Dilemmas in Matters of Human Life – Lesson 13

This transcript was produced automatically using artificial intelligence. There may be inaccuracies in the transcribed content and in speaker identification.

🔗 Link to the original lecture

🔗 Link to the transcript on Sofer.AI

[Rabbi Michael Abraham] In cases of saving life, do my relatives get preferential treatment over others? We saw that the answer is no. Only in situations of rescue. Meaning, if say there are two sick people and I need to spend money to save one of them, then of course there is legitimacy to spending the money on my relative and not on someone else—your own life takes precedence. But when the question is whether to sacrifice one person for the sake of another, it makes no difference how close he is to me. You can’t kill someone distant in order to save someone close to you. There’s no such thing.

Okay, in any case, in the first question we were talking about similar questions, but there’s an interesting twist here. In the first question, yes, we’re talking about a road, and in front of the car there are many people. If I swerve to the side, then I’ll kill one person and save the many people. So the quantitative consideration, the consequentialist consideration, says to swerve.

But before that, what would happen if there were one person in front of me and I swerved and killed another person? What do you do in that situation? Passive omission is preferable, right? In other words, in the end we saw that there’s no justification. I started this whole discussion with the difference between a pursuer on the one hand and saving oneself at the cost of another person’s life on the other. And the claim was that if they threaten me to kill someone, and if not they’ll kill me, I’m forbidden to kill him. Meaning, I have to die and not kill him.

On the other hand, if someone threatens me, I can kill him in order to save myself, or someone else can kill him so that I’ll be saved. That is the law of a pursuer. Now, what do you mean, “Who says your blood is redder?”—when they threaten me to kill someone and if not they’ll kill me? So I brought the Tosafot in Yevamot. Tosafot says that there, if I’m falling from the second floor and below me there’s a person such that if I fall on him, then he’s probably finished—even after a diet. And if I move myself to the side, then of course I’ll smash into the ground and die, and he’ll be saved. The question is whether I have to tilt myself to the side. Seemingly yes, because after all I’m forbidden to kill one person to save my own life, right? What do you mean, “Who says your blood is redder?” If they threaten me and I can kill someone in order to save myself, it’s forbidden. I have to die and not kill him.

But that’s not correct, as Tosafot says, and it seems obvious to me that it’s not correct, because in the first case, when they threaten me so that I kill someone else, I have to perform an act of killing. I kill him in order to save myself. There is no justification for killing him. In order to perform an act, you need justification. And if I’m going to kill someone, I need justification. Now, what’s the justification—that I’ll be saved? My blood isn’t redder than his. But that doesn’t mean that the other person’s blood is always redder than mine. True, my blood isn’t redder than his, but his blood also isn’t redder than mine. Right?

So if I’m there—if I’m falling and the person is down there—now, if I do nothing, I fall on him, he cushions my fall, I stay alive and he dies. That’s the situation right now. If I have to tilt myself to the side in order to save him, what I actually have to do is perform an act that kills myself. There’s no justification for that either. After all, his blood isn’t redder than mine, just as mine isn’t redder than his.

Doesn’t it depend on whether the situation is voluntary? For example, if you dove headfirst?

No, no, obviously it does. I’m not dealing with a voluntary situation, because then you’re just a murderer—in the first case, a clear murderer. I’m not talking about that. I’m talking about a case where I fell, they pushed me, I don’t know, something happened. Now in such a case, I’m saying that the consequentialist consideration is a consideration—even though I went deeply into it—but it’s a consideration whose significance is not absolute. Meaning, you have to pay attention: it’s not always true that the outcomes determine the situation. In this case the outcomes are one life here, one life there, in both cases. And still, if they threaten me so that I kill someone, I’m forbidden to kill him. But if I’m falling and I have to tilt myself to the side to save him, thereby killing myself, I’m also forbidden to do that. Not only am I not obligated—to most halakhic decisors I’m forbidden to do that. I’m forbidden to do that because to kill myself also requires a reason. The other person’s blood is not redder than mine, just as my blood is not redder than his. The rule is basically: passive omission is preferable. Once you have one life against another life, there’s no justification for doing anything. Whatever happens, happens. You can’t do anything, because the moment you do something, you have killed—yourself or the other person. Those are the facts.

Now the question that comes up here is: what happens if someone is facing a group of people—five people, ten people—and I’m driving toward them, and if I swerve to the side, then I’ll kill one person. Now here the consequentialist calculation is already ten versus one. Before, we were talking about one versus one, and then I say passive omission is preferable. One versus one is an even balance. There’s no justification for doing anything; I’m supposed to let the situation do what it does. But if there are ten versus one, then a consequentialist consideration seemingly prefers the ten over the one. And here it becomes a real dilemma.

If it were ten versus one in a neutral situation—say I had to give water either to ten people who are about to die of thirst on one side, or to one person who is about to die of thirst on the other side—and I can’t get to both of them, so I either send the water there or send the water there, in that case I have no doubt that you have to send it to the ten. I think that’s also the topic in Horayot, about precedence in life-and-death matters. But in a case where I’m driving toward ten and I need to turn the car aside and then I’ll kill one person, that’s already not a simple question. Because on the one hand the consequentialist consideration says: turn the car. On the other hand, turning the car is an act. It’s my act. And by what right do I give myself permission to sacrifice his life in order to save their lives? If he decides to sacrifice his life to save them, fine. But I can’t decide for him that he must sacrifice his life to save them.

