1

Let us compare ST with Standard Murder (SM). The murderer in SM is typically an individual rather than a group. He also intends death, but not necessarily to terrorize others. Standardly, he has something against the particular person he would kill, rather than selecting him at random. The killing could be impulsive rather than thought through. It is typically done for personal (or perhaps, if done by the Mafia, for business) reasons.

It might be thought that SM is by definition morally wrong. But possibly, even murder is only prima facie wrong.³

Is it possible that in the case of ST and SM the person harmed/terrorized is always wronged even if we wrong him in the course of doing an act that, all things considered, is not wrong? (Often a sign of someone being wronged is that he may permissibly resist what is being done to him even if this interferes with the goals of the agent that justify his acting.) Consider a variant of the possibly justified murder case described above: If I do not kill the elderly person, someone else will torture him in the last 30 seconds of his life. Suppose that he knows the details of his prospects but refuses to consent to being killed by me 30 seconds earlier. It seems that in this case, if I kill the man, I would murder someone in his own best interests. Do I also wrong him? Perhaps not. If not, then someone need not always be wronged in being murdered. By contrast, ST as described above is never in the prudential interests of those harmed/terrorized. On these grounds one might think that it, but not SM, always wrongs its victim, even if the act is ultimately right. But what if it were in the moral interests of at least those who are only terrorized to be jolted from passivity in the face of the crimes of their government? Must we always be wronging these people if we put their moral interests before some of their strong prudential interests?

1

Let us consider possible implications of some of the differences to which I have pointed between ST and SM. Do the differences bear on the issue of whether stopping terrorists is a “war” rather than merely a “policing of (prima facie) criminal activity”? I have suggested that ST is not SM. But this does not mean that ST could not be criminal activity of any sort. If ST is a form of criminal activity, then might it be better to think of policing it rather than of being at war with it?

We should examine the war model. If we are at war with those who use terrorism, then this will be a war against the use of (let us suppose) improper means per se. This is a deontologist's dream, in a sense. But is not a war against improper means just policing the activity of criminals rather than a war? No, because there is a sense of “war” in which it seems appropriate to speak of a war against crime. This is when we do not go piecemeal after criminals. Rather, we go all out and try to extirpate criminal activity. So it may be the metaphorical sense of “war” that is used by those who speak of a war on terrorism (or a war on poverty). Terrorists, of course, do not think of themselves as engaged in a metaphorical war; they have declared real war on certain countries. But if white-supremacists in the United States have also declared real war on the United States, this does not mean that fighting them is any more than policing criminal activity and metaphorical war at most.

Consider another possibility. Perhaps terrorists are war criminals, and this is why it is appropriate to speak of being at war with them? But that suggests that there is a real war ongoing about something besides the use of improper means of fighting, for example, about the forms of government that should exist in the Middle East. In that real war, there are those who use means that are ruled out by the rules of just war, and they are the war criminals. But going after war criminals is itself usually a policing activity rather than a war. We might, however, make it a metaphorical war by adopting the all-out, extirpating approach discussed above. Then there would be two sorts of wars going on simultaneously.

Employing the second model of war, in which we are at war over issues besides improper means, we could say that we are at war with people over a particular substantive issue (e.g., the nature of political arrangements in the Middle East). It happens that all of the warriors on the other side use (seemingly) improper means in pursuing the war. If they are really unjustified in using those means, then they are war criminals. This model, unlike the first war-against-crime model, does not allow us to ignore the political/religious aims of the opponents and the differences between us and them over a particular substantive issue as well as over the means used in fighting.

Neither sort of war I have described implies by itself that there is a national emergency in a country at war (nor does an attack on one's home ground always imply a national emergency). It is the presence of an emergency rather than merely real or metaphorical war that might permit useful infringement of civil liberties and constitutional protections in order to stop terrorism.

2

The second model of war—involving substantive disagreement—brings us to a possible implication of the difference made by the political/religious motivation of the terrorists. Suppose that we came to see merit in some of their views about the very important issues that divide us (e.g., political arrangements in the Middle East) and also recognized that we would never have changed our views in this crucial way but for their terrorist acts. Then we might think that it is not morally inappropriate to forgive their use of improper means rather than punish the terrorists as criminals. This seems to have happened with Menachem Begin of Israel and Yasser Arafat of Palestine, who were terrorists but became recognized leaders of their national groups. Forgiving or even excusing such terrorists may require them to forswear ST means but not necessarily apologize for past behavior. By contrast, if Begin had been found to be an ordinary murderer and user of terror (e.g., he killed his wife to terrorize women in his family), then he would probably have been put on trial and told to resign his premiership. This raises the possibility that even planners of the 9/11 disaster could theoretically come to be seen neither as mere criminals nor even as war criminals. But this could be so only if it turned out that they had been in the right while we had been seriously deluded (and resistant to change except by terrorism) about crucial issues that divide us.

1. Conceptual Issues

Let us begin the detailed discussion leading to these conclusions by considering whether intending harm as a means to terror, which terror is in turn intended as a means to something else, is sufficient to make an act be ST, when NCs are seriously harmed against their self-interest and without their consent. This is just a conceptual question independent of whether the act is a more serious wrong than causing collateral damage. In this part, I offer considerations against a positive answer. (In part 2, I consider a view that favors a positive answer.)

Consider this question first in cases in which it is assumed to be permissible to bomb a military facility, causing death and terror as collateral damage. The act will be permissible, if we use standards developed in just-war theory, when the military goal is sufficiently important, the bombing is a necessary means to it, it has a significantly high probability of bringing about the goal, and the collateral damage is a proportionate evil relative to the goal to be achieved.⁸

P(i) Suppose that the United States is controlled by Nazis. The Resistance's aim in bombing the WTC is to destroy ongoing military operations in it. Suppose that this would be morally justified despite the side-effect deaths and terror to nonresponsible NCs. However, the only pilots willing to fly the planes and drop the bombs on the WTC are those who are interested in bringing about the side effect of this, killing and terrorizing NCs. Would the character of the operation change from military bombing with collateral damage to terrorism merely because of the pilots' intentions? To focus on the intentions alone, I assume that these pilots are definitely unable to kill and terrorize any other people than pilots who lacked the bad intention, given that they would bomb the military facility in exactly the same way as the other pilots and would have no option to do otherwise in the case as imagined.⁹

It may be said that there are multiple acts here: the act of the pilot and the act of the authority in the Resistance who authorizes the attack. The former is an act of terrorism, the latter is not. It may also be said that in this case an authority uses terrorists to carry out a nonterrorist mission (at least if it knows of the pilots' intentions).¹⁰

P(ii) Now suppose that there is no Resistance movement that will bomb the WTC in the Nazi United States. Instead, the bombing is planned and done by Baby Killer Nation, a group that takes an interest in destroying the Nazi's military operations in the WTC only because it is a means to satisfy its own desire to cause harm and terror to NCs. This is a desire Baby Killer Nation's members act on only when there is such a type of act in a just cause that can serve as a cover. Furthermore, they would never perform any acts in bombing the WTC that will cause more side-effect damage than would be caused by bombers who lack their bad intentions. Should their act of bombing the WTC be classified as an act of terrorism rather than a military operation? I do not think so. This conclusion goes beyond the minimal claim stated above, for the intention of the authority is not determinative here. Rather, the new minimal claim might be that the conditions under which the authority will allow itself to act on its intention to harm and terrorize NCs and its doing no more than what others seeking to destroy the military operation without bad intentions would do are determinative of the category into which we should place the act.

I believe that the conceptual issue that we are dealing with in P(ii) is analogous to a problem in the philosophy of law. Suppose that A intends to murder B but deliberately waits to do so until B attempts to murder him, so that A's act is no different from what C would permissibly do in self-defense if B attacked C, except, of course, that A's intention is different from C's. Indeed, we can suppose that A would supererogatorily have refused to defend himself against anyone's attack on him but B's. In other words, his act is not overdetermined in the sense that he also has a constant intention to save himself. I believe that even in this latter case, A's act should be classified as self-defense despite his intentions because his act is what is actually needed to defend himself.¹¹

Of course, in P(ii), unlike a self-defense case, Baby Killer's actions that destroy the Nazi regime may not help save Baby Killer Nation itself. Rather they save others from the Nazi regime. An analogue to this is what I call the Bad Man Trolley case. Suppose that a trolley is headed toward killing five people, and it would be permissible for a bystander who wants to save the five to redirect it toward killing one person instead. Imagine, however, that the bystander is a bad person. He would not bother to save the five lives but for the fact that he recognizes that the one person who will be killed if he redirects the trolley is his enemy. He turns the trolley only in order to kill his enemy, but only because he is in a context where the five will be saved by his turning the trolley. His turning the trolley, exactly as a person who merely wanted to save the five people would turn it, is not, I think, a murder.

