The Coventry Dilemma (Column 764)

With God’s help

About two weeks ago, a military incident from Lebanon took place that stirred our public discourse (see here for an overview). It reminded me of another ethical issue that stirred the literature and public discourse surrounding World War II, regarding the bombing of Coventry. Some of what will be discussed here already came up in column 608, but a number of substantive points have become clearer to me, especially the connection to the Coventry dilemma.

Description of the event

On 11 March there was a very massive Hezbollah barrage on the northern communities. This was an unusual and surprising attack that forced everyone into protected spaces and prevented them from functioning (see, for example, here). Already while it was happening, a public argument began, with accusations exchanged against the IDF (see, for example, here, already the next day, and the argument continued for several more days after that). The claim was that there were elements within the IDF that prevented preemptive action and also did not notify the residents, and according to several journalists this was done so as not to expose an intelligence source. There was a dispute among various IDF officials, mainly the head of Military Intelligence, the commander of the Northern Command, the IDF Spokesperson, and others, each casting the blame on the other. As far as I saw, no final conclusion on this matter was published.

The issue itself: the Coventry dilemma

The very attempt to evade responsibility, beyond the question of what factually happened here, indicates that all the relevant parties share the assumption that such conduct is improper. That is why each of them tried to dissociate himself from the matter. For the purpose of our discussion, let us assume that there really was such a need on the factual level; that is, there truly was a dilemma between exposing an intelligence source (human or otherwise, HUMINT or SIGINT) and the residents’ security. Is it really so clear that such conduct is improper? Protecting intelligence sources is of tremendous value, and of course that too concerns human life. So why are the lives of the people who would be harmed now necessarily preferable to the lives of those who would be harmed in the future because of the absence of intelligence?

This immediately brought to mind what is sometimes called the ‘Coventry dilemma’ (see here). The Coventry affair is frequently mentioned in intelligence literature. Some claimed that Britain’s prime minister during World War II, Winston Churchill, refrained from evacuating the city of Coventry before it was bombed by German Luftwaffe aircraft, thereby causing harm to hundreds of its civilians. The claim is that he did so only because the information about the expected bombing became known from the content of a decoded German cipher telegram, and Churchill feared that evacuation measures would become known to the Germans and cause this highly valuable source to be shut down. The myth of sacrificing Coventry on the altar of source security serves both as a description of the importance of safeguarding intelligence sources and as an expression of the dedication and resolve the British showed in their war against the Nazis (this took place in 1940, before the other Allies had joined the war, when Britain faced a very real danger of defeat and even Nazi occupation).

The background to this was apparently the deciphering of the German Enigma codes, one after another, which was a trump card in the war as a whole. This was done mainly by Polish teams already before the war and during it as well, and was later brought to completion at Bletchley Park, Hut 8, by a British team that included the famous Alan Turing. It is no wonder that it was extremely important for the British to keep secret the fact that the code had been broken, even though this did not endanger any particular agent. It endangered the entire war effort. If the Germans changed the code, Britain would be unable to defend itself and many would be harmed.

From the mid-1970s onward, the secret of the Enigma decryption was revealed, and then various memoirs by British intelligence personnel began to be published, some of them apparently not entirely accurate. Among other things, it was published that during the air blitz on Britain in November 1940, a German cipher operator made a mistake, and instead of using a code word to describe the target in the encrypted telegram, he stated its explicit name—Coventry—as the city that was about to be heavily bombed from the air. The story continues that Churchill received this precise warning two or three days before the bombing, and because of the sensitivity of the source he was required personally to decide whether to reinforce the air-defense array around the city, whether to rush fighter planes into the airspace above it, and especially whether to begin evacuating the population. The story says that he called these decoded telegrams "the goose that lays the golden eggs for us," and therefore sought to guard those eggs at all costs in anticipation of the severe tests that still awaited him in the future. He determined that apart from increasing the alertness of the rescue services, no preventive action would be taken in advance of the bombing. He also ordered his staff and aides to ‘bury’ the information and not report it to anyone. That night, 14 November 1940, hundreds of German bombers dropped 450 tons of bombs and more than a hundred mines on the city, causing the deaths of about 550 people, the injury of 860 more, and the destruction of some fifty thousand (!!) homes.

