Monday, September 23, 2019

Fulfilling requests

One of the most moving stories in Rosenbaum’s deeply moving Holocaust and the Halakhah tells of how one can be a great moral hero even when acting out of mistaken conscience. A man in a concentration camp comes to his rabbi with a problem. His son has been scheduled to be executed. But it is possible to bribe the kapo to get him off the death list. However, the kapo have a quota to fill, and if they let off his son, they will kill another child. Is it permissible to bribe the kapo knowing that this will result in the death of another child? The rabbi answers that, of course, it is permissible. The man goes away, but he is not convinced. He does not bribe the kapo. Instead, he concludes that God has called him to the great sacrifice of not shifting his son’s death onto another. The father finds a joy in the sacrifice amidst his mourning.

The rabbi was certainly right. The father’s conscience presumably was mistaken (unless God specifically spoke to him and required the sacrifice). Yet the father is a moral hero in acting from this mistaken conscience. (Here are two relevant features of this case. First, while he was mistaken, he was not mistaken in a way that shows moral callousness—on the contrary, he is obviously a man of moral sensitivity. Second, while he was mistaken in thinking the sacrifice was morally required, nonetheless the sacrifice was—I think—at least permissible.)

The analytic philosopher will see this as a variant of a trolley case (with some complications, such as that the deaths were mediated by the free agency of the kapo). It is permissible to redirect the trolley away from one’s child and towards a stranger’s child. This is another way in which the proportionality condition in the Principle of Double Effect is not a utilitarian calculation: the agent has a proportional reason to save their own child even when it is foreseen (but not intended) to cost another’s their life.

But at the same time it would not be permissible to redirect the trolley away from one stranger’s child towards another stranger’s child. Such redirection would be a grotesque toying with lives. It would be a needless and callous making of oneself into a cause of another’s death, even if unintentionally.

Here, however, is a case that puzzles me. Suppose Alice’s child is on the track the trolley is speeding towards, and a stranger’s child is on another track. Alice is physically incapable of redirecting the trolley but Bob is capable of it. Alice and both children are strangers to Bob. Would it be permissible for Alice to ask Bob to redirect the trolley?

Here is an argument to the contrary. It is impermissible for Bob to redirect the trolley from one stranger to another: that is just playing with lives. But it is impermissible to request someone to perform an impermissible action. Hence, it is impermissible to ask Bob to redirect the trolley.

That seems mistaken. The case of asking Bob to redirect the trolley need not be that different from begging the kapo to take one’s child off the death list, depending on the details of the latter story. So what is going on?

I think there are at least two ways to justify Bob’s acquiescence to the request and hence Alice’s making of the request:

  1. Once Alice asks Bob to redirect the trolley, Alice is no longer a stranger to Bob. There is a way in which Bob in receiving her request can become an agent of Alice’s, and hence those that Alice cares for become ones that he has a special reason to care for.

  2. On receipt of the request, Bob has two options coming with distinctive incommensurable reasons. The first is not to redirected, with the reason being promote equality, in this case equality between children who don’t have a parent in place to speak up for them and ones who do. The second is to fulfill the request of an anguished parent to save their child. Both reasons are grave, and it is permissible for him (other things being equal) to act on either reason. Requests really do add weight to reasons.

There is another complicating factor. I do have the intuition that if Bob is an employee in charge of the trolley, he should do nothing. The reason is this. Insofar as he is in charge of the trolley, Bob has a role duty of mitigating damage done by the trolley. It is generally good policy that such a role come along with a significant independence from outside influences, such as bribes or even requests. So, in that case, Bob should act as if he did not receive the request. But if he did not receive any request, he shouldn’t do anything, for it is better not to become the cause of the child’s death—as one would if one redirected.

Here is a variant case. There are three tracks. The trolley is on track A with five people. The other two tracks, B and C, have one person each, and Alice is asking Bob not to redirect to track B, as her child is there. Bob has to redirect to either track B or C, but everything other than Alice’s request is equal between these tracks. Here it seems to me that Bob should flip a coin (if there is time; if not, just act as randomly as he can) if he is an employee. And if he is not an employee, then he has a choice to accede to Alice’s request or flip a coin.


Philip Rand said...

Feser uses Dostoevsky... you use Roesenbaum... both you chaps are nuts...

Russell Jones said...

While the father may have been mistaken, I believe the Rabbi was most certainly incorrect in his judgment. Bribing the kapo would be enriching the evil doer. If we take the whole scenario together, version 1 involves a father grieving over the death of his son, and version 2 involves a father guilt ridden for both enriching a murderer and sending another child to death.

I know at first glance this seems to be dodging the thought experiment, but I think that most moral thought experiments are far too loose in their description. Take the example of Alice asking Bob to change the direction. The only thing Bob is certain of is that Alice and her daughter will be happier if he switches the tracks. He knows by not switching the tracks, Alice will be unhappy and one child will be happy. The comparison here seems trivial.

To really tighten up these examples, we need to remove a great deal of uncertainty.

Helen Watt said...

Maybe it's not so much that you're enriching the evildoer (is it always wrong to pay kidnappers, for example?) Maybe it's more that the evildoer's plan to replace your child with another child on the list is a practically necessary condition to your getting what you want:
in practice, the evildoer will only plan to strike out your child's name if he simultaneously plans to replace it. Even if that's not a means that you yourself are intending to your getting what you want, it's a 'causal' means of sorts - so falls foul of at least one version of the 'means' condition in double effect reasoning.