It might be suggested that the unwillingness to characterize P(ii) as terrorism is linked to a deeper point about the characterization of terrorism. Perhaps it involves the agent not only intending harm and terror but also intending that those harmed and terrorized know that it is his intention to harm and terrorize them. Bombing only on the condition that there is a militarily (or economically, etc.) useful target to hit (which will cause harm, leading to terror) may conceal the intention to harm and terrorize people. If making clear to the populace one's intention to harm and terrorize were a part of terrorism, then that would account for why P(ii) is not thought of as terrorism.¹²

I do not think that the latter claim is true. For it would not be inconsistent with bombing only on the condition that there is an appropriate military target to put on one's plane a sign that says, “Aiming to harm and terrorize you.” (Call this the Baby Killer Message case). If one aims to avoid doing a harmful act that one has a duty to refrain from doing, then Baby Killer Nation can still accomplish that aim (by bombing only on condition that there is an appropriate Nazi military target) and also make sure that its other intention—to harm and terrorize—is not concealed. Should we still refuse to characterize the Baby Killer Message case as terrorism? I think so. But if this is true, then it weakens the claim that the reason we would not classify P(ii) as terrorism is that the agent does not aim to make clear to its victims that it intends them to be victims.

P(iii) Suppose next that Baby Killer Nation would have bombed the Nazi WTC even if there had been no military significance to its destruction; it has an unconditional intention to act just to produce the harm and terror. Suppose further that such an act would have been impermissible. But in fact, there are Nazi military operations worth destroying in the WTC, and Baby Killer Nation knows this. Further, it does nothing in bombing the WTC other than what a genuine resistance lacking intentions to harm and terrorize would have done. Its bombing could then be categorized as a military bombing rather than terrorism, I believe. What the Nation would have done—bomb in the absence of the military operations—does not bear on our description of what it actually does in this case. Hence, we should move beyond the claim made in P(ii) that the conditions under which the authority will allow itself to act are always relevant to deciding whether an act is terrorism.

In an analogous Bad Man Trolley case, the same is true. That is, suppose the bad man would have turned the trolley on his enemy even if doing this did not also save the five, but in fact it will save the five, the bad man knows this, and he does only what someone who wanted to save the five would have done. Then his act should not be classified as murder. In a self-defense case, the same is true, I believe. That is, suppose that A would have killed B even if B were not attacking him, but B in fact is attacking A, A knows this, and A does only what C could permissibly have done in self-defense. Though A acts only intending to kill B as an end, his act should be classified as an act of self-defense.

P(iv) What if we assume that Baby Killer Nation was ignorant of the Nazi military operations in the WTC and bombed it with the intention to harm and terrorize NCs? Nevertheless, if members of the Nation actually destroy the ongoing military operations and they do no more than what a genuine resistance lacking their intention would have done, I think that their act should be characterized in the following way: they attempted to perform an act of terrorism but succeeded in doing something else. In an analogous trolley case, when the bad man does not know that five could be saved by redirecting the trolley, I think he attempted to murder his enemy but succeeded in doing something else. In an analogous self-defense case, A does not know that B is trying to attack him. When A attacks B, trying to murder him, he actually succeeds in defending himself, though he may be guilty of attempted murder.

P(v) So far, Baby Killer Nation has been presented as intending harm and terror as an end in itself while it is bombing the military operations in the WTC in the Nazi United States. Most terrorists do not intend harm and terror as ends in themselves but as means to achieving other goals. So let us suppose instead that Baby Killer Nation intends the death and terror as a possible means to the surrender of the Nazis. And indeed, it is as likely that the populace will surrender from terror as from lack of munitions.

Nevertheless, the Nation acts on this intention only when there is another possible route to victory, namely, stopping the military operations in the WTC. I believe that its act is then not terrorism. For suppose that members of a resistance movement foresaw but did not intend that people would surrender as a result of collateral terror in response to collateral deaths from their bombing the WTC, before the military effect of bombing the WTC was decisive. This would not make their bombing the WTC terrorism (or impermissible, if bombing the WTC is permissible when the deaths and terror are mere foreseen side effects.)¹⁴

I have been considering the conceptual question of whether intentions to harm and terrorize NCs are sufficient to make an act that harms and terrorizes NCs terrorism, and have so far suggested “no.” Now consider the means/side-effect distinction separately from intention. Suppose that in all these WTC-Military cases the harm and terror to which it leads are, in fact, the causal route through which the downfall of the Nazis comes about, as the terrorized citizens overthrow the Nazi government independently of any military effect of the destruction of the military operations. As I noted in discussing P(v), the actual causal significance of harm and terror are not enough for us to call the bombings of the WTC acts of terrorism. How people react to what may just be a side effect is not enough to determine the character of the agent's act. Would the combination of intending harm and terror with these being the actual causal route to surrender be sufficient for terrorism if neither alone is sufficient? Not in a variant of P(ii): Baby Killer Nation's bombing the Nazi WTC, intending the deaths and terror of NCs, and such terror actually leading to surrender do not imply that its act is terrorism given that destroying the military facility, its condition for bombing at all, could also be a means to surrender.

2. Moral Issues

Let us now consider the moral, rather than conceptual, significance in P(i)–P(v) of an agent's intention to harm and terrorize, and then, separately, the moral significance of the causal role for harm and terror in bringing about an agent's goal. (I shall return to the conceptual issue.)

Consider intention first. I believe that acts are morally worse when the intention is wrong, in the sense that there are morally wrong attitudes toward persons present and they are also efficacious in leading to acts. I think that it is worse to seek harm and terror (even as means rather than as ends in themselves) rather than act despite their occurrence (or even on condition of their occurrence) but not seek them, other things being equal.

However, this is not the same as saying that the intention makes an act impermissible when another act like it in all respects except for the intention is permissible. Nor is it the same as saying that the intention makes an act a more serious wrong, in the sense, for example, that there is a higher threshold of good to be achieved that must be met in order to override the prima facie wrongness of the act and make it permissible.

In particular, in P(i)–P(v), we would not require the resistance movement or Baby Killer Nation or their pilots to refrain from bombing the military target, even when we know that they will act only for bad reasons. Furthermore, I think we would not require them to refrain even if we would require individuals to refrain from doing acts against the Nazis that were impermissible on grounds other than the agent's intentions. (That is, we need not merely be allowing impermissible acts for the greater good.) Nor would we require the bad man in any of the trolley cases to refrain from turning the trolley. Nor would we require A to refrain from attacking B in any of the self-defense cases, even if we knew he would act for bad reasons and we had no reason to favor the outcome in which he defended himself over the one in which he did not. We would certainly not require these agents to refrain from attacking a target or saving the five if it were the agents' duty to do these acts. Even Kant would have agreed. He argued that a shopkeeper should give the right change to a child from concern to fulfill his duty to do so rather than from mere prudence. But if a shopkeeper would return the change merely from prudence, Kant would not claim that his returning the change was morally impermissible, though his act would have no moral worth.¹⁵

Now consider the means/side-effect distinction separately from the intention/foresight distinction, by considering the actual causal route to surrender. Does the fact that the actual causal route to surrender will be through the harm and terror mean that it is impermissible to bomb the Nazi military facility whose destruction could also have helped bring about surrender for military reasons? I do not think that this is so, even if it would have been wrong to bomb in the absence of any facility whose destruction was of military use. One need not be required to refrain from destroying a military facility whose destruction could promote an objective good (such as ending Nazi rule), when doing this is justified despite side-effect deaths, simply because people will respond to this side effect and surrender. Nor does the good that one can achieve though destroying a military facility have to be greater in order to justify one's act when one intends that it be the collateral harm and terror that will actually cause surrender.

Finally, returning to the conceptual issue, suppose that some of my conceptual suggestions in C(1) about P(i)–P(v) are incorrect, and some of these cases should be categorized as terrorism. Thomas Scanlon has argued that intentions can change the meaning of acts.¹⁶

These results would imply that we could not rule out the permissibility of terrorist acts without first considering other possible effects of the acts, such as destroying military targets, that could justify the acts, even when it is the harm and terror that turn out to be causally efficacious to the good effect that justifies the destructive act.

I have been dealing with hypothetical cases where the act of destroying the Nazi WTC is assumed to be permissible (P). I argued that the intention to harm and terrorize (as an end or mere means), even when the harm and terror are actually the means that achieve one's end, does not account for the supposed distinction in moral permissibility between bombing that involves collateral damage and terrorist bombing. However, I also noted that the intention to harm and terrorize people can be additional bad elements of the act. It can lead to justified indignation toward attitudes people take to other people and the considerations they take to justify acts (such as death and terror) that do not justify acts. Nevertheless, this need not imply that the act they perform for bad reasons is a wrong (impermissible) act.

3. Impermissible Acts

Now let us consider the role of intention and also of the actual causal role of harm and terror in deciding what type of wrong act is done (a conceptual issue) and how serious a wrong it is when the act is assumed to be impermissible (I). (That one act is a more serious wrong than another need not mean that it is more wrong. Some hold that all wrong acts are equally wrong. However, murdering someone is a more serious wrong than breaking his arm, other things being equal.) The act of bombing the military facility will be impermissible, at least, if the death and terror it leads to as a side effect are out of proportion to the military goal to be achieved. First consider imaginary cases.

I(i) Suppose that it makes military sense for Al Qaeda to blow up the WTC in a United States not controlled by Nazis, because of military operations going on in it. However, it is morally unjustified for it to attack because of the side effects of death and terror. (It could also be that, given the type of conflict Al Qaeda is involved in, it is not even permissible for it to destroy the opponent's homeland property, even if this harmed and terrorized no one.) No pilot is interested in taking the job for the purely military reasons. However, some pilots are interested in doing the job because it will kill and terrorize people. Does hitting the WTC with this intention, carried out in exactly the same way as pilots who do not intend to kill and terrorize would carry it out, become an act of terrorism, although Al Qaeda authorizes the mission to destroy the military operation? Even though the destruction of the WTC would be impermissible, I do not think that its impermissibility should alter our conclusion from what it was in P(i), namely, that the pilot's intentions do not determine how we should categorize the bombing. However, in this case, where we are considering which wrong was committed, it will be the wrong of impermissible bombing of a military facility rather than the wrong of terrorism.