I should note that over the years it became clear that, in truth, neither the codebreakers at Bletchley Park, nor the intelligence officers of the Royal Air Force, nor even the British political leaders knew sufficiently far in advance what the target of the bombing would be. The truth was grayer. It later emerged that about three days before the attack, on 11 November, a decoded Enigma message revealed that the German air force was planning an important operation and that the planes would be guided to the target by means of radio-beam navigation (this was a method the Germans developed to guide aircraft at night or in poor visibility), in which directional radio beams are sent from two different places toward the target, each with its own distinctive tone, and the planes flew along them. When the beams converged, the pilots understood that they had reached their destination and dropped the bombs. They did not know that the target was Coventry until the final minutes, when they saw the beams converging in real time, and by then there was no time left to take any meaningful action.

Churchill in fact made no decision to withhold adequate defense from Coventry, and he did not even know that it was the intended target, but the question of principle obviously remains. Is it proper to act in this way? Do we sacrifice the lives and property of hundreds and thousands of residents for long-term strategic goals? This is certainly similar to what was described in our case regarding Hezbollah.

Differences between the cases: preliminary considerations

There is, of course, a major difference between the cases. One side of the comparison favors Churchill. Israel is not in an inferior position vis-à-vis Hezbollah; on the contrary. We are the stronger side here. Therefore, the consideration of sacrificing residents for the sake of victory in the war seems more problematic in our case. On the other hand, there is a simple consideration in Bibi’s favor. In our case, most residents have reasonable protective solutions, and whoever follows the instructions is under reasonable protection. Hezbollah’s rockets are nowhere near the intensity of the massive aerial bombardments that took place in Coventry. Therefore there is more room to withhold the information and forgo preventive action, since this is not really a matter of sacrificing human lives. True, one is sacrificing their property and quality of life (and even that can in principle today be handled with reasonable compensation).

Beyond the differences between situations and cases, what remains for us is the principled question: Is it proper to sacrifice present lives for the sake of future gains?

The Coventry Dilemma: Discussion

At first glance, on the level of principle this seems like a simple calculation. I can sacrifice the lives of N people now in order to save the lives of M people in the future. Assuming that M>N, it seems self-evident that this is justified. There is, of course, a technical problem of estimating what N is and what M is (especially when we are dealing with an estimate of future risk), but on the principled plane, given M>N, there does not seem to be any special problem here.

On its face, this dilemma resembles the trolley problem and also the dilemmas regarding the hostage deals in the Swords of Iron War. In both of those cases, the basic schema is similar: sacrificing N in order to save M. And in the case of the hostages, these were precisely the claims of the opponents of the deal: after all, it is clear (to them) that M>N, so what is the question? In what way are the lives of the hostages preferable to the lives of future civilians who will be murdered?

But the matter is not so simple. Let me recall that in the trolley problem, a train is moving along the track, and farther ahead five workers (5=M) are sleeping on it, and it is about to run them over. I am standing by a switch and can divert the train to another track on which one worker (1=N) is sleeping, and he will die. The question is whether to divert it or not. Apparently it is obvious that it is better for one to die than five, and therefore one is sacrificed in order to save five. This is a consideration that makes everything depend on the outcome (how many people lost their lives).

But there is an approach, with which I personally tend to agree, according to which in such a situation it is not right to sacrifice the life of one person in order to save the lives of five others. There are two main considerations here:

• Diverting the train is an action that I perform, and that action kills one person. If I do not divert it and remain passive, five people will die, but not at my hand. I am not a murderer. That is, we do not look only at the result (how many people lost their lives) but also at the actor and the action (whether I am a murderer). In my eyes this is the less fundamental difference. In my article on separating conjoined twins, I argued that the fundamental consideration is the consequentialist one, not the actor-centered consideration (whether I am a murderer). Therefore, there, although all the halakhic decisors prohibit performing the surgery and instruct that both be left to die, I hold that one is obligated to perform the surgery, even though I thereby kill one of them (or at least deprive him of whatever brief span of life remains) in order to save the other.

Therefore, the more significant consideration in my view is the following:

• Diverting the train causes the death of a person who is not involved, and there is no reason he should sacrifice his life in order to save other people, so there is no justification for forcing him to do so. This is sometimes expressed with a phrase like: "I am not supposed to play God" and decide who will live and who will die.

