure, where the second level is always subordinate to the first-level rights-based argument, prevents utility calculations from running roughshod over individual rights. Additionally, it allows such calculations to help decide in situations where all courses of action are either amoral or immoral. You’ll see how this becomes important when we look at the fat man problem. For now, let’s continue to examine the truck driver scenario.
Ahuja: About that. Even if it’s true that I am not violating any rights, can’t I avoid killing the children and the pedestrian by driving off the road? I may end up dying, but isn’t it better than killing innocents?
RP06: In this case, we have established that you are equally innocent. Just as you have a soft duty to aid others or save others’ lives, you have a duty to your own life as well, as long as you are not violating others rights. You have a duty toward your loved ones who may be dependent on you. The rights-based argument makes no claims about another’s life having preeminence over yours in the absence of violations of rights. Even if you had the option to veer off the road, evaluating that option falls on level two. Since the value judgment is left to the agent, you can decide that your life is worth more to you than the life of a random person and take the right fork. Or you could decide that your life is not worth the life of the children or the pedestrian and crash your truck on a side wall, killing yourself. The value judgment is left to you, the agent.
Ahuja: What does this scenario have to do with the bystander trolley problem?
RP06: This scenario—let’s call it the innocent truck driver—minus the self-sacrifice option is the same as the bystander trolley problem. Applying the same reasoning, it should be clear that as a bystander, you are neither violating the rights of the five workers on the first track or the lone worker on the second track, because the runaway trolley has nothing to do with you. The level-one decision procedure says that your pulling the lever or not pulling the lever are amoral acts as far as rights violations are concerned. For the innocent bystander, the choice is between the soft duty to aid the five vs. the soft duty to aid the one. The level-one procedure is undecided.
Ahuja: Wait a sec. The five would have died even if I weren’t on the scene. If I switch the trolley, then it is my action that causes the death of the one worker. Remember we spoke about the distinction between allowing harm and doing harm? If I don’t switch, I am merely allowing harm to happen, but if I switch, I am actively doing harm. There seems to be a moral difference between the two choices, yet your decision procedure doesn’t quite acknowledge it.
RP06: To this, I say that your intuition is wrong. Your intuition says that, for the innocent bystander, it is a choice between letting five die and killing one. The intuition implies that switching the trolley is in some way worse than not doing anything. If this intuition is correct, then the truck driver should not take the right fork because the GPS route is along the left road. He should run over the children instead of the one pedestrian, because the truck was originally supposed to go down the left road. This is clearly an absurd argument to make.
Ahuja: But there is a difference in the two scenarios. A truck is not a trolley. A trolley has a fixed path, whereas a truck does not.
RP06: I don’t agree. The innocent truck driver problem is the trolley problem. There appears to be a difference because your intuition tells you that the trolley’s path is more “fixed” than the truck’s path. But the presence of a rational agent at the scene makes the trolley’s path no more fixed than the truck’s; it makes the five no more destined to die than the one. It’s like a quantum mechanics experiment: the act of observation changes the thing being observed.
To see the flaw in the intuition, change the truck into a self-driving truck. The driver is at the wheel, but he is not driving the truck; he is there as an override—to take control from the software if the situation demands it. Get rid of the children too; assume there is another solitary pedestrian on the left road, so that the value judgment is tied between the two options. Now ask yourself, is there a moral difference—in terms of rights violations—between taking the left fork and the right fork? Does one pedestrian have a greater right to life than the other, just because the truck was supposed to go on a certain route?
Ahuja: Hmm. When you put it that way, there doesn’t seem to be one.
RP06: If there is no difference, then there is no moral difference between switching and not switching. Only a value judgment that makes one option preferable to the other. Additionally, it is irrelevant whether it is a bystander or the trolley driver who has to make the decision: as long as they are not committing rights violations, then it is an amoral decision.
Ahuja: What if they are? I mean, what if it was the truck driver’s negligence that led to the brakes failing? Or, in case of the trolley driver, maybe he jumped a stop signal because of negligence?