Yes, but what if the ten are old people who in any case have only two days left?

That doesn’t matter. We talked about this—that the value of life isn’t measured by life expectancy. In other words, life—say in a simple model—has infinite value. So it doesn’t matter whether you have one day or a year or something like that. We discussed a responsum about short-term life, and there we did try to define some lower value for short-term life, but again, that still wouldn’t justify any act.

What happens if you’re falling? They pushed you, and you’re about to kill ten, and you can tilt yourself and kill one?

Same thing, yes, same situation. That’s the situation here. In the discussion there from the Jerusalem Talmud with the enemies—we’ll get to it in a moment; we’ll get to the Jerusalem Talmud too.

Yes.

Driving is not one specific act; it’s an active act every moment. In practice I’m acting the whole time as I drive, I’m constantly pressing the gas, so I can’t say I stopped acting, as in falling.

No, no, in this case that’s exactly the point. Since you’re currently moving at high speed, the situation is that even if you take your foot off the gas now, there will still be impact.

Right, but I brought myself into this situation, unlike when I jumped out the window.

No, no. If you brought yourself into a situation where you knew you were going to kill them, that’s something else. But you brought yourself into a situation where you didn’t know. I jump out the window, I didn’t look down because I’m currently committing suicide, and then afterward I look and see there’s someone there. Okay, so now it’s not like a case where they pushed me—I brought myself into the situation.

You didn’t know. And here too you didn’t know. But still there’s a distinction: when you jump out the window, you’re not doing a normative act. When you’re driving on the road, you are doing a normative act. You’re not supposed to think there’s someone there. But it’s a disruption on the road, a disruption on the road—people are on the road, what can I do? I’m doing a reasonable, permitted act. I’m driving on the road. What can I do? I also have no intention to kill anyone—not even myself when I get on the road. I’m just driving.

Yes, right.

In any case, the claim now is that there is a dilemma here. I said that at the very beginning I spoke about consequentialist considerations. I said that the accepted halakhic approach is that the results don’t have that much significance; the significance is the act. Therefore, for example, the Jerusalem Talmud that you mentioned earlier—we mentioned that Jerusalem Talmud about gentiles who besieged a city and say: give us one of you, and if not we’ll kill all of you. So it says in the Jerusalem Talmud that you do not hand anyone over, and that is also the halakhic ruling in Maimonides. And that’s also how people argued with me about Siamese twins; I brought all those situations that they compared to the Jerusalem Talmud. And the claim was that when we hand someone over to those threats, those besiegers, those who are besieging the city, that is, then we are doing some act of killing. And therefore, even though we are sparing many lives, it’s forbidden to do that.

And there I said that I protest against this; it cannot be right. It can’t be right, first because the consequentialist consideration is so significant that it’s impossible that we ignore it, and second because even if we do not hand him over, he too will die—he will die together with us. After all, the gentiles will kill all of us, including him, if we do not hand him over. So if he is going to die anyway, then all that remains is the question whether to hand him over, in which case he dies—which in any event is what will happen to him—but at least we’ll be saved; or not to hand him over, in which case we all die. In a case like that, it seems to me obvious that you have to hand him over. But in the Jerusalem Talmud and in Maimonides it doesn’t say that.

So I talked about this and said that I think what’s written there is a law in the laws of sanctifying God’s name, not a law in the laws of preserving life. And indeed, in Maimonides it appears in the Laws of the Foundations of the Torah and not in the Laws of Murderer and Preservation of Life. And the claim is that this surrender to terror is a surrender that one must not make. In other words, like desecration of God’s name—but you can also see it in not necessarily religious terms. Rather, the idea is that a community does not allow itself to be seen as one that hands over one of its people to such terrorists, to those who are going to kill them. It’s a public consideration that we are not willing to do this.

But in the private consideration of an individual person—for example, in fetal reduction or Siamese twins, which are situations exactly like that but on a completely private plane—no one is threatening, there is no sanctification of God’s name and no desecration of God’s name. In such a case, it is obvious that one hands over the one who will in any event die, in order to save all the others, so that at least they survive. And certainly if we do a lottery, then each one can be given a chance. But I’ll come back to that again.

Why if they single someone out do they say yes, you hand him over, right?

Because then it’s also sanctification of God’s name?

First of all, even when they single him out, according to Jewish law it is still forbidden.

Forbidden?

No, only if he is liable to death. And even then, if he is liable to death by the kingdom, even then Elijah the Prophet says to Rabbi Yehoshua ben Levi that this is not an act of piety—even when he is liable to death and they singled him out. But as a matter of Jewish law there is a dispute in the Jerusalem Talmud about this; that’s the interpretation there.

So I claim—and I said that Professor Anker wrote this in an article, I think he is the father of this idea—I think he is entirely correct. This is not in the Laws of Murderer and Preservation of Life; it is in the laws of sanctifying God’s name. Therefore in the same situation, as with fetal reduction for example—in fetal reduction, the mother has many fetuses in her womb, three, four, I don’t know how many, and the doctors’ assessment is that they cannot all survive, it’s too crowded, they will all die. The question is whether I’m allowed to remove one of them, to “reduce,” as it is euphemistically called in these contexts. I remove one fetus so that the other two or three can remain alive. That’s the situation. So the accepted claim is that in principle this is forbidden, in light of the Jerusalem Talmud in Terumot: “Give us one of you”—it’s exactly the same situation. Rabbi Zilberstein writes, by the way, that it is permitted, but only because they are fetuses, since with fetuses there is no prohibition of murder. So therefore—but on the principled level, if there were such a situation with people after birth, people who already have full value…

full value of life—then it would be forbidden to do this. And I claim that it is permitted to do this; it has nothing to do with the Jerusalem Talmud. In the Jerusalem Talmud it is the laws of sanctifying God’s name. In fetal reduction this is not sanctifying God’s name. I’m not surrendering to anything. I’m simply trying to save the maximum number of lives. To hand everyone over to die is simply absurd in my eyes. To hand everyone over to die because I’m not willing to remove one, when that one too will die. It’s not that I’m saving him; it’s that I’m killing someone. I’m killing someone who in any case will die.