These two considerations are not identical, but of course they are also not entirely disconnected. The fact that, by diverting the train, I am performing an act is what underlies the perception that I am ‘playing God.’ Remaining passive and not diverting it is not perceived by us as ‘playing God’ in the sense that I am sending the five to their deaths. Ostensibly, this is the same consideration as the first one, which hinges on activity versus passivity. But that is not the full picture. They are still two different considerations. The first focuses on my prohibition of murder (that I did the killing), that is, on the actor; the second focuses on the result (that a person was murdered). The second consideration is actually telling us that even if we focus on the result and not on the person who brings it about, there is still room to prefer inaction, because the five simply happened to find themselves in mortal danger, and the one person who is unrelated to the situation is not supposed to pay with his life in order to save them. Could we seriously imagine that that one person would be obligated to give up his life in order to save the five? Who is to say that their blood is redder than his?! Why should he, who is unrelated to the situation, have to step in and sacrifice himself in order to save them? If so, by the same token, someone else cannot send him to his death. And again, this is a consideration that can be made even if we do not focus on the actor, that is, on the question whether I am a murderer, but rather on considerations of outcome. In the initial discussion we thought that, if the consideration is consequentialist, the conclusion is self-evident. Now we understand that this is not necessarily so.

An example from the prohibition of interest

At the beginning of the chapter Eizehu Neshekh, the Mishnah discusses the prohibition of neshekh and the prohibition of tarbit (because in the Torah itself they appear as two prohibitions). The Talmud there tries to understand the difference between them:

It follows that by Torah law neshekh and tarbit are one and the same thing. But are not the verses written: “interest on money” and “increase on food”? But can you really think that there can be neshekh without tarbit, or tarbit without neshekh? What would neshekh without tarbit be? If he lent him one hundred in return for one hundred and twenty, and at the outset one hundred was worth a dinar while in the end one hundred and twenty was worth a dinar—there is neshekh, since he is deducting from him, taking from him something he did not give, but there is no tarbit, for he has no profit, since he lent him a dinar and is taking a dinar from him. In the end, if you go by the initial value—here there is both neshekh and tarbit; if you go by the final value—there is neither neshekh nor tarbit. Furthermore, what would tarbit without neshekh be? If he lent one hundred for one hundred, and at the outset one hundred was worth a dinar while in the end one hundred was worth a fifth more—if you go by the initial value, there is neither neshekh nor tarbit; if you go by the final value, here there is both neshekh and tarbit. Rather, Rava said: You cannot find neshekh without tarbit, nor tarbit without neshekh, and Scripture distinguished them only so that one violates two prohibitions.

That is, there is in fact no difference between the two, and all that the Torah separated them for was so that one would violate two prohibitions. But if so, why does the Torah use two expressions? Moreover, Maimonides, in the ninth principle, rules that one should not count two prohibitions or two commandments with overlapping content (unless it is a prohibition and a positive commandment; see the sixth principle). That is why the obligation to observe the Sabbath is counted as one commandment, even though the Torah repeats it 12 times. So why, in the case of neshekh and tarbit, where the content overlaps, does one violate two prohibitions?

In Maimonides’ Sefer HaMitzvot you indeed will not find two separate prohibitions, but at the beginning of Hilkhot Malveh Ve-Loveh (and in the discussion itself as well) he rules that one does in fact violate two prohibitions (Hilkhot Malveh Ve-Loveh 4:1):

Neshekh and marbit are one and the same, as it is said, “You shall not give him your money with interest, nor give him your food for increase,” and elsewhere it says, “interest on money, interest on food, interest on anything lent at interest.” And why is it called neshekh? Because it bites, for it causes pain to his fellow and consumes his flesh. And why did Scripture distinguish them? So that one would violate two prohibitions.

What is the difference between these two? Maimonides hints at it here, but Rashi on the Mishnah explains it explicitly:

What is neshekh? — That it bites, for he takes from him what he did not give him.

“One who increases through produce” — one who increases his own gain through produce; and in the Gemara it explains that with a loan of money or produce as well, this is tarbit, for his wealth increases; and here it explains rabbinic interest, by way of sale and purchase.

And likewise on the Talmud:

Can there be neshekh? — where he bites his fellow, yet his own wealth does not increase.

And can there be tarbit? — where the lender profits, yet the borrower is not bitten.