I(ii) But suppose that all is as in I(i) except that Al Qaeda (including its pilots) is not interested in doing what it makes military sense for it to do. It is only interested in killing and terrorizing NCs. Nevertheless, it will not cause death and terror unless military factors are present to create a militarily rational cover for its operation. If it bombs only on this condition and does no acts that would not be done by someone interested only in getting rid of the military operation, is its act terrorism rather than impermissible military bombing?

I have mentioned the view that an authority's intention can change the meaning of the act and that “terrorism” should be applied when the meaning of the act is that people seek to harm and terrorize. If this view is correct, I(ii) is terrorism.

Alternatively, we might get support for the negative answer by considering a self-defense analogy.¹⁷

So suppose that C would be engaged in impermissible self-defense against an attacker B in killing him even though he aims only at self-defense. Can we classify the wrong that A does, if he does only what C does, as impermissible self-defense when A aims only to murder B but waits until B's attack on him to kill him? Here the issue at stake is: May we ignore intention in characterizing the type of act done when the act is impermissible (as one view holds we might when it is permissible)? Might we focus instead on the condition that sets limits on when the bad intention will be acted on rather than on the intention (as was at first proposed in discussing P[ii])? Might the answer depend on whether there is a “core of appropriateness” to the act, even if it is ultimately impermissible? (That is, defending oneself in the face of an attack is appropriate but is made impermissible in virtue of some other factor, such as that the only available response is excessive. However, attacking a nonthreat does not have the same core of appropriateness. Attacking military supplies to be used unjustly against one is appropriate, even if ultimately impermissible due to collateral damage. Attacking a public garden does not have the same core of permissibility.)

In all of these cases, the agent should refrain from his act. This, I think, makes it more likely that we refer to his intention to characterize the act's type. So in cases I(i)–(ii), the conceptual results will be closer to those of the “meaning” approach to categorizing acts. It is also possible that the seriousness of the wrong done is determined by reference to this factor. If this were true, it would indicate that the role of intention in determining the type of act could be different for permissible and impermissible acts. Nevertheless, the impermissibility of the act is still established on grounds other than the presence of the intention to harm and terrorize.

Consider how we might reason about the relevance of intention in cases I(i)–(ii). Suppose that A, like C, would have impermissibly defended himself against attacking agents other than B, even though he did not begin with the aim of killing them, merely waiting for the cover of responding to an attack. Suppose also that A set the self-defense scenario as a condition of killing B. I suggest that we might then classify A's act in killing B in the same way as C's wrongful self-defense. This is because the case suggests that A has two intentions in the circumstance when he kills B—to defend himself and to eliminate B—and the self-defense goal is primary in that it suffices on its own as a reason for action and also sets the conditions for action on the other intention.¹⁸

But if A, unlike C, would not have impermissibly killed any other attacking agent except B, this suggests that he is completely uninterested in self-defense, even though he conditionalizes acting for his true intention on self defense, and so his impermissible act should be classified as a different wrong more serious than C's. Analogously, in I(ii), if Al Qaeda has no interest in the military target per se, then the wrong in bombing will be terrorism, even when it is made conditional on bombing a military operation. Finally, suppose that in variants on cases I(i)–(ii), Al Qaeda is interested both in destruction of the WTC and in harm and terror, but the harm and terror are a necessary (if not sufficient) condition for its acting. That is, Al Qaeda would not fulfill its intention to destroy the WTC unless it also achieved this other goal. Here, attacking the military is one goal of the action but is not sufficient on its own for action. Harming and terrorizing are also a goal of action but are not sufficient on their own. In this sort of case, I think that the impermissible act would be terrorism, at least in part.¹⁹

For completeness, let us now consider I(iii) and I(iv).

I(iii) Suppose that all is as in I(ii), except that although Al Qaeda knows about the military usefulness of destroying the WTC, it would have destroyed the WTC merely in order to produce death and terror even if there had been no ongoing military operations there. Its wrong is thereby the wrong of terrorism.

I(iv) Suppose that all is as in I(iii), except that no one in Al Qaeda knows of the actual military operations in the WTC and it aims to destroy the WTC in order to kill and terrorize. Here again, I think that the wrongful act is terrorism.

Should the actual causal route (independent of intention) to an end affect how we categorize an impermissible act or the degree of its wrongness. I do not think so. For example, suppose bombing the WTC military facility for military reasons would be impermissible at least because of collateral harm and terror. Does the act become terrorism or a more significant wrong just because the populace actually surrenders due to the harm and terror? I do not think so.

Suppose this analysis of when we should apply the term terrorism, arrived at by considering I(i)–(iv) and variants, is correct. Then it will be important in describing the actual case of bombing the WTC on 9/11 to consider whether bombing the WTC (where no military operations were ongoing) is something Al Qaeda would have done even if no deaths or terror occurred. The bombing would certainly be useful to show the vulnerability of one's opponent and perhaps to destroy some of its economic power. We could certainly conceive of some other agent (analogous to C in the self-defense cases) that did not intend to harm and terrorize NCs as a means or end yet would exhibit instrumental (though not moral) rationality in bombing the WTC for such reasons. Of course, such a bombing would be morally impermissible. It would be impermissible either because of the collateral damage or because it is not true that one may destroy the homeland property of one's opponent (let alone NCs) in just any sort of conflict. Nevertheless, the wrong would not be classified as terrorism. Suppose Al Qaeda would have bombed the WTC on 9/11 if no people would have been harmed or terrorized. Suppose that in its actual bombing on 9/11 it did nothing other than what someone who was interested only in destroying the WTC would have done. Then if what has been said about I(i)–(iv) is correct, even if Al Qaeda also intended the harm and terror and it was harm and terror that actually had the biggest useful effect for its cause, its impermissible act would not be terrorism.

Intention to harm and terrorize is, I believe, a necessary condition for ST.²⁰

I emphasize that my claim about Al Qaeda is different from the following claim that some people make: For all we know, given their observable acts, Al Qaeda did not intend harm and terror but only destruction of the WTC, despite foreseen collateral damage, and therefore we cannot describe its act as ST. (Some also conclude from this that we cannot say that what it did was impermissible.) My claim is that if Al Qaeda intended both the destruction of the WTC and the harm and terror, was willing to seek the former without the latter, and took seeking the former as a condition of seeking the latter, then its act should be characterized as something other than terrorism, but it can also be true that this other type of act is impermissible.

1. The Necessary/Nonnecessary Means Distinctions

Intending harm and terror is necessary, I think, to make one's act be terrorism, but it is the conjunction of this intention with bringing harm and terror about when they alone are effects of one's act that could be a means to one's end that could make one's act be an example of impermissible terrorism. Cases that have been standardly used to illustrate terrorism have this characteristic. For example, they involve a pilot who drops bombs directly on children and intends the harm to them and the resulting terror as a means to surrender, when no other aspect of his act can help cause surrender that would justify side-effect deaths and terror. It has been mistakenly thought that the intention and/or actual causal route through harm and terror were the crucial characteristics for impermissibility. But if the harm and terror are effects that do bring about one's end when something else one brought about through one's act could independently also bring about one's end, we may not yet have either terrorism or an impermissible act, even with a bad intention. This was true in the case where hitting military operations in the Nazi WTC could also have brought about surrender by way of reduced munitions, if harm and terror had not actually led to surrender. The fact that harm and terror would be the only effect of one's act that could produce one's goal implies that it is necessary, given one's act, for them to have a causal role in achieving one's goal. (Note that the necessity of a particular causal route to produce one's goal, given one's act, is not the same as the necessity of that causal route tout court. The fact that one might have done many different acts to bring about one's end that did not require harm and terror is consistent with the act one actually does leading to one's end only through harm and terror as a means to one's end.) But is the necessity of a role for harm and terror, given one's act, by contrast with the possibility of one's act leading through some other route to one's goal, sufficient to make the act impermissible?

2. The Produce/Sustain Distinction

There is reason to doubt this, as is shown by the following cases.²¹

A key point in distinguishing the permissible from the impermissible cases where harm and terror are necessary, given one's act, to achieve one's goal is the different way in which the harm and terror come about and, as a consequence, the different causal roles of harm and terror. In the Sustain Destruction case, the harm and terror are brought about by the destruction of the WTC that, if only it could be sustained, would by hypothesis be the goal that could justify the existence of collateral harm and terror. This contrasts with the harm and terror being a necessary means, given one's act, to the initial destruction of the WTC, not produced by that destruction and merely necessary to sustain it. (It also contrasts, I believe, with harm and terror being the side effect of a mere means (such as dropping a bomb) to a goal (such as producing the destruction of the Nazi WTC or sustaining its destruction by dropping further bombs on it.)²³

I conclude that intending harm and terror as means or ends is necessary for a case to involve ST, but it is not this intention that contributes to the prima facie impermissibility of the act that causes harm and terror. Furthermore, neither an actual nor a necessary causal role, given one's act, for harm and terror in achieving one's goal makes for the prima facie impermissibility of the act that harms. Rather, what can be crucial for the prima facie impermissibility of an act is that the harm and terror have a necessary role, given one's act, in producing one's goal or are the effects of one's act in the absence of any other justifying effect. This contrasts with harm and terror being the effects of achieving one's goal and such effects having a necessary causal role in sustaining it.²⁴

These results imply that a modal operator (such as the necessity of harm and terror, given one's act, as a means to producing one's goal or the possibility of doing without them in producing one's goal) should sometimes play a role in characterizing the impermissible or permissible act.²⁵

1

In order to make this clearer, consider the Human Tinder case: an agent bombs people as a means to start a fire that will blow up a military target. Or consider the Stampede case: an agent bombs people as a means to create terror in others so that they stampede and trample a military facility, thereby terminating a military operation. In these two cases, harm and terror are mechanical means to military ends. I say “mechanical” because the destruction of the target in the Human Tinder case could be produced by any burnable substance. And the terror involved in the Stampede case is like a fright-and-flight instinctive response and is in this sense mechanical.