The same thing with Siamese twins: to perform separation surgery on Siamese twins when the alternative is that both of them die. If both are going to die, then obviously I need to do the separation surgery, and one of them will have to die, but at least the other will remain alive. And all the halakhic decisors forbid it. With Siamese twins, by the way, these are no longer fetuses; these are twins who were born. So there the situation is exactly that. All the halakhic decisors forbid it. I spoke with people who checked this with all the major halakhic decisors. Everyone forbids it. In my eyes that’s outrageous. I mean, it simply cannot be right.

But here our logic is that it’s not necessarily the case that the one person will die, right? In the example you gave where there are seven people against one.

No, the assumption is that he will certainly die. For purposes of the discussion we’re talking about certainty.

Yes, he will certainly die. There’s no doubt here—you remove him for sure. When the halakhic decisors forbid it, it seems to me that the theoretical assumption, when there is no probabilistic consideration, never really exists in practice.

No, no, it’s not about probability. Why probabilistic? Everything is deterministic.

There is never complete certainty that he will die, and because of that lack of certainty there is—

No, no. So I brought the Lechem Mishneh on Maimonides. The Lechem Mishneh on Maimonides indeed writes about this. He says: don’t hand over the one. He asks: how can it be that you don’t hand over the one? After all, “Who says your blood is redder?” That’s the reason one may not kill one person in order to save oneself. But here that one whom I am killing will in any event die too. So that reasoning of “Who says your blood is redder?” doesn’t exist. So the Lechem Mishneh asks that. Everyone asks that.

So the Lechem Mishneh answers it. He says: don’t hand him over to them—who knows, maybe afterward they’ll change their minds and not kill us. Fine, that’s a technical solution. I’m asking conceptually. Let’s say it’s certain they’ll kill us, okay? A heavenly voice came out and said they are going to kill everyone. The question is what I am supposed to do in such a case. Beyond that there are the considerations of doubt, which of course can be taken into account.

When the halakhic decisors—because the theoretical case never exists—they hold this of course in theory but not in reality. Their justifications are not because of the concern that maybe it won’t happen, but because we cannot hand over one person. And again, what does that mean? It means that in the halakhic world, the practical consideration is a more important consideration than the consequentialist one. The deontological consideration is more important than the teleological one. Meaning, we don’t merely calculate where more life will be gained; rather, the question is when we are committing an act of murder. And supposedly I want to remain clean—I didn’t murder. The question is not how many people will remain alive and how many will be lost, but how I come out looking good. Of course I’m presenting it too crudely. But that is basically the approach. The approach is: when I stand before the heavenly court, can it be said that I did not murder? No problem, I’m clean, I didn’t do anything. But in order for you not to murder, many people lost their lives on their own. You didn’t murder them.

So the question is whether I need to calculate how much less I will murder, or whether I need to calculate what should happen in the end—how I will save as many lives as possible or prevent as little loss of life as possible. Okay? Therefore I claimed throughout that the consequentialist consideration is in fact a foundational consideration. Where the consequentialist consideration is unequivocal, I do not think there is room for hesitation because of the practical consideration—the question whether I am performing an act of murder or not. Ultimately the goal is to save lives. I brought various proofs for this from within Jewish law as well. This is not the accepted approach in Jewish law, but in my opinion there are proofs for it that are hard to argue with.

So at the level of an autonomous car, that will be the distinction between an autonomous car and a driver?

Wait, wait. I’m not yet at the autonomous car, one second. For the moment I’m still with a live driver.

As far as the car is concerned, they won’t accuse the car of murder. They’ll accuse the programmer.

I’ll get to that in a moment. I’m doing this in steps—precisely because of that point. First let’s see what happens with an ordinary driver; afterward we’ll see what happens with an autonomous car. Okay?

So with an ordinary driver, what happens now when there are ten people in front of me and one person off to the side who will be killed? The first question is: in such a case there is a dilemma between the consequentialist consideration and what we may call the practical one—whether I will be performing an act of murder. Because from the consequentialist point of view, if I perform an act of murder—if I swerve to the side—then the cost is that I am performing an act of murder, but the benefit is that I saved ten people rather than one. So here we have a case where the consequentialist consideration is in conflict with the practical consideration. And here the question really is not simple.

There are halakhic decisors—I mentioned Tzitz Eliezer, for example—who want to claim that while it is true that I may not kill the one in order to save myself or in order to save someone else, if it is many other people, then maybe it is permitted. Tzitz Eliezer argues, as far as I remember, that this is only with regard to a public, not merely many people. So here there is not only the question of how many people will be saved, but once there is a public interest of three or ten people against one—I think, as far as I remember at least—and he argues that it is still forbidden; it is like one versus one. The claim is not about the number of people but the question whether it is a public or just a collection of several individuals. Part of this may be a slippery slope, because the question is what counts as one versus two, two versus three—where do you draw the line?