Alexander R Pruss said...


It is permissible to enrich evildoers. If a gun is pointed at you and it is said "Your money or your life", typically you should hand over your money.


Very interesting. I had been thinking about the other child dying as clearly not a means (after all, it happens after the danger has passed your child over, and a means can't come after the end (without time travel or the like)). But it is also wrong to intend the evildoer to have an evil plan. And that does complicate matters. But look: this "practical necessity" isn't all that necessary. Suppose that the kapo takes one's child off the death list and then just happens to forget that he is going to be in trouble if he doesn't put in a replacement. Your plan to rescue your child is still successful. But if forming the plan to put a replacement on the list was a means, then your plan would not count as successful if that didn't happen. So I think it's not a means.

Helen Watt said...

But isn't the kapo's plan (if not the actuality) of putting a different child on the death list the practical means here? The parent's plan is successful when (a) the kapo forms a plan to strike off the child's name and (c) he actually does so. But (a) and (c) will not happen without (b) the kapo forms a plan to put another child's name on the list instead. So isn't the kapo's forming that plan a practical means (if not necessarily a means intended by the parent) to (b) which is after all the route to (c)?

Alexander R Pruss said...



Consider three machines with subtly different programming. In each case the machine administers a death list and you can select a name on the list and pay money to the machine.

1. The machine first removes the selected name and writes in another random name. Nothing else of relevance is done.

2. The machine first writes in another random name, checks that it was successful in doing so, and if it was successful in doing so, it removes the selected name.

3. The machine sets in motion a program with a timer that will shortly write in a random name, checks that the program has been set in motion, and before that timer kicks in, it removes the selected name.

In case 1, it's clearly permissible to pay the machine to remove the name of one's child. It's just like the trolley case.

In case 2, it sure looks like the writing of the random name is something like a causal means to the erasing of the selected name. In case 3, it looks like the setting up of the delayed write-random-name code is a causal means to the erasing of the selected name.

Still, in the machine cases, I think it's still permissible to pay the machine in cases 2 and 3, because a machine's writing of a name on a list isn't in and of itself bad, even if that list is going to be used for figuring out who is going to be killed.

The kapo seems to be more like machine 3: the kapo will only erase the name because he is still planning to write in a new name. So his planning to write in a new name does seem to be a part of the father's plan. AND there seems to be a crucial difference between the kapo and the machine. The machine's having code to write in a random name is only instrumentally evil, while the kapo's having a plan to write in a new name seems to be intrinsically evil given the kapo's knowledge of what the name will be used for.

But maybe there is some wiggle room here. Imagine this scenario. You've hacked into the unjust executioner's computer. There is a list of ten names of people to be executed. The executioner has a poor memory for numbers. If there are eight or nine people on the list, he won't notice any difference. But if there are seven or fewer, he'll notice and kill a dozen extra people because he'll be mad. So ideally you'd erase two names from the list. But unfortunately because of weird technical details of how your hack works, the only erasing you can do is to erase the first three names on the list, nothing less and nothing more. But if you erase the three names, you will have to insert a name, or else the executioner will notice there are only seven. So, you decide to erase the three names, and then put in a name. And it has to be a real prisoner's name, or the executioner will notice.

Suppose the name you put in is your own. Surely then you are to be praised. You've done nothing wrong. In particular, you haven't intended your own death or done anything else that's evil. But perhaps you can't put in your own name: you aren't one of the prisoners and so the executioner will catch on to the hack if your name is on the list. A fair thing seems to be put back one of the three names you erased, at random. But suppose you can't do that. Maybe all three are close associates, and if the executioner sees one of these three names on the list, he'll notice that the other two aren't there and get mad. Then, is it really wrong to put in a random prisoner's name? You're not intending that the person with that random name be killed, but only foreseeing it. You're just redirecting the trolley.

But if this is OK, then it is not guaranteed that the kapo is acting wrongly in replacing the name on the list. Just like the hacker need not be intending that the person added to the list be killed, the kapo need not be intending death for the people on the list.

Helen Watt said...


Very interesting! The machines may be a bit different from trolleys though which aren't normally aimed at the person on the track. (Of course, machines and trolleys can't really aim at things but we can aim at things for them.)

If the name were just a code that did nothing to harm the random person, then obviously there wouldn't be a problem. But if it directly activates say a DNA-seeking missile with the random person's DNA substituted/awaiting substitution as a target so that your child can be saved, is that maybe like getting a savage dog to chase another child?

In the case of Kapo, would we be happy with getting him to aim a trolley at some random person (albeit not particularly intending that it hit them) to distract him from our escape attempt?

With the covert name substitution, it's one thing not to raise Kapo's suspicions - he sends the email list and just doesn't notice anything has changed - but something else to get him to have a positive thought: that's good, I have the right number of people on my list to hand over (as it happens, to certain destruction).

It sounds a bit better when you're replacing the names of people who were already on the list but to choose a random name looks particularly like instrumentalising that person - pulling them into the situation to benefit third parties. Isn't that a bit like using people as human shields? (And might you be instrumentalising even yourself if you put your own name on the list?)

Russell Jones said...

Dr. Pruss,

I don't think it is permissible to enrich evildoers without a sufficient moral good coming from it. In the scenario discussed, the parent would enrich an evildoer without producing any moral good (unless you consider the pleasure of the parent without considering the pleasure of the hypothetical parent/siblings/friends of the replacement child).