That is, neshekh is the prohibition against biting into the borrower’s property, while tarbit is the prohibition against unlawfully increasing the lender’s property. As noted, there is no simple practical consequence that distinguishes them (and that is why Maimonides does not count them as two separate prohibitions), yet these are still two different prohibitions. This is not like the Torah writing the same prohibition twice, and therefore one who lends on interest violates two prohibitions. These are two different ways of looking at the same act: from the standpoint of the injurer and from the standpoint of the injured party. One who lends on interest has thereby violated two prohibitions: he has both ‘bitten’ the borrower and increased his own wealth.

The same applies to our case. The two considerations I presented above are parallel: one speaks from the side of the person who killed, which is one aspect of the prohibition of murder, namely that the perpetrator is a murderer. The second aspect is that some person was murdered, and that is the consequentialist consideration.

Who is the one saved and who is the one sacrificed?

Our assumption is that there are M future victims and N hostages, with M>N. As we have seen, this still depends on the question who is being sacrificed and who is being saved. If the N are the ones being sacrificed, then there is no justification for sacrificing them in order to save M others, even if a larger number of people will be saved (as in the trolley problem). But if the N are not being sacrificed, and we are simply not saving them in order to save the M from danger, that is a different situation, and there there is room to make a numerical calculation.

In the trolley case it seems fairly clear that the saved are the five, and the one who is sacrificed in order to save them is the single worker. The reason is that the five are under threat and the one is not. The question there is whether it is permitted to sacrifice one in order to save five. I explained why I tend toward the approach that forbids this, for both reasons, though the second is, in my eyes, the more central one. What about conjoined twins? There it is clear that both are under threat, and the one who is sacrificed is no different from the one who is saved. Moreover, the one sacrificed would die even without this. Therefore there it is clear that there is justification for sacrificing one in order to save the other. I explained in my article that the lottery that determines which of them will be saved and which will be ‘sacrificed’ is a rescue for both of them, since in that way they at least receive a 1/2 chance of being saved (otherwise both die). That is not so in the trolley dilemma, where it is clear that the one is the one sacrificed and the five are the ones saved.

And what about the hostage deal? There it is not at all clear who is the one sacrificed and who is the one saved. Are we sacrificing the hostages in order to save the future potential victims, or are we sacrificing the future victims in order to save the hostages? In fact, there is a present threat to the hostages, and releasing the terrorists will create a new threat to the future victims. Therefore, at first glance, the ones sacrificed there are the future victims and the ones saved are the hostages. If these are the definitions, then at least according to my approach there is no justification for sacrificing the future victims for the sake of saving the hostages, even if the number of future victims is smaller.

How should this be applied to the Coventry dilemma (or to the Galilee and Hezbollah)? Are the residents of the Galilee the ones sacrificed, and the residents of Israel who will be harmed in the future if the intelligence source is exposed the ones saved? Or is the opposite true? Apparently the answer depends on the question who is the one saved and who is the one sacrificed. Before turning to clarify that question, one qualifying remark.

Qualification: an uncertain threat

Up to this point I have assumed that, with regard to the hostage deal, quite apart from the question whether M>N, everything depends on who is defined as the one sacrificed and who as the one saved. But I have already written in the past (see column 608) that there is room to reject that argument because of the following consideration. The future threat in this case is not certain but doubtful. Even if we release terrorists in a deal, there is no concrete threat here to any specific person; rather, each of us faces only the possibility that he may be harmed. Let us assume for the sake of the discussion that there is no doubt that there will be attacks and that people will be harmed (although even that is not correct. There is always doubt, and certainly with respect to the number of casualties). Even so, even if we assume that the expected harm is M>N, that is, that the number of expected casualties is greater than the number of hostages we saved, it may still be justified to release the terrorists. One can argue that each of us is willing to take upon himself a small risk of being harmed in order to free dozens of hostages who are in present and certain danger. Thus, for example, I am willing to accept a 0.001% (one-thousandth of one percent) risk of being harmed (each of us takes such risks all the time) in order certainly to save dozens of hostages who are in certain and present danger. Notice that this is so even though, when we calculate the total number of casualties, we get M=0.0001X10,000,000= 100, which is more than the number of hostages eventually released. That is, when we are dealing with a doubtful future threat, the number of casualties hardly plays any role, and it also does not depend on who is defined as the one saved and who as the one sacrificed.

How do we decide who is the one saved and who is the one sacrificed?