So here is the difference between the people in the road, who committed an offense, and if it’s on the road versus the one standing on the sidewalk?

Interesting question. The legitimacy of the sidewalk? Sharing responsibility? I don’t know what to say. Interesting question.

But the question is: what happens if the ten people on the road violated the traffic laws? They came down into the road carelessly. Now as a result of that I have to swerve the car and kill someone. Where is the dilemma? The dilemma is that true, violating the law does not carry the death penalty, but on the other hand that other person also isn’t guilty. When they break the law, they cause him to die. So if they broke the law, let them bear the consequences.

Is there a ruling like that?

I don’t know. I don’t know.

Here’s an example of a decision an autonomous car is supposed to make, but the problem is not the autonomous car—I don’t know how I’m supposed to program it. Meaning, I don’t know how to make that decision. Okay.

But the question is what would happen if I were driving myself. That’s why I say again: this does not mean that because of this you shouldn’t let the car go on the road, because otherwise I also shouldn’t be allowed to go on the road. Okay.

So here I say there is a dilemma between the consequentialist consideration and the practical one. And in this case the halakhic decisors are divided. I don’t know what to say in such a case. I tend toward the consequentialist consideration, but I don’t know how to justify it all that well. Meaning, I think in light of the considerations I brought at the beginning of the series, when I talked about the preference for the consequentialist consideration against the accepted view that speaks about whether I am performing an act of murder or not, then I would say that in this case too I would think one should care for the ten and not for the one. I’m not talking right now about when they committed an offense, but generally.

The second question is: what happens if a person stands in front of me, or I myself will die? Do I have to kill myself in order to save him? Here it is really similar to the case of Tosafot in Yevamot, right? Someone who is falling—a person is not supposed to kill himself in order to save the other person. Your own life takes precedence. Meaning, I cannot kill someone else in order to save him; I can’t tilt this so as to kill someone else in order to save him—that is certainly forbidden to me. Even if the one who would be killed is me, that too is forbidden to me. Because what—my blood is not less red than someone else’s. It’s not more red; I have no privileges. But it is also not less. So just as I’m forbidden to kill someone else in order to save him, I’m also forbidden to kill myself.

And the third question is: what happens if it’s a matter of killing myself in order to save many? And again, this is the same dilemma, because in my view it’s the same question as killing one person in order to save many. But there is a difference. Why? Because if the difference lies in the laws of doubt—meaning if there were a clear ruling, it would be the same for me and for another person. But if there isn’t a clear ruling, then I can say, look, anyone who wants to volunteer can volunteer, but I’m not willing to be killed in order to save the other ten. If I were obligated, then there is no wisdom and no counsel—I’m obligated; Jewish law obligates me to die in order to save them. But if I’m not obligated, and it’s a matter of doubt, there is no clear ruling on this. So true, maybe I can volunteer and die in order to save those ten people—maybe I can; even that is not agreed upon—but to say that I am obligated as a result of this, even though in the laws of doubt perhaps with respect to someone else I would have to, with respect to myself, your own life takes precedence.

If it’s an autonomous car, then the programmer is the one deciding.

Right. Now I’m getting to the autonomous car. Up until now I was talking about the driver being a normal driver. Now what happens when we talk about an autonomous car? I was simply trying to go in stages.

Here it’s like someone giving you an order.

Yes, but I said: I’m driving in the normal way on the road; that’s not called that I performed the act. If I know there are people there or I did something careless, then maybe it’s more similar. But if I’m driving as usual on the road—this is what the road is for. Now I’m in the situation.

I’m coming back from a course for traffic offenders, and there they say that if you’re involved in an accident, there is always a significant percentage of responsibility on you because you were supposed to drive in a way that prevents any accident.

Okay. Meaning that following what this gentleman said, my driving is an active act that endangers others by the very fact that I am on the road.

Fine, that’s words. In the end, if I’m driving in a normative way, that’s what people do, the road is intended for that. I’m supposed to prevent an accident, and there is always… it’s not completely clean. I’m saying: you’re supposed to prevent an accident and you can’t. What do you do with the fact that you can’t? Wait, no—we haven’t gotten to the autonomous car yet, we’re talking about a driver. Now I’m moving to the autonomous car.

What happens with an autonomous car? All right, let’s continue. In an autonomous car, there are two additional factors that enter. First, the car makes the decisions, not me. Okay? So the question of “your own life takes precedence” and so on is less relevant here. Why? Because I’m not the one deciding whether to swerve or not. The question is what the programmer is supposed to put into the car’s software. And in a certain sense, when the programmer writes that software, in the end, in a certain sense he is the driver. Meaning, he is in that situation and he is the one making the decisions—he just makes them in advance.

There’s no end to this.

Okay, so what follows from that? No, when you program the car, you need to examine whether the software—you are taking responsibility for that. Here you’re introducing another parameter. This additional parameter is the question whether we establish a standard, and then you can’t decide anything. The government decided, or the consensus—you buy this. So buy it—but others buy it; that’s not important. People buy such a car; that’s the autonomous car there is, there is no other car. Don’t buy a car because cars cause traffic accidents? You know that cars cause traffic accidents? Then don’t buy a car. If you buy a car, the point is that there are things human beings do; this is the normal way to live, and there are costs, what can you do. If a person drives on black ice now in Minneapolis, then you’re putting him on the ice—how does control of the car work? Do you turn left or right? So what’s the question then? Then say Ford is responsible because the wheels were such-and-such? Ford is not responsible because you committed a negligent act; you drove the car on ice. What does that have to do with anything? But that’s what the car is for. You don’t have control, so don’t drive there—what do you mean? No, fine. Let’s continue.

