I. Special Immunity or Moral Parity?
Andrew Altman and Christopher Wellman say, “Surely, it would have been permissible for somebody to assassinate Stalin in the 1930s.” Footnote 1 If so, is it also permissible to kill a president, Member of Parliament, bureaucrat, or police officer from a democratic regime, if killing is necessary to stop them from harming the innocent? If it is permissible to assassinate Hitler to stop him from invading Poland, is it also permissible to assassinate a U.S. president in order to stop him from invading the Philippines or from ordering the genocidal slaughter or forced relocation of Native Americans? If it would be permissible to kill a Gestapo agent to stop him from murdering innocent people, is it also be permissible to kill a democratic police officer who uses excessive violence?
Philosophers and laypeople often assume not. They assume that in liberal democracies, only nonviolent resistance to state injustice is permissible. Footnote 2 They assume we must defer to democratic government agents, even when these agents act in deeply unjust, harmful, and destructive ways.
This assumption is puzzling. The prevailing view is that, when it comes to government agents, the practice of killing in self-defense or defense of others is governed by different moral principles from those that govern defensive killing in other contexts. This presupposes that it makes a difference to the permissibility of killing an aggressor in self-defense or defense of others that the aggressor is wearing a uniform, or holds an office, or was appointed by someone who was in turn elected by my neighbors. According to the prevailing view, my neighbors can eliminate my right of self-defense or defense of others by granting someone an office. Footnote 3
To illustrate this asymmetry, consider the following three cases:
A. Shooter in the Park
A masked man emerges from a black van holding a rifle. He starts shooting at children in a public park. Ann, a bystander, has a gun. She kills him before he kills any innocent children.
B. Health Nut
Health guru John sincerely believes caffeine is unhealthy, that it causes laziness, and that it induces people to use hard drugs. John announces that he and his followers will capture coffee drinkers, confiscate their belongings, and imprison them in John’s filthy basement for years. Ann, who is too poor to move away from town, loves coffee. She secretly drinks it in the morning in her kitchen. One day, a henchman breaks into her house and attempts to capture her. She struggles to defend herself, and, in the process, kills him.
C. Terrorist
Cobra Commander, leader of the terrorist organization COBRA, has a device that allows him to launch the United States’ nuclear arsenal against Russia. Ann, a private civilian, somehow stumbles upon COBRA’s secret control room. Just before Cobra Commander enters the launch code, Ann shoots him.
I expect most people believe that it is permissible for Ann to kill the wrongdoers in A–C. Probably only radical pacifists would deny that killing is wrong in A–C: Footnote 4
But now consider three new cases (D–F) that seem analogous to the first three (A–C). In these new cases, the only obvious major difference is that the wrongdoer is a democratic government agent:
D. Minivan Shooter
Ann witnesses a police officer stop a minivan with a female driver and three children in the back. Ann sees that the woman is unarmed. The police officer emerges from his car and immediately starts shooting at the van’s windows. Ann has a gun. She shoots the police officer before he kills any of the children. Footnote 5
E. War on Drugs
Town leaders decide to make marijuana illegal, even though there is overwhelming evidence that marijuana is in every respect less harmful than alcohol, a drug that is legal for any adult to consume. Footnote 6 Ann has a pot stash in her house. One night, a bunch of police officers raid Ann’s house in a no-knock raid. She recognizes that they are police officers. She also knows that if they capture her, she will be imprisoned for a decade. Her government issues overly punitive sentences for drug possession and is unresponsive to citizens’ demands to overturn the law. Ann struggles to escape, and, in the process, kills the cops. Footnote 7
F. Hawk
Ann, a janitor, happens to be cleaning the Situation Room when the president and his staff enter and lock the door. She hears the president inform the Joint Chiefs and his cabinet that he intends to unload the United States’ nuclear arsenal on Russia. The head of NORAD, who is on the screen, has already entered his launch code. Just as the president is about to enter in his own launch code, Ann tries to restrain him. The president breaks free and is about to enter the code. Ann takes a gun from a secret service agent and shoots the president.
Many people judge these cases differently from A–C. They think killing in self-defense or in the defense others is wrongful in (at least some of) D–F, though it was permissible in A–C.
Thus, most people seem to subscribe to what I will call the Special Immunity Thesis:
Democratic government agents enjoy a special immunity against being killed in self-defense or defense of others. The set of conditions under which it is permissible to kill a democratic government agent, acting ex officio, is much more tightly constrained than the set of conditions under which it is permissible to kill a civilian.