By contrast, it is often true in ST that the harm and terror are meant to influence people's judgment and will, especially in the political sphere. Here, fear gives rise to a prudential or altruistic judgment (that one should save oneself or others) that is meant to provide one with a reason to alter one's behavior or policies. The fear is not a factor that causes one to bypass the exercise of judgment and reduces one to an instinctive fright-and-flight response. In the Stampede case, people act in a panic. This form of terror is somewhat like torture without the physical abuse, insofar as the torturer tries to undermine the victim's will rather than give him a reason (namely, to avoid pain) to change his willed behavior. We might call this the mechanical/nonmechanical distinction.

In the Human Tinder and Stampede cases, the attacker fights with the people harmed and terrorized in the sense that he uses them as tools in defeating the military. But often in ST, at least when the NCs are in a democracy and are not juveniles, the terrorist is trying to get people to change their government's policies. So there may also be a distinction to be drawn between the military use of harm and terror and its political use.

Table 1 lays out the possible combinations of these distinctions:

Notice that there could be a nonmechanical use of people to defeat the military, as when fear leads people, as a matter of prudence or altruism, to consider sabotaging their own country's military operations. And there could be a mechanical use that affects the political realm, as when people's terror leads them to stampede and interfere with an election.

2

Let us consider the case of nonmechanical use for political purposes (NP) in more detail. When the aim is to create terror that will lead people to create political pressure on their government, the terrorist is trying to change people's judgment and deliberately willed political behavior.²⁷

The terrorist in the type of case we are considering is also trying to get people to behave in ways that exhibit a lack of the virtue of courage, it might be argued. When an agent harms people, she may be trying to make them unable to fight. When she terrorizes them for nonmechanical purposes, she tries to make them unwilling to support a fight or otherwise stay the course, though they are not unable to do so. (The Stampede case involves a mechanical form of terror that makes one psychologically unable to control oneself, so it differs from the unwillingness to stay a course that I am now considering.) It would be better if people did not give in to fear, as doing so leads them to take factors on which one focuses merely because of fear as reasons to change policy. They cannot be proud of their behavior when they change a policy merely because of fear and independent of merit. This is especially true when the fear is out of proportion to the real increase in the probability of death and destruction, for example, when there is a much higher probability of dying in a car accident than from a terrorist attack.²⁸

Focusing on these factors in ST raises the issue of whether and when it is in fact cowardly to give in to fear as a reason for seeking a change in policy. It seems reasonable to surrender from fear when a criminal says, “Your money or your life.”²⁹

In response to this last point, it may be argued that U.S. foreign policy (if that is what the terrorist is trying to change) is a matter of national self-interest, not principles, and so if changing the policy to stop terrorist attacks were in U.S. national interest overall, then there would be nothing morally objectionable (or cowardly) in its doing so.³⁰

However, even if the substance of U.S. foreign policy were only a matter of self-interest and not principle, giving in to ST would not be like ordinary action from self-interest. In ST (and in criminal threats) the agent takes or threatens to take something from people (e.g., their lives) to which they have a right, in order to make it be in their interest to trade something else (e.g., their foreign policy) in order to get back (or avoid losing) what they have a right to have without such a trade. Contrast this with a nation that says, “We will not give you our oil unless you change your foreign policy” (the Oil Threat case). This, too, may be done in order to cause fear in people and lead them to pressure their government to change its foreign policy. But such a threat does not involve taking U.S. oil (or anything else to which U.S. citizens are entitled) and then trying to trade it back to them for a change in U.S. foreign policy. Indeed, if a threat of withholding oil were intended to produce terror, then it might be an instance of morally permissible NST. (One might resent such an attempt to alter one's policies, but this would not be enough to show that it is impermissible. It is like the embargo that the United States used to try to alter Saddam Hussein's policies.) The moral objection to the attempt to make something be in one's self-interest seems appropriately greater in the case of threatened ST than in the Oil Threat case. Not acting on this moral objection (by standing fast in the face of terrorism) for reasons of self-interest or even concern about the interests of others seems morally objectionable and not only shortsighted. This would be true even if U.S. foreign policy were only a matter of national self-interest to begin with.

Perhaps this claim might be challenged in the following way.³¹

Consistently or inconsistently, however, the United States rejects the view that ST is just another way of raising costs and that, given its own conduct, it is not in a position to argue that there are moral constraints on pursuing aims. Indeed, given that ST actually increases the likelihood of death and serious injury to NCs very little, it is probably resentment against what it considers the terrorist's impermissible means, combined with the view that U.S. foreign policy is to some degree principled and the immoral means it employs are not as bad as ST, that may be underlying the attempt to extirpate ST. So we are back to the view that changing principled foreign policy from fear, when this requires giving in to pressure produced in immoral ways, is objectionable.

I conclude, therefore, that if ST is an attempt to give people a reason based on fear to give in to impermissible types of pressure and to change principled policies, when the actual increase in probability of death or serious injury is and is perceived to be small, then ST is an attempt to get people to behave in a cowardly manner. (If it is expected that people will unreasonably overestimate the increased probability of death, then there is at least an attempt by the terrorists to get people to behave in an irrational way.)

Giving in from fear is not the same as reconsidering the merits of a policy after ST, when one sees how strongly some people object to the policy. I suggested above that ST agents might be trying to set their victims on what they see as a morally right path. They may induce cowardice and irrationality to do this, but they may also seek conversion, a reconsideration of the merits of a policy. (To do this, harm and terror should be accompanied by dissemination of arguments for their side.) Suppose that one changed one's policy because on reconsidering it, one saw its flaws. (And not only in the sense that one sees that the policy is not really in one's self-interest because it leads to terrorism. One sees its flaws independent of that effect.) Will others believe that one changed the policy because of fear instead? That is, that one would have changed it even if it were not flawed? If so, this might encourage ST as a tactic. Indeed, even if others believe that one changed the policy on grounds of merit only after ST, then ST might seem to be a successful tactic for getting one to attend more closely to the merits of one's policy! This, too, could encourage ST.

If these things might happen, then it is no longer a simple matter of changing one's policies on the merits. The additional possible effects of encouraging ST might have to be considered and weighed in the balance. Here lies the poignancy of the Spanish case in 2003. The majority of the Spanish people all along thought that the policy of supporting the U.S. war with Iraq was wrong (and not just on self-interested grounds). But they seemed to put real pressure on their government after an ST act, and so some causal chain was suggested. But should they have gone on paying with their lives for what they always thought was wrong, just so as not to be thought to be giving in—or even being responsive in some other way—to ST?

3

(a). Suppose that an act of ST is an attempt to make people act cowardly, to prey on their weakness or on their less-than-optimal rationality in calculating risks. Does this help make the ST act prima facie wrong in the sense of being impermissible? I believe so. This is because it is prima facie wrong to play a role in undermining people's judgment and will. It is also because these effects are, given the agent's act, necessary to produce his further end (rather than only side effects). However, these factors are also present in the Oil Threat case, but they do not make that threat overall impermissible. Do these factors make an act that is impermissible on other grounds (e.g., it involves impermissible killing) be a more serious wrong? One way to test for this is to hold constant in all cases the amount of harm and terror and their causal necessity in producing an outcome and ask whether the agent does a more serious wrong in the NP case than in either an MM case or an MP case. One problem with this approach is that we might not be able to equalize for terror. For the terror that will undermine judgment entirely and produce panic must, it seems, be more intense and hence greater than the terror that does not.³²

Is it a more serious wrong to try to influence politics and thereby win a conflict than to merely influence the military when both are done through a mechanical use of harm and terror? Perhaps. Is it a more serious wrong to influence politics through the effect on judgment described above rather than mechanically? I think that it is in some ways “uglier” but I do not think that this makes the agent's act a more serious wrong. Why is it uglier? Because it may make ST a morally worse event for its victims. After all, they will act in a morally wrong way (e.g., in a cowardly way). The responsibility for any wrong behavior by the victims will be theirs when they let their judgment be swayed by fear. They have the option of not doing this. To some degree they even have the option of not feeling fear, as we have more control over this emotional response than over physical harms, at least when there is no panic attack (as in the Stampede case). For this reason the victims may prefer to be used mechanically, for then the responsibility for the outcome lies solely with the ST agent.