With an autonomous car, what we are really talking about is the question of the programmer, not the question of the driver. Now the question is to what extent I can treat the programmer as the driver. Is the programmer the driver? On the one hand, seemingly yes, because in the end I’m talking about that aspect in the software that deals with these specific situations. Meaning, it’s not that there is some general software here and in certain situations this software, which is supposed to drive the car in all situations, reacts in this situation in this way as a side effect—as some outcome that emerges. No. I’m talking now about the programmer putting into the car a specific element in the software that addresses this situation. In this situation he tells it: do this or do that, okay?

What practical difference does the question make right now? Because if I’m talking about general car software and in these specific situations that is what the software will produce, but it’s not that he specifically thought about this software in advance, then it’s not relevant. The programmer is not putting a particular software into the car. He tells the car: drive this way. It’s like a car manufacturer. Our halakhic question right now on this topic is what he must program, or how. I am the programmer now. Yes. Because after all there is no question what the driver will do; the car does it, not the driver. Okay? So now I’m saying that my assumption is that he inserts into the car a specific instruction for this situation. Not that this is a situation to which the car’s general software simply responds, because that is something else. That is something similar to a regular car. A regular non-autonomous car that we drive also sometimes doesn’t respond well in difficult situations, but that’s how it’s built; it’s not that the manufacturer built it negligently.

That’s indirect causation.

The question of the programmer—exactly, that’s the question. That is exactly the question. But the question is whether it is like an arrow, or whether he programmed properly and then it isn’t called an arrow even if it kills. It’s exactly the same thing. So the question is whether the programmer is essentially considered the driver. In principle yes, I think. In principle yes. More than that: in the end the programmer works for the driver. Meaning, he is the driver’s agent. True, he doesn’t know him at all, but he is the driver’s agent. The driver buys a car from him or buys a service from him; he built the car for him, and in the end the responsibility belongs to the owner of the car. You bought a car whose software is this software. The programmer says: I built a car. If you don’t want it, don’t use it; if you want it, use it. I built a car, and you knew everything. Meaning, you knew what the car does, assuming he knows everything. Taking the car is your fault, or else don’t enter such situations, or take responsibility.

And then there is room to say that the programmer is not the driver, because in the end he is only indirect causation. The programmer is only indirect causation. He built the car; you took the car and used it in a certain way, and therefore it is actually hard to speak about the programmer as the driver.

More than that, many times when we are before the problematic situation, we make different calculations than when we are inside the problematic situation. For example, regarding autopsies there are all kinds of discussions. Is it permissible to dissect the dead in order to train doctors or in order to save lives, for medical research and things of that sort? There are many halakhic decisors who say no. They say: if right now there were someone who needed an organ from this dead person, which is already a case of saving life, which overrides all the prohibitions in the Torah, then take from this dead body what you need and give it to him. But for a future situation that may happen, to do something now that is certainly a transgression, so that in the future perhaps lives will be saved—who knows whether the doctor will learn anything from it, and whether something he learned from dissecting this corpse will help him save lives someday in the future? You can’t know. And therefore the present situation does not justify transgressions for the sake of uncertain future benefit—and perhaps not even uncertain, but future benefit. Therefore also in the context of the autonomous car, when you program the car, in the end you are still far before the problematic situation. And the question is whether you are supposed to make calculations now about what will happen in the future.

But there’s a difference, because there we’re talking about something I do for a certain benefit, yes? And the price is—I do it in order to gain human lives in the future. Say a doctor who learns by dissecting a body; then in the future perhaps he’ll succeed in saving a person’s life because of what he learned, and right now he’s committing a transgression, right? He is harming the dead, violating the honor of the dead, not burying certain organs—there are several transgressions involved. The question is whether it is permissible to commit a transgression now for the sake of some future benefit.

In the situation of the autonomous car, that is not the case. In the autonomous car situation, right now I’m doing some act that is not a transgression and nothing of the sort; I am asking what I should do. And it’s not… therefore in this context, the future situation is the situation that stands before me. It is the situation standing before me because in the end, as I said earlier, I am now putting into the software some part that is supposed to deal with these situations. So from my standpoint, that’s what I am doing right now. There is no dilemma here about whether it is permissible to commit a transgression now in order to gain future benefit, or gain benefit now at a possible future cost. I am not gaining any benefit now, except perhaps financial profit. I simply want to know what to do. That’s all. Therefore I think that precisely in this context, the time gap is not relevant.

Especially since those programmers, or that software you buy, has some kind of disclaimer. Okay? That makes you a partner in the whole system. Why are we talking about the programmer? What is the programmer? The programmer is not the one who decides. It doesn’t matter who decides, that’s not important to me right now, although perhaps we could also talk about the programmer. Because if my boss says to me: program it this way, but it’s forbidden, then I don’t do it—just as if he tells me to desecrate the Sabbath, I still don’t do it. So it may still be the programmer. The question is whether it’s the programmer, the company, or perhaps the public. So I said: you can say it this way or that way, and that’s another question that is less important for our purposes. I’m saying: in principle, if this thing is halakhically forbidden, there is room to impose responsibility on the programmer too. You should have refused. You should say, I’m not willing to do this, just as if someone tells me to work on the Sabbath—I don’t work on the Sabbath; it is forbidden to me.