The Special Immunity Thesis holds that killing the agents of democratic governments, at least when they are acting ex officio, faces a special justificatory burden.
In contrast, one might reject the Special Immunity Thesis in favor of the Moral Parity Thesis:
If there are conditions under which a civilian may, in self-defense or defense of others, kill a fellow civilian, there are also conditions under which a civilian may kill a democratic government agent, even an agent acting ex officio.
The Moral Parity Thesis holds that killing the agents of democratic governments is at least on par morally with killing private civilians. Strictly speaking, I leave it open that it might be easier to justify killing government agents than civilians. The Moral Parity Thesis implies that if killing is permissible in any of the cases A–C, it is permissible in the analogous cases D–F.
This paper defends the Moral Parity Thesis and attacks the Special Immunity Thesis. My thesis is that the conditions under which you may kill a civilian, in self-defense or in defense of others, are also conditions under which you may kill the agent of a democratic government, even when that agent acts ex officio.
Note that I focus solely on the ethics of defensive killing against immediate threats from democratic government agents. I am not here discussing punishment, whether anyone might ever deserve to die, whether killing anyone might ever be justified as an end in itself, or whether violence is useful to overturn laws or produce social change. Many philosophers and activists believe that nonviolent civil disobedience is both morally superior to and more effective than violent resistance in changing unjust laws. Footnote 8 They might be correct, but that is not my concern here. Footnote 9
II. A Commonsense Moral Theory of Defensive Killing
My thesis in this essay is conditional: Whatever conditions render it permissible to kill a fellow civilian in self-defense or defense of others also render it permissible to kill a democratic government agent, even an agent acting ex officio. Officially, I need take no real stance on what those conditions are. However, in this section, I’ll offer a brief account of defensive killing, though the precise details go beyond the scope of this essay.
Here is a sketch of a theory of defensive killing, taken from Jeff McMahan’s Killing in War. Footnote 10 By default, killing is presumed wrong. However, a person can become liable to be killed by performing certain wrongful or unjust actions. A person is liable to be killed when he is doing something deeply wrong, unjust, or harmful to others, and when killing him would serve a defensive purpose, such as self-defense, the defense of others, or to prevent him from causing greater injustice. Killing is also restricted by a doctrine of necessity: at minimum, when a nonlethal alternative is equally effective at stopping someone from committing injustice, it is not permissible to kill him. Most people accept this broad outline, though they dispute the details of the theory.
One way to discover what commonsense moral thinking says about the ethics of killing is to examine English common law. As John Hasnas says, “The doctrines of self-defense and defense of others are doctrines that developed through the common law process that embody centuries of experience regarding how best to discourage violence and resolve violent disputes.” They “ . . . represent what fifty generations of juries and judges believed to be a fair and proper response to [wrongful] attack.” Footnote 11 The common law is a reliable guide to people’s moral intuitions about permissible killing.
The common law assumes that people have a right to protect themselves and others against “unlawful” threats such as assault, battery, rape, and murder. Footnote 12 According to the common law doctrine of self-defense, one person (the “killer”) may justifiably kill another (the “adversary”) when:
-
1. The killer is not the aggressor, and
-
2. He reasonably believes he is in imminent danger of severe bodily harm from his adversary, and
-
3. He reasonably believes that killing is necessary to avoid this danger. Footnote 13
Note that the common law regards meeting these conditions as justifications, not merely excuses, for homicide. Footnote 14
A. Some comments on condition 2
In common law, killing in self-defense is justified only if the threat of harm is severe enough. I can kill you to stop you from raping or dismembering me, but not to stop you from throwing mud at me. There is no sharp line between what threats count as severe or not severe enough to warrant killing. Also the “imminent danger” proviso does not literally mean that a victim must wait until the last possible second to defend herself.
B. Some comments on condition 3
The common law requires the killer to have a reasonable belief that deadly force is necessary to protect herself. The test here is whether a reasonable person might hold the belief, not whether it is impossible for a reasonable person to doubt the belief. Finally “necessary” here means there are no good alternatives, not that there are no alternatives. So, for instance, in Shooter at the Park, Ann could attempt to wrestle the shooter or reason with him. But the probability of success is sufficiently low that she is allowed to kill him instead.
The common law doctrine of defense of others is almost identical to the doctrine of self-defense. According to one popular law textbook:
. . . one is justified in using reasonable force in defense of another person, even a stranger, when he reasonably believes that the other is in immediate danger of unlawful bodily harm from his adversary and that the use of such force is necessary to avoid this danger. Footnote 15
These conditions should be understood the same way as the conditions for self-defense.