As stated, in the trolley case it seems clear that the ones saved are the five and the one sacrificed is the single worker. In the case of conjoined twins there is no such distinction at all. In the case of the hostage deal there is a dilemma as to whether the ones saved are the hostages and the future victims are the ones sacrificed, or the opposite. At first glance, it seems that the ones saved are the hostages, who are now under threat, and the ones sacrificed are the future victims, who are not presently threatened by anything. Exactly as in the trolley dilemma. The same applies to the Coventry dilemma. There too it seems that the residents of Coventry are under a present threat, and it is not right to sacrifice the future victims (who are not currently threatened) in order to save them. Up to this point we have assumed that the determination of who is sacrificed and who is saved is made according to the question of activity versus passivity. Whoever would not have been harmed had I done nothing is the one sacrificed. Whoever is under threat even without me is the one saved. That is also what we saw in the analysis of the trolley dilemma.

But here another variable enters. A state has a duty to protect its citizens. A state is not a private, uninvolved third party that acts solely according to moral rules and nothing more. A state has a duty to protect its citizens, and that duty applies to every citizen in every situation. This becomes especially forceful when we remember that the hostages entered mortal danger because of the state’s failure to fulfill that duty (the failure of 7 October). It is true that the future victims also are not to blame, and the state must protect them as well, but there is room for the claim that when what is at issue is the relationship of a state to its citizens, the distinction between those sacrificed and those saved loses its force.

As an analogy, let us consider the law of a custodian who was negligent. A custodian undertook to guard the animal entrusted to him. He left the door open and the animal ran away. In the case of an ordinary person (for example, someone who comes and opens the door of my cowshed and my animal escapes), this would be damage by indirect causation, and he would be exempt. So why is the custodian liable? Because he has a contractual and halakhic duty to guard the animal. When you have such a duty, even your omission is regarded like an active deed. So too, according to Jewish law, in the case of an expert, such as a money-changer whose opinion led to a loss, he must pay even though it is only indirect causation. If I came to him as an expert to ask for an opinion, he has a professional responsibility to give me a correct opinion (see ‘showing a dinar to a money-changer,’ Bava Kamma 99b, and Shulchan Arukh, Choshen Mishpat 306:6; see also here).

Incidentally, in many legal systems as well (including the Israeli one), if a person bears responsibility for something by virtue of his office, his negligence by omission can be regarded as active harm (see, for example, here). This applies to a corporate officeholder and also to a public official (this is ‘officer liability’). For a broader discussion, see Roni Rosenberg’s book, Between Act and Omission in Criminal Law.

The meaning of this is that if our discussion concerns the relationship of a state to its citizens, then the state has a duty to look after their security, and therefore its negligence that brought them into danger counts as an active deed. If so, then the state has in effect already sacrificed the hostages and now does not want to save them because of fear of future harm to citizens and soldiers. At that point it is no longer at all trivial to see the hostages as the ones saved and the future victims as the ones sacrificed. The situation here seems symmetrical, and perhaps there is room to view it as a case like that of conjoined twins. True, there is still room for the claim that passive inaction is preferable, but there is certainly also room to make the numerical calculation: is M>N or not?

Back to Coventry

You can now understand that the case of Coventry and that of Hezbollah are different from the hostage deal. True, in these cases as well we are dealing with the state’s duty to protect its citizens, but the threat that currently exists to the residents of Coventry (or the Galilee) is not the result of negligence or a blunder on the part of the state. Here there is a present threat to residents, and in order to save them the state endangers future victims. In such a situation, the case for preserving the confidentiality of intelligence sources (in order to prevent future harm) becomes much stronger, since the position of the future victims is entirely equivalent to that of those presently under threat. Here there is more room for the numerical consideration: is M>N or not?

Still, one can say that the future threat is uncertain, and therefore, if there is now a real and immediate threat to Coventry and the Galilee, present protection should be preferred even at the price of future risk. An ordinary resident of Britain or Israel is willing to take upon himself a small risk of being harmed in order to save people from a certain and present danger. Here, of course, the considerations I presented above regarding the differences between the cases must also be taken into account. In Israel there is no fear that Hezbollah will win, and the whole question is only how many will be harmed and who will be harmed. In Britain there was fear of losing the war and of German occupation. Here the future risk was enormous, and there is more room to take future risk into account and preserve intelligence sources. In their case, N was the population of a city, which is no small number, but M was enormous (the entire country). It is very doubtful whether an ordinary citizen in Britain would have been willing to accept a substantial risk that the whole country be conquered in order to protect the residents of Coventry.