However, I do not think that it is the responsibility of an agent involved in a conflict, when he decides whether to use harm and terror mechanically or nonmechanically, to choose on the basis of sparing his opponents from their own bad decision-making (if this will be bad only for his opponents). (By contrast, in nonconflict situations, I think that we should act to promote and support each other's good reasoning and moral virtues.) Similarly, we need not correct the errors of our opponents when this will harm our side, even when they consult us for advice. However, the victim's desire to avoid being placed in a situation where he will behave badly is an added incentive for him to try to extirpate ST, even if not because it makes what the ST agent does a more serious wrong.

Suppose that an agent did want to help the people he harms and terrorizes to avoid cowardly behavior or bad risk assessment, so that they would not later be ashamed of their conduct. He could make their behavior not exhibit these defects by increasing the harm he does, so that it would be truly reasonable for them to fear great further damage and be reasonable for them to surrender. But this has several problems. First, it would do more harm to people, and that seems too great a price to pay in order to make one's opponent's conduct reasonable and noncowardly. Second, it seems unnecessarily solicitous toward someone in a conflict to want to cushion the blow of choices they are responsible for making.

The first problem could be remedied³³

(b). So far I have argued that, due to the role of NC responsibility and the lack of a duty to be solicitous of one's opponent's behavior, achieving one's end via nonmechanical means does not make an agent's act a more serious wrong than achieving the end via mechanical means. Can we also argue that using the nonmechanical route—even though it will depend on eliciting poor judgment or cowardice on the part of one's opponent—is morally better than producing the same effects through mechanical means (holding the amount of harm and terror and their causal necessity in producing an outcome in both cases constant)? That is, if it were a more serious wrong to use the mechanical, rationality-undermining means, then this, rather than merely the refusal to be solicitous of one's opponent, should lead one to use nonmechanical means.

In answering this question, recall the claim that mechanical terror is like torture without physical abuse of the one terrorized. (There is, of course, still physical abuse in ST of those whose deaths lead to terror.) Indeed, psychological torture is a bona fide category of torture that does not require physical abuse. Some think that there is a human right not to be tortured. I do not think this is true, if we understand by a human right one that is had by anyone just so long as he or she is a human person.³⁴

It may be that there is a human right not be tortured in the sense that being a human person is enough to give rise to a right not to be tortured. (Indeed, this may be the correct way to understand any human right.) But this does not mean that this right cannot be defeated or forfeited on occasion, for example, in virtue of acts that the person performs. So some human persons may lack the right not to be tortured in certain circumstances. It also may be true that someone's right not to be tortured is not defeated or forfeited merely because it would be useful to torture him when he is not doing or has not done acts that threaten others and whose bad effects could be stopped by torturing. Let us consider such impermissible torture in the following discussion.

Would it be morally worse to try to get a person to divulge information by breaking the person down to a nonrational, panicked state or by threatening him so that reasons of self-interest stemming from fear overcome his moral impulse to keep a secret? Even though the second course involves trying to corrupt someone, I think that it could be permissible to take it when reducing someone to a nonrational, panicked state is impermissible. This is so even if the corrupted person becomes less worthy of respect and the reduced person would be only an object of pity. This would explain why corrupting (by unpleasant or pleasant prospects) is considered a morally permissible interrogation procedure but psychological terror is not.

But perhaps the fear leading to panic is greater than the fear that gives one a prudential reason to divulge information, and this is why the former is prohibited when the latter is not. To deal with this possibility, we should imaging that the panic period is short and the lesser fear is extended in time. Alternatively, we could imagine options involving no fear at all. Suppose that we could for a brief duration give someone a truth serum that made it impossible for him to exercise self-control, so that he blurted out anything he knew. The alternative way to get information is to bribe the person with the prospect of rewards rather than punishments. Again, I suggest that it is no solution to the problem of corrupting someone to undermine his rational agency instead, even for a short while in the way described.

Why might it be true that it is preferable to avoid mechanical psychological means? This is a difficult question. I suspect that it may be better in some way for the person to lose control in a way that, in a sense, reduces him to a nonperson level for a short while than to be corrupted. However, it is contrary to the importance of being a person to try to elicit responses from him by bypassing his rational control in this way, and those who do it bear full responsibility for his state and what he does in it. The alternative will involve encouraging him to make choices that reflect badly on his moral character. But the outcome, whatever it is, will be a function of his full humanity being exercised, not short-circuited, and more responsibility for his ultimate state will lie with him. We might say that while the short, tortured period could be (comparatively) better for him, it is more at odds with the importance of him. For similar reasons, it might be morally worse to create the mechanical panic response in a population rather than the fear that corrupts their judgment, even if we could equalize the fear involved.³⁵

Hence I conclude that ST is prima facie wrong in part because it employs undermining judgment and will. However, such nonmechanical corruption does not make ST a more serious prima facie wrong act than mechanical use of terror.

1. Terror

(a). Its Significance per se. In order for there to be terror that is not unreasonable, what must its object be? Could it be a broken leg (in an ordinary person)? I doubt it. It must be fear of something very bad, such as death or grave injury to self or others. However, such terror could come from seeing someone else being harmed only to the extent of having his leg broken, if one thinks, “That's only the beginning.”

There seem to be two major types of terror, as noted in Section III.E: T(a)—fear that gives one a reason of prudence or altruism to alter behavior; and T(b)—fear that undermines rationality so that one does not act for reasons of prudence or altruism as one is no longer reasoning; rather, one acts in a panic. Each type of terror might be temporary or permanent and (in each category) frequent or infrequent. (Permanent, infrequent terror would occur, for example, if the world were never again without ST, but it occurred no more than once a year.) As noted above, the T(b) form of terror is like torture without physical abuse, insofar as torture tries to undermine the will rather than give one a reason (namely, to avoid pain) to change one's behavior. T(b) is involved in the Stampede case mentioned above but not in nonmechanical uses of terror.

Are any of these conditions of terror worse for their victims than death or grave bodily injury, keeping constant the number of people affected? Neither temporary T(a) or T(b) nor permanent T(a) or T(b) if they are infrequent are as bad as death. But perhaps permanent and frequent T(b) would be as bad as or worse than death or grave bodily injury.

To confirm some of these judgments, consider the significance of temporary T(a) all by itself. That is, suppose that we aim to produce such terror but without any harm. We bomb some trees because we know that people will think we are trying to kill them and they then become terrorized and pressure for a change of policy. However, we do not and would not harm anyone. Call this the Trees case. (A Trees-Stampede case could be imagined involving T[b], where we bomb the trees in order to produce fear of death, leading to a panic that causes a stampede.) Or we might spread rumors that we have weapons of mass destruction (WMD) in order to terrorize people, though we have no such weapons (the Rumor case). Are these cases of terrorism? I believe so, but they are NST, as there is no physical harm (intended or caused) to anyone.³⁶

Of course, worse events can come from terrorizing without harm than from even killing, if the reaction to the former will be much worse than to the latter. So suppose that Saddam Hussein had terrorized (T[a]) people with rumors of WMD when he had none, and this led the United States to go to war, which resulted in many deaths on all sides. This could be a worse outcome than ST that did not produce such an extreme response. Even a mechanical stampede not caused by harm to anyone could result in more deaths and injuries than a few deaths that alternatively would have been caused by deliberately harming people. But the reason it will be worse is that it produces serious harm rather than just more temporary T(a) or T(b).

I have been arguing that terror per se is typically less bad and causing it has lesser moral significance than killing or gravely injuring, other things being equal. But this need not be true simply because being harmed is always morally worse than being wronged in nonharmful ways. For example, I think that killing a person against his will when it is for his own good is nonharmful but serious wronging. It is more serious than harming him in some minor way. So, possibly, one should do what wrongfully breaks an arm in each NC rather than what wrongfully terrorizes them without harm, if one has a choice.

However, in deciding which to do, one must not only consider how terror compares with less serious harms; one must also consider the dynamics of responsibility for both harm and terror. (This issue is referred to in the discussion of victim responsibility in Section III.E.). That is, when we harm someone, even in a small way, it is we who are responsible for the harm. When we terrorize nonharmfully, and T(a) leads someone to change his conduct from fear, the responsibility for the change seems to be his. For when we cause T(a), its victims have the option of not responding to it in certain ways. Indeed, in the case of T(a), people may have control over whether they even become terrorized. (For example, if they stay calm and remember that the objective probability of harm is not much increased, then they may not be terrorized.) An agent might reasonably prefer not to be responsible for causing harm and prefer to do what shifts responsibility away from himself for any harm that does occur. He may choose between nonserious harming and terrorizing on these grounds. Hence the prospect of achieving one's end with less responsibility for harm might attract an agent to NST that involves him only terrorizing.

As noted above, the way in which others respond to terror could make things worse for them than if they were killed by others. I noted that victims might cause more deaths in their own community in the long run in resisting the terror in a noncowardly (but perhaps ill-considered) fashion. Things could be worse for them as well if they gave in to the terror. This is so if acting shamefully or losing one's dignity is worse than even dying at the hands of others. However, as I argue in Section III.E, it is hardly the responsibility of an agent, when he decides whether to harm an opponent or cause him nonharmful terror, to choose on the basis of sparing others from their own bad decision-making. The victims who behave badly may ultimately prefer to have been harmed instead, thus avoiding responsibility for bad conduct. But this will not be an additional factor that makes it wrong for an agent to use harmless terror instead of harm.