III. Legitimacy and Authority
There appears to be an obvious justification for the Special Immunity Thesis: Democratic governments have a special moral status. Unlike lone shooters, criminal organizations, health nuts, or totalitarian dictators, many democratic governments (and, by extension, their agents acting ex officio) are both legitimate and authoritative. They have a right to rule and we have a duty to obey them. Therefore, while it is permissible to kill an evildoing terrorist, it is not permissible to kill an evildoing president, even when that president is doing the same thing as the terrorist.
A. The concepts of authority and legitimacy
To evaluate this purported justification of the Special Immunity Thesis, we must first clarify what the terms “legitimacy” and “authority” mean.
A government is legitimate just in case it is permissible for that government to stand, and to create, issue, and enforce rules using coercion. A government is authoritative (or “has authority”) over certain people just in case those people have a moral duty to obey that government’s laws, edicts, and commands. Legitimacy might make it permissible for the government to tax you. Authority might make it impermissible for you to refuse to pay. In short, “legitimacy” refers to the moral permission to coerce, while “authority” refers to a moral power that induces in others a duty to submit and obey. Footnote 16
B. Legitimacy and authority are logically independent properties
Legitimacy and authority are independent moral properties. Most theories of legitimacy and authority try to ground both properties on the same principles, such that governments have both or neither. But, at least as a matter of logic, a government (or a rule-making entity) could have one without the other. Having legitimacy does not suffice to have authority; having authority does not suffice to have legitimacy.
To see how a government-like entity could be authoritative but not legitimate, imagine a theory of “pacifist democracy.” This theory holds that we should all obey a central rule-making deliberative council. However, the theory holds that this council may never coerce people to comply with its rules. The pacifist theory thus holds that the council is authoritative, but not legitimate. Footnote 17 This pacifist theory may be false, but it is not incoherent.
Governments could also be legitimate but not authoritative. That is, a government might have moral permission to stand and to create laws, even if no citizens have the duty to obey or defer to that government. The government would have permission to force citizens to obey, but citizens would have no obligation to obey. In fact, as I’ll discuss in 3.4, there’s reason to think this view — that governments have legitimacy but not authority — is the dominant position among political philosophers who write about authority and legitimacy.
C. Legitimacy is irrelevant to this debate
If a government has legitimacy to do X, then it has moral permission to use violence (within appropriate limits under appropriate conditions) to enforce its ability to do X. If a government has legitimacy to issue rule X, then by definition it has moral permission to force you (within appropriate limits under appropriate conditions) to comply with X.
Yet, as we just discussed in subsection B, the fact that a government has legitimacy to do X does not imply citizens must let the government do X or obey the government when it does X. That a government legitimately does X tells us nothing by itself about what citizens may or may not do in response. Citizens might instead have no duty to obey. They may be free to resist or even to fight back violently. Footnote 18
If that seems puzzling, consider a boxing match. In a boxing match, both boxers have waived their right not to be punched. However, they both retain the permission right to block or evade the others’ punches. Thus, while they both have legitimacy to punch the other boxer, they do not have authority to do so.
So it might go in cases D–F. Even if the government agents have legitimacy to commit these injustices, this fact would not imply that Ann has any duty to refrain from killing the government agents. Even if we supposed the president in Hawk could legitimately nuke Russia, Ann might still be free to kill him in defense of others. We would need a further argument that it is not only legitimate for the president to nuke Russia, but also that he has authority over Ann such that she must let him.
Thus, I could agree that democratic governments may legitimately create and enforce laws that forbid private civilians from killing government agents who act ex officio. One can hold, without contradiction, both (i) that citizens may permissibly kill agents in self- or other-defense, and (ii) governments might legitimately forbid citizens from killing their agents in self- or other-defense. Indeed, it may even turn out, empirically, that legally prohibiting civilians from acting on their own judgment tends to produce the most overall justice. But, still, all of this is compatible with holding that civilians nevertheless have no moral duty to obey these legal prohibitions, and may instead permissibly kill democratic agents whenever they would similarly be justified in killing private civilians.
D. Authority probably does not exist
If, for some reason, governments and their agents had authority, this might explain why democratic government agents enjoy special immunity. By definition, authority is a moral power that induces in others a duty to defer and obey.