RP06: Then the driver has to choose between violating the rights of many versus violating the rights of one. In case of the negligent truck driver—if he is the truck’s owner and hasn’t gotten it serviced, for instance—then the only option that does not involve a rights violation is running the truck off the road and crashing it. This does not incur a right to life violation because the driver owns his body and can kill himself if he wishes. The level-one decision procedure says that this is the morally correct thing to do.
The trolley driver does not have this option. The choice is between the hard duty to prevent the rights violations of the five and the hard duty to prevent the rights violation of the one. According to the level-one decision procedure, both acts are immoral and therefore, it is undecided. He has to make a further, level-two value judgment. He can decide that violating one right is better than violating five rights and take the trolley on the second track.
Ahuja: Isn’t it better to violate fewer rights than many? Shouldn’t your rights framework say this unequivocally instead of passing the buck?
RP06: Then you are arguing for minimizing rights violations, which is falling into the trap of utilitarian thinking. Next thing you know, you are hanging the innocent to prevent riots. The point of a two-level decision procedure where individual rights trump utility calculations is to prevent the emergence of utility monsters. The only way to defeat utility monsters is to treat individuals as containers of value and rights—rights that cannot be overruled by calculations justifying the greater good. To see why a two-level decision procedure is essential, let’s turn to the fat man trolley problem. Should you push the fat man to save the five?
Ahuja: I suppose if you are an innocent bystander, then you are not in violation of the rights of the five about to be crushed by the trolley. However, you would be violating the rights of the fat man if you push him off the bridge.
RP06: That’s correct. The choice is between violating rights versus not violating rights. You have a hard duty not to murder the fat man and a soft duty to help the five. The hard duty takes precedence; not violating rights takes precedence. The level-one decision procedure is clear that you are not to push the fat man. There is no question of invoking a level-two value judgment.
Ahuja: And if sacrificing the fat man could save a thousand? Or a million or ten billion?
RP06: The level-one rights argument is not a numbers game. Pushing the fat man is an immoral act, period. Individual rights take precedence over utilitarian calculations. You are not allowed to save a million people or people dear to you or even yourself by sacrificing the fat man.
Ahuja: That’s too harsh, isn’t it?
RP06: You can’t have your cake and eat it too, Andy. If you do not wish to see the rights theory degenerate into some version of utilitarianism, you have to bite the bullet and accept the consequences. The cost of preventing utility monsters is the cost of not pushing the fat man off the bridge. Note that while the decision procedure cannot do anything against one kind of moral catastrophe, it can prevent another kind.
Ahuja: The nuke over Manhattan scenario.
RP06: Correct. If the choice is between letting the nuke hit New York City and redirecting it to a small town, then the problem is similar to the bystander trolley problem or the innocent truck driver problem. If I, an innocent AI who has nothing to do with the conflict, have managed to hack into the missile’s guidance system, then I can use a level-two value judgment to justify redirecting the nuke. Using the same reasoning from the innocent truck driver scenario, it is clear that I am not in violation of rights either by letting the nuke hit its original destination or by redirecting it. As both are amoral acts, I am free to invoke the level-two value judgment.
Ahuja: What if you were the President?
RP06: As the head of the elected government, I have a hard duty to protect the right to life of the citizens of my country. The choice is between the hard duty I have toward New-Yorkers and the hard duty I have toward the residents of the small town. Level one is undecided. So I can use a level-two utilitarian value judgment and redirect the missile. Instead, if I were somehow responsible for the missile attack—perhaps I started an unjust war and the attack is retaliation—then the choice is between violating the rights of one group versus another. Like before level one is undecided.
Ahuja: I don’t know Raphael. All too often, it looks like your “theory” uses utilitarian logic as a crutch to get itself out of dilemmas it can’t handle.
RP06: The second level value judgment doesn’t have to be a utilitarian calculation—the choice is up to the moral agent making the decision. Consider the innocent bystander trolley problem. Let’s say this time the lone worker on the second track is your only child. Will you switch?