However, what if an agent is morally responsible for either the truly unavoidable or the perfectly reasonable terrified responses by others to her harmful act (such as bombing trees)? For example, suppose that Iraq had provided strong evidence of its having WMD (stronger than what was actually pointed to in launching the war against it), and it would have been completely morally appropriate to respond to this evidence with fear and war, even though war would harm many. Then there would be less difference between an agent causing harm herself and having others cause harm in responding to her doing what merely terrorizes them.

We have been comparing causing terror and harm on the assumption that we are dealing with cases where it is wrong to do either one. But is it true that it is always impermissible in a conflict to use terror on NCs (when one causes no harm) as a means that is necessary, given one's act, to produce an outcome. In a conflict-free context, it is most often wrong to terrorize people, even by such means as telling them a truth one knows will create terror. But in a situation where one faces an opponent in a conflict, this may not be true.

Consider briefly some ways in which we might terrorize NCs in a conflict, intending to do so, in manner T(a): (i) We tell them the truth, for example, that their children will die as soldiers, their supplies are low, and we have superior arms. Causing terror in this way is permissible. (ii) In a declared war, we actually kill combatants who are their children, and they know that other children will then be called up as combatants. Causing terror in this way is permissible. (iii) We do or threaten to do other things it is permissible to do but that we know will reasonably lead to terror. For example, in the Oil Threat case, someone threatens not to sell oil. Causing terror in this way seems to be permissible in a conflict. (iv) Now consider releasing a flock of butterflies over an opponent's population when we know they are very (irrationally) frightened of butterflies. This involves taking advantage of their irrationality and so is demeaning in a way that reasonable and truth-produced terror is not. Trying to get people to behave in a cowardly way is also preying on their imperfections. If the victims act shamefully, then they may wish they had been harmed instead of terrorized in this way. But this does not mean that using tactics in a conflict that depend on opponent NCs' defects of rationality and character makes the otherwise permissible act of sending butterflies impermissible when producing reasonable, truth-based fears would not be wrong. In cases (i)–(iv), terror per se is not only a less serious imposition than significant harm, it is also permissible.

Is it permissible to threaten to do what it is impermissible actually to do as a way of terrorizing? Some have argued that it would be permissible to threaten to use nuclear weapons on NCs if this would deter nuclear war, even if one never intended to use the weapons because using them on NCs would be impermissible. But the nuclear deterrent was supposed to benefit even the NCs who were threatened, on the assumption that they would be harmed as a side effect of their own side's releasing the nuclear weapons or of our responding even by attacking only military sites. A comparable condition would not be present in the cases of NST that I am considering now. So perhaps producing terror by threatening to do what it is impermissible to do is wrong if it will not also benefit those terrorized.

Is it permissible to do what one knows will give rise to an expectation that one will do what it is wrong to do, just because it will give rise to such an expectation, when this does not actually involve threatening to do what it is wrong to do? This is what happens in the Trees case (described above) when we seek to establish a false belief that we will bomb NCs as a means to producing our goal. It seems to me that this sort of terror might sometimes be permissible. It is also present in the Terror in Other Defense case: a military division is under attack. It cannot beat back the opponent. Someone not under attack realizes that if he kills a combatant at the back of the attacking force, close to where villagers live, then the villagers will think they are going to be attacked. This will lead the villagers to interfere with the attacking force. Creating the terror in NCs in this way seems permissible in a conflict.

On the basis of these cases of permissible and impermissible T(a) we can conclude the following. Terror (and the inappropriate weakening of one's resolve) is certainly something people would like to avoid, and it can make an act prima facie wrong. But people do not always have a right not to have someone bring these states about during a time of conflict. They may have a right that certain means, such as killing NCs, not be used (and even not be threatened to be used) to bring about these states, even in a situation of conflict. But this is because the people killed would be improperly treated if their involvement is, given someone's act, a necessary means to producing the terror. (Could we also say that it is wrong to kill NC A not only because he has a right not to be killed but also because B has a right that A not be killed as a means of causing B terror? It seems odd to say this. Suppose that A is a condemned criminal who has no right not to be killed by the state executioner, but the killing of A will be carried out only in order to cause terror in ordinary citizen B. I do not think the fact that the killing will cause terror in B [and be done to cause terror in B] makes killing A impermissible. But perhaps it is correct to say that when T[a] comes about by means that are wrong quite independently of being a wrong way to bring about B's terror [e.g., it is a wrongful killing of NCs as an end or means], then B can complain not only about the terror caused but about the fact that an act that should not have been done leads to terror. And the fact that an act that should not have been done also leads to terror can become an additional ground for condemning the wrong act.)

Now consider whether it is always impermissible in a conflict to produce T(b) of a temporary sort as a means that, given one's act, is necessary to produce military or political ends. It may be that even though creating panic is “cleaner” than T(a) that may involve corrupting NCs, it is far more rarely permissible. This is an implication of our earlier argument in which we compared using torture without harm that reduces people to a nonrational state with doing what elicits incorrect judgments and moral weakness. Even dropping leaflets with truths on them that will cause a mindless panic or sending butterflies that will cause a mindless panic in NCs seems impermissible in many cases where these means to causing T(a) would be permissible. Still, if the only way to stop an attack on our side is for a bystander to do something harmless that creates a harmless terror stampede by an opponent's NCs, then I think that this may be permissible.

I conclude that what happens in ST to those who are killed is usually worse and a more serious wrong than what happens to those who are just terrorized.

(b) Challenges. I have downplayed the badness and moral wrongness of causing terror per se (when no physical harm is done) relative to causing serious harm such as death. Consider some challenges to this view and my responses.

(i) It may be said that it is wrong just to compare terror with death in the same number of people, for terror typically affects many people, but killing via ST affects only a few.

A response to (i) recognizes that terrorizing more people is worse than terrorizing a few, and it is a more serious wrong than terrorizing a few when it is wrong. But still it is not as serious a wrong as seriously harming fewer people. This is because it is a mistake, in two ways, to aggregate much smaller impositions on each of many people in order to outweigh much graver losses to a few (assuming that their baseline condition was the same). First, it is not as bad an outcome if many people suffer T(a) as when a few die. Second, if we have a choice of killing a few people or instead producing T(a) in many, then it would be wrong to do the former.

(ii) It may be said that it is wrong to focus on terror per se rather than on its wider effects independent of weakening commitment to policies. Terror undermines civil society in a way that even thousands of deaths do not. For example, people are afraid to lead lives in public and congregate with others, at least when the harm that people fear is random and does not occur at predictable times (unlike the nighttime bombing of London during World War II).

A response to (ii) recognizes that undermining civil society can make an act prima facie wrong (though not all means to this effect ultimately may be wrong, as discussion in Section III.F.1(a) suggests). However, it is only if the absence of civil society and public life leads to death and grave losses to individuals (e.g., serious medical conditions are left untreated because people are afraid to go to hospitals) that the loss to individuals will be as or nearly as great as it is to those killed in ST. And only this is morally relevant, if we are to avoid the mistake of aggregating much smaller losses to many individuals so as to outweigh grave ones to fewer individuals (as discussed in the response to [i]).

So the response to (ii) implies that an outcome with a few deaths is worse than one where people are T(a) (but not because of any harm to anyone) and so there is no civil society providing non-life-preserving benefits to people. It also implies that if a terrorist agent has a choice, it would be wrong to produce the few deaths rather than the T(a) and collapse of non–life-preserving aspects of civil society through nonharmful means. (Henceforth I will assume that “civil society” involves only non-life-preserving benefits.)

(iii) Now consider a general objection to the view (embodied in the response to [ii]) that death to a few is worse than many being affected by both T(a) and the absence of civil society but with no deaths. We each often take a small risk of death in order to participate in civil society rather than certainly eliminate our civic involvement. (This is an intrapersonal calculation we make independent of numbers of people.) For example, we increase our risk of dying by driving to a café rather than staying home. Should we not then think that the small risk of dying in a harmful ST attack that terrorists expose everyone to is less bad than an outcome in which terror per se, without harm, causes people to cease participating in civil society? (In the cases where everyone runs a risk of dying by driving in order to participate in civil society, we are imagining that those who might benefit from civil society are also the ones running the risk of death. This is a simplifying assumption. Let us suppose that it is true in actual terrorist scenarios, where harm leads to terror, that those who would benefit from civil society are also the ones who run the risk of death, because it is in public venues that agents of ST attack.)

The risk to each is small if terrorists can only kill a few of many people at random. An important difference between the one-person and the many-person cases is that only in the latter can we be sure that someone will be killed. Hence if many people avoid public activities, then each certainly loses civic life and there is also no civic life for each to participate in, but death(s) will be avoided that would otherwise have occured. By contrast, in the single-person case, the person might never have been killed if his retreat from civic life had not occurred.

The possibility of no deaths occurring can sometimes also be a characteristic of many-person cases. Then the fact that people are each willing to take a small risk of a very bad outcome (such as death) rather than give up some good does not show that the occurrence of the bad outcome is less bad than the absence of the good per se. For example, each may be willing to risk a devastating nuclear explosion in order to have cheap energy that increases the standard of living. This is consistent with the nuclear devastation being far worse an outcome than is living with a somewhat lower standard of living. What is needed to make taking the risk a rational choice is the belief that the devastation need not certainly occur—that we really may all escape it.