In short, the claim is that the programmer is supposed to do what the human driver ought to do. That seems to me basically the conclusion. Now what should the human driver do? That itself is a dilemma. That itself is a dilemma and, as I said earlier, because who knows—there is some conflict here.

We said that the driver may save himself, but the programmer is not saving himself.

No, but the programmer—that’s why I said: once I identify the programmer with the driver, then I am basically saying the programmer works for the driver. Meaning, he sells him a product according to what the driver asks of him, as it were—of course he didn’t explicitly ask him, but that’s the product. And there is a question in a case where either he himself is killed or he kills many people. I said: the programmer is the driver; the programmer works for the driver. So the driver says to the programmer: do it in such a way that I don’t crash in order to save them. That’s what the programmer does. That’s the end of the question they asked, where they said that most people say one should kill the one in order to save the ten, but we don’t want to ride in such a car. The claim is that if you don’t want to ride in such a car, then what would you do? Whatever you would do is what you are telling the programmer to build for you. This is how you want him to build the car for you, since the programmer works for me.

It seems to me that technically the driver can always override the software and take the wheel into his own hands, so maybe all these questions…

Fine, again, I’m not getting into technical questions because for me what matters are the principled questions. Is that clear? Clear, yes. Okay.

The Chazon Ish in this context brings here—that is the Chazon Ish I mentioned earlier regarding that accident in Bnei Brak—the Chazon Ish also writes… in his novellae to Sanhedrin and also in Yoreh De’ah:

“One who sees an arrow going to kill many people, and can divert it to another side, and then only one will be killed and the many will be saved, and if he does nothing the many will be killed and the one will remain alive—it may be that this is permitted. And it is not comparable to gentiles who said, ‘Give us one of you to kill, and if not we will kill all of you,’ as in the Jerusalem Talmud in Terumot, where the law in that case is that they should all be killed rather than hand over one Jewish soul. And this is not comparable to diverting the arrow, because handing over a life to be killed is a cruel act of murder, and in this act, in which they hand over someone from the city to death, the saving of the others is not inherent in the nature of the act. Rather, it just happens that this act of murder now saves the many. Diverting the arrow from one side to another is, in essence, an act of rescue, and in this it may be that one should strive to save Jews in every way possible.”

This is classic Chazon Ish reasoning. What this reasoning basically says is as follows. I brought this, by the way, also in the article on Siamese twins, because it really is a very interesting point. The Chazon Ish actually dismantles the whole discussion we have had until now. He says: leave it, the whole discussion we’ve had until now is irrelevant. What he says is that the question is: what is the nature of the act that I am doing? The act I am doing right now—I’m speaking again about a human driver, under the assumption that the programmer and the human driver are the same thing—what act am I doing? If I divert the arrow, then basically I am performing an act of rescue, because the arrow is about to strike the ten in front of me, say, or the one in front of me, and I divert the arrow in order to save them. If I divert the arrow in order to save them, I am doing an act of rescue. True, as a result someone else will die, but this is not an act of murdering him; it is an act of rescuing them whose side effect is that someone else dies.

By contrast, when I take someone from the city and hand him over to the gentiles, that is not an act of rescue—on the contrary, it is an act of killing. The side effect is that we will be saved. The question is how I relate to the act, how I see it with simple human eyes—not halakhic eyes or anything else. How does a person who looks at this act see it? Does he see an act of murder here, or an act of rescue?

So then you could kill ten in order to save one?

Exactly. And that’s the next step. He basically argues that the Jerusalem Talmud in Terumot is not comparable to diverting the arrow. And the claim is that with the car it is like diverting the arrow. When you turn the car so as not to run over the ten, you are basically saving the ten. It is true that in the end someone is killed, but you didn’t aim the car in order to kill him; you turned it away from there, and it just wound up there.

And if you aim it in order to kill him?

Then that’s also one versus one, here too. Here’s the question.

Now in light of these things, there is certainly room to ask: what happens if the situation is the reverse, where there is one person in front of me and I turn the car and kill ten? Seemingly the Chazon Ish’s reasoning is the same reasoning. I am only doing an act of rescue. So what’s the problem? The fact that ten were killed—they were killed, but I performed an act of rescue. That’s not true. Why is it not true? Because the question is what counts as an act of rescue. Meaning, when you perform an act in order to save one person, and you know that the price is going to be that, incidentally—it doesn’t matter as a result—ten will be killed, that is not called an act of rescue at all. You do not save one at the price of ten. No, wait, what am I doing? I am saving.

I’m saying the difference is that this is a layered perspective. On the one hand I say I’m not looking at this against that; I’m looking at the question of how the act is defined. But that’s not entirely true. After all, I know that the one or the ten standing to the side will die—it’s not that I don’t know. Rather, right now I am doing an act of rescue, not an act of killing. Fine, but if you know ten will be killed, don’t tell me you’re going to perform an act of rescue for one person when you know ten will be killed. That I’m not willing to define as an act of rescue. Therefore it seems the Chazon Ish would of course not agree to diverting the arrow if the one standing to be killed in front of me is one person and those who will actually die are ten. Even though his definition is one that seemingly fits here too.

Does Barak agree?

Barak agrees. No, there it was one versus one.