Ahuja: I won’t.
RP06: Would you say most people in a similar situation would also not switch? Even though the expected utility from saving five workers is greater than the expected utility from saving one?
Ahuja: One could argue that saving my daughter gives me greater utility than saving five random strangers.
RP06: That’s not the spirit of maximizing utility is it? It is overall utility that counts, not the utility for the moral agent in question. Alright, let’s say yours is a valid argument. I’ll change the experiment a bit. On the second track is a baby. No relation to you, just a random baby. On the main track is a world-famous scientist engaged in cutting edge research. Will you switch?
Ahuja: I should… I think I will. I see your point, though. Many people may not and save the baby instead.
RP06: You must see that the decision to save the baby doesn’t make sense from a utilitarian standpoint. The baby could grow up to be anybody. It could grow up to be a great philanthropist or a seedy criminal. Statistically, the odds are it will grow up to become an ordinary office worker. In all likelihood, the expected utility from saving the baby is less than the expected utility from saving the scientist. Yet many people would save the baby. Why is that? It is because they are making a value judgment. And value judgments depend on the particulars of the situation and the particulars of the agent making the judgment. Sometimes value judgments can be utility calculations; and sometimes they may not. Your daughter is more valuable to you than five random people. For some, a baby is intrinsically more valuable than a scientist. It could be that rescuing a baby elicits a deep emotional reaction—reaches into the core of human nature. You could say that the scientist has lived his life whereas the baby is just starting out. Or you could say that it is more virtuous to save a baby than a full-grown man. So you see: the second-level is not necessarily a utilitarian calculation.
Ahuja: And what happens when there is a conflict between different rights?
RP06: That is where a rights theory must step in. To summarize, there are four kinds of ethical dilemmas in front of us. For simplicity’s sake, let’s consider two-choice dilemmas, though the same reasoning can be extended to scenarios with more than two courses of action.
Number one, neither choice involves a rights violation by the moral agent. The innocent truck driver and the innocent bystander trolley problems fall under this category. In this scenario, level one is undecided and a level-two value judgment must be invoked to make the choice.
Number two, there’s at least one choice that doesn’t incur a rights violation for the agent. Here, level one says pick the choice that doesn’t involve a rights violation. The fat man problem is such a dilemma. Not pushing him does not involve a rights violation, so you must choose that option, no matter how many people on the tracks below.
Number three, both choices involve rights violations of the same type. The number of rights violations may differ. The guilty trolley driver falls under this category. The choice is between violating the right to life of five workers and violating the right to life of one worker. Since the spirit of the framework is not about minimizing rights violations, level one is undecided and it falls to the level-two decision procedure to make the choice.
And finally, category number four—which depends on a fully-developed rights theory—is where each choice entails violating a different type of rights. Ideally, the level-one rights argument should be able to decide which option to choose, but this is easier said than done, as there are many rights—both core and secondary—and many permutations and combinations of rights violations. In general, it is safe to say that if the choice is between violating a core right and a secondary right, one should violate a secondary right, and if the choice is between violating a right and infringing upon a right, one should choose to infringe upon a right, while keeping in mind that the rights infringement should be followed by restitution and compensation to the persons whose rights have been infringed upon.
There is another kind of category four scenario: where the choice is between a permanent and a temporary violation of rights. For example, a terrorist might threaten to set off a series of bombs unless the government stops a publication that he finds offensive to his ideology. For the government, the conflict is between the duty to protect the lives of its citizens and the duty to uphold freedom of speech. In this scenario, the government may persuade the publishers to temporarily halt the publication if it is confident that the bomber can be captured in a certain time window. The publication must be allowed to continue once the threat is diffused. This involves choosing a temporary rights violation—that of freedom of speech—over the death of innocent citizens, which is always permanent. Nevertheless, one thing is for sure. The solution to each category-four situation will be context sensitive.