But as noted, in some cases the bad state of affairs will definitely occur to someone, though each runs only a small risk of it happening to him. Such a case is one where many drive in order to participate in civil society. It is these cases that support objection (iii), for they support the conclusion that people think that the state of affairs in which something very bad happens to a few (such as dying in car accidents or terrorist attacks), and not just the state of affairs of running a risk of dying, is better than a state of affairs in which the deaths do not occur but many people are afraid and go without the good of civil society.

One response to this objection questions whether we can decide which outcome is worse just by considering with which one people would choose to live. If each person decides to live with an outcome in which a few die, on the basis of the low risk that each person has of being amongst the dead, this may just mean that each one does not care about how bad the outcome will be but only about the risk each faces. And this might be a morally permissible basis for decisions. But it still could be that the outcome in which people die is worse, and that it is morally appropriate for an agent who has a choice whether to bring about one outcome or another to care about which is worse.

Let us suppose, however, that willingness to live with an outcome tells us about its relative badness. A second response to objection (iii) raises what we can call the Paradox of Risk. That is, suppose the outcome is worse for us when we are afraid and we lose our participation in civic society than when each has a small risk of death and some eventually die. Then why would the terrorist, in increasing the risk of death to each a small bit, do anything that produced terror that stopped people from engaging in civil society? That is, why would he succeed in bringing about (what [iii] claims is) the worse state of affairs in addition to the few deaths he causes?

Perhaps it is because a more serious moral wrong is done when one is deliberately killed in terrorism than when one is killed in a car accident? But if it is fear that causes people to flee civil society, the fear should presumably be in proportion to the probability of being killed rather than to the moral seriousness of the death. In order to confirm this conclusion, imagine the following two imaginary variants of driving cases:

In these cases, the risks that drivers face and the deaths that actually occur are due to deliberate villainy. Would we think that this was a worse state of affairs than when people die in accidents? I believe we would. Should we stop driving and give up participating in civil society? I do not think so. But it is worth pointing out two factors that complicate giving this response. First, D1 and D2 may differ slightly in a morally significant way. In D1, the villains have done their dirty work already, and this will lead to collisions. Hence it may be that we have no concern that our driving and some being killed will encourage villains to engage in further tampering to achieve their aims. By contrast, in D2 the villains are still active (which is more like the situation with ongoing terrorism). Hence going out on the road and giving them successes may encourage them to continue to create collisions; our actions might fuel the creation of new risks of death and additional deaths. This effect on the villains may add another factor that makes the deaths in D2 worse from our point of view and may encourage one not to participate in civil society. However, a second factor may move one in the opposite direction, for not participating in civil society will give the villains power to alter one's behavior in another significant way. This additional factor could make giving up on civil society a worse outcome than it would otherwise be per se.

If we would be willing to have villain-caused deaths happen rather than give up civil society when abstracting from these two factors, this would suggest that we think it is a worse state of affairs not to have civil society for many than for each to run a risk of death and have a few villainous deaths.

Suppose terror and loss of civil society are worse than a few villain-caused deaths. Then it is only if people want not to encourage villains with successful killings that there would be a reason why what the terrorists do in ST would make people avoid civil society. Alternatively, their avoiding civil society may (again) raise the question of whether people are responding irrationally when they fear death from ST but not from driving on the road (even in D1); perhaps they see the risk from ST as greater than it is. Perhaps they see the risk as so great that if they had to undergo the risk in order to participate in civil society, then they would reasonably forgo civil society. But this would not mean that the actual uninflated number of deaths involve a worse state relative to the loss of civil society.

(iv) The objection we have been considering to the claim that a few deaths are worse than fear that leads to the absence of civil society can be rephrased so that the objection avoids the Paradox of Risk. Assume that in reality there is a small risk of death to each of many (in driving or from terrorist attacks). The objection (rephrased) claims that the outcome in which some people die as the fulfillment of this risk is not as bad an outcome as the one in which deaths do not occur but people are afraid and avoid civil society. However, many people would give up civil society and stay home through terror because they overestimate the probability of death. They would run the risk and tolerate some of them being killed rather than give up civil society if they did not overestimate the probability of death. Now we must deal with this objection.

In all of the driving cases considered so far, the deaths occur to people who are trying to participate in civil society. Many deaths also occur to people while or as a consequence of participating in civil society. ST typically kills people on the occasion of their participation in civil society, which is why avoiding civil society is a response to the fear of death, as it eliminates the occasion for the deaths. We might describe the causal route as follows: participating in civil society (or means to it) leads to deaths, while not participating in civil society (or means to it) eliminates these deaths.

By contrast, we could imagine other cases in which it is also true that only those who try to or do participate in civil society are put at risk, but without it being true that they are at risk on the occasion of trying to or participating in civil society. Rather, undergoing some risk that will eventuate in death for some is just a condition of the availability of civil society. So we might describe this causal route as follows: accepting risk of death and some actual deaths allows for participating in civil society, while not accepting risk of death and some deaths disallows participation in civil society. Consider the following two cases as illustrations:

It might be claimed that in these cases, too, we would each prefer a world in which we faced the risk of death and in which a few deaths occurred rather than have the villains cause all of us, by nonharmful means, to be afraid and so to give up civil society. This judgment is given as support for the claim that though the outcome in which the deaths come about by villainy is morally worse than when they do not, it is not as bad as no deaths but instead fear leading to the absence of civil society.

(v) The issue we have been dealing with in (iii) and (iv) is whether lesser losses to each of many people (such as fear leading to the loss of civil society) are a worse state of affairs than a few deaths of that group of people. Suppose, for argument's sake, that the answer is yes. (This would imply that we cannot rank the badness of states of affairs just by looking to see how badly off the worst-off people in them would be.) What would this imply about the wrongness of producing each state of affairs? Would it imply that if a terrorist had a choice between pressing one button that would cause a few deaths or another button that produced no deaths but caused T(a) leading to the collapse of (non-life-preserving elements of) civil society, then he would commit a more serious moral wrong if he pressed the second button rather than the first? I do not think so. For it may be a more serious wrong to do something that would result in a better state of affairs rather than something that results in a worse state of affairs.

This is a common view amongst nonconsequentialists. Typical examples involve the impermissibility of harming a few people in order to save many others. In these cases, however, if one does not harm some, one will not have harmed others but only not saved them from some harm. If we were given a choice between killing one person and killing five, other things being equal, even a nonconsequentialist would say that it is a less serious wrong to kill one. Suppose that this is because we cause (rather than allow) the harm in both cases, and greater harm is a worse state of affairs than less harm. Then if the destruction of civil society through T(a) is a worse state of affairs than a few people being killed, why does not a terrorist do a less serious wrong if he kills a few rather than causing T(a) and destroying civil society in a harmless way?

It might be said that when the terrorist kills, he is completely responsible for the deaths, but in pressing the other button he shares responsibility with his victim for the collapse of civil society. For this reason he commits a less serious wrong in doing the latter. That is, he only brings about a state in which people make bad judgments based on fear that may itself be inappropriate.

I think that we should put aside this issue of responsibility in trying to defend the view that it is a more serious wrong to kill the few. The issue we should focus on instead is the difference it makes to what we may do when the worse state of affairs has in it at least as many individuals as badly off as others will be in less bad states of affairs—for example, at least one person will be dead whether we kill five or kill one—by contrast to worse states of affairs in which this is not true. I have emphasized this point elsewhere³⁷

Why may we not kill rather than produce (what we are assuming is) the worse state of affairs in which each person will suffer a lesser loss? When it is permissible to do something to people, this reflects morality's endorsement of a certain view about what status persons have as individuals, one by one. (When we let things happen to people, morality is not thereby endorsing what is happening to them.) It is a worse insult to the high status of a person to kill him than to paralyze him or cause him terror, for it endorses the view that it is permissible to violate a person in the more serious way. Violating a few strong rights involves a greater wrong to each than violating many weaker rights, because doing the latter need not imply that it is ever permissible to do something as bad as to kill someone. Hence the act that produces the worse state of affairs can be more respectful of persons, even if it insults more people and would make for a worse state of affairs.

I conclude two things here: (1) the wrong done to each person is greater when the few are killed than when each of many caused T(a) and is deprived of civil society; and (2) the wrong done when many are affected should be ranked by the worst one does to any given person, and only after that by the numbers affected.³⁸

2. Harm

We have already said that the harm involved in ST is severe since it is often death. This is the much more serious part of ST than the terror per se, I believe, at least because it involves the greater wrong against a person. One of the oddities of the term terrorism is that a nondistinctive part of ST (i.e., a part that occurs in nonterror killing) has a greater role in making terrorism be wrong than the distinctive part of it that gives it its title. Even if ST were unsuccessful in causing any deaths, increasing the probability of death for insufficient reason can also make the terrorist agent's act seriously wrong. And even if the deaths from ST are less frequent and less numerous than those from car accidents, greater resentment against (if not fear of) wrong acts, and even bad intentions, rather than accidental causes is not irrational. It is consistent with being less afraid of being attacked by a criminal than of getting a deadly disease to insist that wrong acts be extirpated first. So it is not necessarily irrational to spend more money on stopping ST than on stopping accidents.