No, it said “a person,” so he agrees. Okay, so in the one-versus-one case too, in the two following cases he indeed says “your own life takes precedence,” like the two people walking in the desert with the flask of water. And in the end he concludes that the programmer has the same law as the driver. Meaning, what the driver may do the programmer may also do, because the programmer works for him. In other words, the programmer builds a product for me. I want it—or for me, or it doesn’t matter if it’s a government standard, that’s not important—but the public determines for him what to do. It’s not that the programmer here is the man with the responsibility. Therefore what is permitted to the driver is permitted also to the programmer.

So one versus one is still called rescue?

What? Yes. You said one versus one is called rescue and one versus ten is not. Right—if ten will be killed. I agree with that definition, but it’s post facto, at that moment. No, in a place where it’s one versus one—then I say you can’t say this is rescue or not rescue; you look at the act as such. This act, of diverting the arrow or the car, is an act intended to save him; after all, I didn’t go to kill him. So true, in the end he dies. I’m only saying that where there are ten there, then already at the stage when I turn the car, you can’t say I merely performed an act of rescue. After all, I know ten will die from this.

And this, you understand, is not so far from the ordinary way each of us thinks when I do various things that damage the environment. The environment—I am harming it, I am doing an act that harms the ozone layer. Now true, this harm is small; for an individual person, this harm is of course insignificant. Therefore I say okay, in the end you can’t say this is really forbidden. Why? Because this act is an act of—I’m basically living my life; I’m not destroying the ozone layer right now. I’m living my life, and this harms the ozone layer, true. But in a place where the situation would be one in which many people would clearly die, then you can’t say I’m living my life and many people just happen to die. Right? And therefore the films that make this so vivid—Al Gore and all those people—the films that make so vivid the present damage caused by what we do, their force lies precisely in that. Don’t think that you are living your life and whatever happens, happens, what can be. No, things are actually happening now in a very clear way. Therefore what I am doing now—it is not true that I am just living my life and that’s it. It’s not; there are consequences to these things.

Let me just get to the enigma.

What? Yes, exactly, I want to get there, so look—there is another question here: what happens regarding a vaccination situation, an example of a vaccine, where as a result some people will certainly die, a certain percentage, but on average lives will be saved. On average this vaccine saves many lives, but practically this vaccine itself will cause the deaths of several people. So that’s the question, right? Whether to do such a thing. I won’t go into that now because of time; I’m getting to the enigma.

Meaning, there is the question: what should an undercover operative do if he has been living a long time among Arabs and as part of his role uncovers things that prevent attacks, etc., like the spy Eli Cohen, and he sees a wounded Jew there in the Arab area where he is, and there is no one to treat him, and he is in mortal danger, and he, the undercover operative, is a Magen David Adom medic who can save him. Okay, it doesn’t matter—you’re undercover, you’re supposed to remain concealed, you’re at some Arab outpost, next to you sits a wounded Jew, there is no one to treat him. If you don’t treat him, he will die. But if you do treat him, then it will be discovered that you are undercover. Okay? No, not that you will die—that’s certain, so then you wouldn’t have to. But no, you won’t die; you will be exposed. Once you are exposed, then your functioning as an undercover operative, which in the end saves lives in the future, is impaired, and you won’t be able to continue being undercover. Okay, so the question is whether you are allowed to treat him.

So in this context he brings the famous story about the Enigma in Coventry, right? When the British cracked the Enigma—Turing cracked it with his group at Bletchley Park—so Bletchley. So they cracked the Enigma. Intelligence people often had dilemmas: they were forbidden to reveal that the Germans knew the Enigma had been cracked. Now information came to them through Enigma that Coventry was about to be bombed—a heavy bombing, meaning many people were going to die there. Okay? The dilemma was whether to warn the residents of Coventry, but then the Germans would understand that the information came from Enigma and would realize that Enigma had been cracked, and you would essentially lose one of the strongest cards they had in the war. Because Enigma gave them information, and as long as the enemy doesn’t know you have that information, that’s very significant. Or not to tell the people of Coventry, and then in Coventry many people would die, but overall the cracking of Enigma is considered one of the major factors in winning the war. Therefore in the end the gain is an immense gain for all the world, or the whole country, on a larger scale than that city. So the question is what to do. The British decided not to reveal it.

Coventry is in England?

Yes. The British. Churchill.

No, that’s different. That’s different. If it were not to reveal it to Bordeaux, obviously they would have revealed it.

All right, in any case, they decided not to reveal it, and the question is what to do in such a case. Here he really brings the consideration of the time gap. He says that both in the case of the undercover operative and in the case of Enigma with Coventry, right now you have life-saving right in your hands. Right now. You have a concern that this thing will lead to problematic consequences in the future. Okay? This is the reverse of a mandatory war. Right now you have a prohibition, but you hope that in the future there will be positive consequences; here it’s the opposite. Right now you are doing something that is a commandment—saving life—but in the future you may have to pay a price for it. The very clear assessment was that yes—and in the end this proved true—that there would be very great future benefit, or there could be great future damage if Enigma were exposed. And then the question is—so he brings in the name of his father-in-law, that is, Rabbi Elyashiv—Rabbi Elyashiv claimed that if right now you need to save a life, save the life. You are not supposed to be dealing with the future. Meaning, the same with Coventry and the same with the undercover operative; in the end there is a present obligation upon you, and future considerations, leave them aside.