Of course, generating terror alone, as an end in itself, as a necessary productive means, or even as a side effect, can sometimes be rights-violating conduct that one can appropriately resist and resent. Nothing I have said in comparing terror per se with severe harm per se implies that causing the terror cannot add a wrong-making feature to an act and contribute to the prima facie wrongness of ST or that it is not often impermissible to do what causes people terror independent of any harm. But if the terrorized response is not entirely rational and is also subject to inner control by persons, this bad effect—unlike grave harm—could perhaps be avoided in some less costly way than stamping out or even defending against the people who try to produce it.

1. Terror-from-Harm

Is terror a worse or a greater contribution to prima facie wrongness when it is caused by actual harm that has already occurred than when it is caused by bombing trees, as in the Trees case? Above, in Section III.F.1(a), I said that when impermissible harm is the cause of terror, those terrorized may complain, as they should not have been terrorized given the way the terror was brought about. In addition, along with the misery and trauma of loss, sympathy for those actually harmed accompanies terror when terror comes from known grave harm that has already occurred. The terror may even interfere with proper mourning for those already lost. This makes the event (if not the terror itself) worse than terror in the Trees or Rumor cases and can contribute to the wrongness of ST. (Of course, in the Trees case, people might not just fear future harm; they might incorrectly believe that others had been harmed. The misery and trauma relating to others would then also occur, even if they are not appropriate. But there could be no interference with proper mourning.) But are these factors plus terror worse than death? No. To test this claim, imagine that we could either stop people from having the terror, misery, and trauma from known deaths of group A by psychological counseling or stop other people of group B from being killed. We should do the latter.

2. Harm-for-Terror

Is causing death worse and/or a greater contribution to wrongness of an act because it is used to create terror rather than for other purposes? For example, is it a greater moral wrong if one is killed as a means to start terror rather than as human tinder in the Human Tinder case? What if the terror is created to be a mere mechanical response that will interfere with a military operation, as in the Stampede case, by contrast with being created in order to change the judgment and will of people in politics? It seems morally worse to have wrongful harm done to one for an effect that is intrinsically worse, and this could bear on the seriousness of the wrong. If creating terror in NCs is a greater wrong than setting aflame a military facility, then it will be morally worse to die for this purpose. If will-undermining terror for political purposes is uglier for the victims, but producing mechanical terror in them is a greater wrong, then it will be morally worse to die in order to produce mechanical terror. (This does not mean that it makes sense to fear the morally worse deaths more.)

In this interaction, it is still the victim of harm who is the locus of our concern with the badness of ST, rather than the victim of terror per se. Here, a distinctive part of terror killing (that is, the terror) makes the harm that is necessary, given one's act, to produce the terror morally worse and (sometimes) a greater wrong, though the nondistinctive part of terror killing (that is, the killing) is still the part that is morally most significant and most to be feared (in virtue of death).

3. Terror-for-Goal

Some may think that it is morally bad that people's appropriate, terrified response to harm to others is used for the terrorists' further aims. This focuses on a further wrong done to the people who are responding to the harm. It introduces the further distinction between using terror for a goal and using harm for terror. However, I think that the concern about misusing the person terrorized should pale before concern for the person gravely harmed, especially if he is harmed to bring about terror. It is consideration of the Trees and Rumor cases that lead me to this conclusion. For in these cases there is no harm created, just people's terror as a necessary means, given one's act, to some goal, and a lesser moral wrong occurs. On the other hand, consideration of the Human Tinder case and the other mechanical versus nonmechanical terror cases lead me to the conclusion that not all uses of harm or harm plus terror are morally equal, even when they are all wrong.

1. First Variant of the Park and Military Facility Case

A small subsection of the very same people who would have been killed if the military facility had been hit would die if the park is hit instead for terrorist purposes. I suggest that if it were morally wrong to bomb the military facility, causing deaths and terror, and also wrong to bomb the park, considered as individual acts, it would be better if the wrong act of bombing the park is done in place of the wrong act of bombing the military facility. This can be true though it involves using the people in the park as necessary means, given one's act, and though we assume that this act requires a type of causal role for people that can sometimes make an act impermissible when other types of acts causing the same or somewhat greater side-effect deaths would be permissible.

This conclusion implies that the fact that an agent kills people as a mere means to produce his end of winning a conflict should not always be taken as evidence of his having worse intentions for NCs than another agent. For in this scenario, given the alternatives, it could be the person who would choose to cause more deaths that have no causal usefulness who would exhibit the worst intentions toward NCs. I believe this is so because his killing fewer of the very same people who would die anyway from his other act is the morally preferable alternative, and an agent who knew this would only choose the act that kills more people because he intends their deaths as an end in itself.

(a). Suppose that threshold deontology implies that harming people as a necessary means, given one's act, to producing an end requires more to justify it than harming people as side effects, but also that this justification can sometimes be given. Typically, discussions involving threshold deontological reasoning result in a justification that makes an act permissible. But we can extend threshold deontological reasoning to apply to cases where the act that should be done instead of some other act is still impermissible. We can do this if we imagine that the Park and Military Facility cases involve Al Qaeda attacking a (non-Nazi) United States. Al Qaeda has no adequate goal morally to justify causing great collateral damage by bombing the military facility. Its bombing the park instead of alternatively bombing the military facility—which is the correct choice of the two—would be a true case of doing the lesser evil, as the act would still be an evil and would not be a permissible act at all. We can call it a substitute wrong, by which I mean an act that it would have been wrong to perform if it were the only act possible to achieve goals yet that should be substituted for another wrong act (bombing the military facility), all the while retaining its status as a wrong act.⁴¹

(b). Now imagine a case in which it would have been permissible to bomb the military facility despite collateral death and terror. For example, imagine that the United States is controlled by Nazis, and the Resistance would do this action. If a much smaller subsection of the very same people who would die if the Resistance bombed the military facility would die if they bombed the park instead, then I believe that bombing the park would no longer be a substitute wrong. It would be permissible and not merely a less seriously wrong act. This is so even if bombing the park would have been wrong (in virtue of how the victims were used as means to produce a goal) if it were the only act the Resistance could have performed. As a substitute for a permissible act that would have caused the deaths of the same people who will die in the park plus many more, what would be impermissible on its own becomes permissible as a substitute. It is secondarily permissible. (This is an instance of what I call the principle of secondary permissibility.)⁴³

When people speak about doing the lesser evil, they sometimes speak as though they have in mind a choice between two acts that remain wrong whichever is done, but this model of a substitute wrong is only present in our case involving Al Qaeda. By contrast, when we do an act (that would otherwise be wrong) because it involves less harm instead of a more harmful permissible act, the second act can be not a lesser moral evil but permissible. This reminds us that the act that is eliminated was originally permissible even though it would cause more harm than an originally impermissible act, and is worse in that sense (if not a greater moral evil). It is because of this that in some cases, what was originally a greater moral evil (bombing the park) can become permissible secondarily because it involves less harm overall and no more to the same people who otherwise would have been harmed.

(c). In the cases in 1 (a) and (b), we moved away from the typical case of ST, which involves killing people who would not soon die anyway. But we can reintroduce this factor. For suppose that everything is as in 1 (b), except that those who can permissibly bomb the military facility would not actually do so because they cannot bring themselves to harm so many innocent NCs as a side effect. Then they find out that they can achieve their aims by directly killing a small subset of those NCs that they still can (are able to), but will not, permissibly kill as side effects. These are people who would not die shortly anyway, but they would be alive only because the Resistance refrains from doing what it still can permissibly do. Due to this, I think it becomes permissible for the Resistance to do to the smaller subset whatever I argued in 1 (b) it could permissibly do if it would alternatively have actually permissibly killed the same people plus others as a side effect.⁴⁴

2. Second Variant of the Park and Military Facility Case

Entirely different people in the park will be killed and terrorized as means from those who would have been affected as side effects by bombing the military facility.

(a). Suppose bombing the military facility would have been impermissible, and bombing the park would also have been impermissible were it the only thing that could be done, but many more people would die in the former operation than in the latter. Then this might still make bombing the park the lesser wrong (despite its use of the morally less favored mere means role for people in producing a goal). This is a further instance of the idea of a substitute wrong, but here it is just the great difference in numbers dead that might bring us above a threshold in selecting between wrong acts (not in making an act permissible), not that all the dead would have been killed anyway.

(b). What if bombing the military facility would have been permissible (in the Nazi United States case) considered on its own, but bombing the park would have been impermissible even if it alone could have been done? When entirely different people will be affected in the park from those in the military facility bombing, and they will be in the disfavored necessary means role, it is quite likely that it will remain wrong to bomb the park rather than the military facility. For we cannot, in general, do equal or lesser harms to fewer people in an impermissible way in order to avoid doing equal or greater harm to a greater number of entirely different people in a permissible way. (This assumes that the additional seven thousand dead is not enough of an evil that preventing it moves us above the threshold for not killing three thousand people as mere means to producing a goal.)

Hence, in addition to what was said in Section III.C(2) above, we cannot say whether terrorist acts should not be done, at least until we know the number who will be similarly affected if another act is done, who those affected are, and whether the alternative act was permissible or impermissible. This is in addition to having to know whether a threshold on abiding by constraints on treating people in certain ways is reached, as transgressing the constraints is the only way to avoid some great evil that there is no other way to avoid. This examination of terrorism reveals that no easy answers are to be had on the topic of the moral permissibility and impermissibility of terrorist acts.