But what is interesting is that he inserts into the reasoning two things: first, that it is future; and second, that who knows—there are all kinds of ways in which in the end maybe in the future what you think will happen will not happen, and to jeopardize the present, the value in the present, for some uncertain future—that is not justified. There are two different reasons here even though he brings them together. Because one reason says: leave me alone about what will happen in the future. It will certainly happen in the future, but it’s in the future; right now I need to save life, so I save life. That’s one reason. But within that he also inserts the statistical, probabilistic reason. Meaning, he says: right now you are certainly saving; in the future who knows whether there will be danger or not, and therefore you do not… It’s like we once talked about slippery-slope considerations. I am opposed in principle to slippery-slope considerations, except in extreme cases. Slippery-slope considerations basically mortgage a certain present for the sake of an uncertain future. Meaning, you do something now that is certainly wrong because in the future you want to prevent the possibility that they may do something wrong. So for that you are already doing something now that is certainly wrong. So leave it—do the right thing now, and in the future we’ll see.

So why don’t you make that consideration with prohibitions, for example dissecting the dead?

I didn’t understand. The opposite consideration—you don’t make it? Why? What, I didn’t understand?

So then you would save the residents of the city?

That’s what Rabbi Elyashiv says, yes.

But the reverse—you don’t do that? You don’t say—?

Because it’s a prohibition! Because it’s a prohibition. On the contrary, he says: save the dead person now from dissection, or save yourself now from the transgression, and whatever happens in the future will happen. It’s exactly the same thing.

But that’s exactly what Rabbi Elyashiv… It’s exactly the same thing, yes.

Now just one point I want to add in light of what we discussed in the last two classes. What very much characterizes both Rabbi Elyashiv and Rabbi Zilberstein, his son-in-law, is a classic Haredi mode of thinking. This diagnosis is very interesting, because in the proofs they bring and the mode of thinking, it is all a way of thinking of private individuals. Meaning, their considerations do not include the question of how a public makes decisions. And I think in the context of Coventry this is very striking. In the context of the undercover operative it is a slightly trickier question, because there it isn’t an open war and so on; it is more of a local consideration. But in Coventry it is really a public consideration, and there I am not at all sure I agree. Meaning, on the level of the personal individual, I also tend to agree with what Rabbi Elyashiv said. Say, you should treat the person in front of you now, and if afterward there are problems, there will be problems—what can I do? You need to save the one standing before you now. But in the calculations a government makes, or the Allies at the time, it doesn’t matter—calculations made on the global level are often different.

I brought the example of a glowing piece of metal in the public domain. According to the Geonim, it is permissible to extinguish it on the Sabbath even by means of a Torah prohibition. Why? Because perhaps someone will pass by and be harmed. But that is only in the public domain; in private property, no. Maybe several people will pass by—is it because it’s several and not one? No, because it is in the public domain. It could be that only one will pass by, but it’s in the public domain. So I said there too there is both a statistical aspect and a non-statistical aspect; there are two planes of consideration. But for our purposes what matters is that a public consideration often does not proceed along lines parallel to the private consideration.

It somewhat reminds me of what Rabbi Yisraeli said regarding the laws of war, with which I did not agree in previous classes, but I do… I don’t agree that we need this, but I do agree that it is true. Meaning, in the end governments do not always make decisions like a private individual, only on a larger scale. No, sometimes a government has a consideration—and what is the idea behind it? The idea behind it is what I talked about in the previous class. Why is a government not just many people? We talked earlier about the difference between a public and many people. Because for me, since the government represents all the citizens, the government is also me. And when the government made a decision, then the people of Coventry made the decision through the government they elected. So the people of Coventry are allowed to decide to sacrifice themselves in order to save the rest. Now, it’s not that they actually made this decision in practice, but the conception of a state as a governing authority that makes decisions for the citizens who elected it—think of the Athenian demos, where the whole state gathers and they vote. So if the state had gathered and voted, including the people of Coventry, then that would indeed have been possible, right? The people of Coventry handing themselves over. And they would deserve the fullest appreciation for sacrificing themselves in order to save all of England, or even the whole world.

Now because of considerations of secrecy, it was impossible to ask the people of Coventry themselves, so what we can do instead is that the authorized representatives make the decision for them, for all of us. There is indeed a difference: when a public, through a government or an authorized representative of the public, makes public decisions, that is in a certain sense like a person who volunteered to do it. And therefore, even if I am forbidden to do this to some person against his will, the person himself is certainly allowed to volunteer to do it. If he volunteers in order to save the public, that is Pappus and Lulianus, whom we talked about—the martyrs of Lod. Right? Now when the government decides, that is basically like the person volunteered, because the government speaks in the name of the public as a whole.

And Rabbi Elyashiv and Rabbi Zilberstein do not make that consideration, and not for nothing. In Haredi thought, that plane simply isn’t there. They do not perceive the public as something halakhically different. It’s not only the attitude to Zionism; it’s not only that the attitude to Zionism is a result of this, or that this is a result of the attitude to Zionism. But there is some unwillingness to see—because for two thousand years we were used to not having a state, not having a functioning public. We were a collection of individuals. Therefore all the halakhic patterns developed on the conceptual infrastructure of private individuals trying to maneuver through life and decide what to do. In a place where someone makes decisions for you—the public as a whole, its representatives make decisions for you—that is not something that exists very much in the halakhic tradition. Therefore Haredi halakhic decisors do not really make these calculations, which in my view are unavoidable in the public context.

And with that we conclude.