However, there is a serious problem with invoking democratic authority to defend the Special Immunity Thesis: There is strong reason to believe that no governments, democratic or otherwise, have authority. The doctrine of government authority has been subjected to sustained and overwhelming philosophical criticism over the past thirty years. As far as I can tell, following A. John Simmons’s seminal work on political obligation, the now dominant view among political philosophers is that while some governments can have legitimacy, none have authority. Footnote 19 (Or, more precisely, they might have authority over a tiny subset of their citizenry.) As Michael Huemer similarly concludes, after reviewing the literature, “Skepticism about political obligation [i.e., authority] is probably the dominant view” in philosophy now. Footnote 20 We now know that none of the traditional arguments (consent, tacit consent, hypothetical consent, fairness, associative obligations, and so on) for government authority come close to succeeding. Following Simmons, others have offered new arguments for authority, but these more recent arguments are highly problematic, and hardly anyone other than the authors of these arguments find them persuasive. Footnote 21
Of course, it’s possible that some governments do have authority, but my point here is that we do not seem to have any good grounds for thinking they do. Since philosophers have spent twenty-five hundred years trying, and failing, to prove that some governments have authority, we should probably assume they don't. At some point, the perpetual failure to prove that X, is evidence that not-X.
I cannot review the entire literature on political authority here. But I can shift the burden of proof. Someone who thinks political authority justifies the Special Immunity Thesis has the burden of producing a plausible theory of political authority.
One worry for me, though, is that if governments lack authority, then this threatens to trivialize my argument in this essay. If governments lack authority, one might think that of course the Special Immunity Thesis is false. One response would be to say that I am just exposing an unnoticed implication of the recent downfall of the doctrine of government authority: Not only do we lack any duty to obey the law, but, in addition, wrongdoing government agents are as fair game for defensive violence as wrongdoing civilians.
However, even if governments lack authority, this does not trivialize my thesis here. There are a number of other seemingly plausible justifications for the Special Immunity Thesis, justifications that do not rely on the assumption of government authority. It takes additional work to show that these other justifications fail. Also, as I explain in the next section, even if someone could show that democratic governments do enjoy a kind of general authority, it would take additional work to justify the Special Immunity Thesis.
E. General authority does not imply specific authority
Suppose, contrary to the state of the literature, that democratic governments do in fact have a sort of authority. Even if, heroically, this were established, it takes further work to defend the Special Immunity Thesis.
To say a government has authority means that there is at least one situation in which a person has a duty to do something because the government commands that person to do it. But a government can have authority over some issues without having absolute authority over everything. Indeed, every extant believer in democratic authority thinks that democratic governments have a limited scope of authority.
Recall that violence in self-defense and defense of others is warranted, on commonsense moral grounds, only to protect oneself or others from severe harm or injustice. To defend the Special Immunity Thesis on the basis of authority, one must show that democratic government agents specifically have authority to commit severe harms or injustices, the very harms and injustices that would render a civilian liable to defensive killing. For the Special Immunity Thesis to be true, it is not enough that governments have some authority, but that they have the specific authority to commit these severe injustices.
As defenders of government authority themselves accept, even if we grant that a democratic government is authoritative overall, it does not follow that every single command, action, edict, or policy of that government is authoritative. Each of the following could succeed or fail in being legitimate or authoritative:
-
1. Regime-types as a kind. (For example, can a theocracy ever be legitimate or have authority?)
-
2. A particular government’s overall right to stand. (For example, is the U.S. federal government legitimate or authoritative overall?)
-
3. Particular offices or branches within a government. (For example, are the Federal Reserve, the Drug Enforcement Agency, or Department of Homeland Security legitimate or authoritative?)
-
4. Particular government practices and procedures. (For example, are judicial review, jury trials, or the indefinite detention of so-called “enemy combatants” legitimate or authoritative?)
-
5. Particular laws, commands, and regulations. (For example, is marijuana criminalization in the United States legitimate or authoritative? Was the Comstock Law legitimate or authoritative?)
-
6. Particular actions and decisions. (Was the Dred Scott decision authoritative? Was it legitimate to wage the Spanish-American war? Did citizens have a duty to let the government fight the war? Was the Indian Removal Act of 1830 legitimate or authoritative?)
These distinctions make a difference. Authority in one does not imply authority in another. A lack of authority in one need not imply a lack of authority in another. Footnote 22
Anyone who wants to defend the Special Immunity Thesis on the basis of authority has a serious burden. It will not be enough to justify a general kind of government authority. Instead, one must produce a theory that justifies granting democratic officials authority to commit severe injustices, the kinds of injustices where we could justifiably kill civilians if the civilians were to try to commit them.
To illustrate this burden: Imagine Bill is the lawfully elected president. Bill demands sex from an intern. Even if Bill is the authoritative president of what is overall an authoritative regime, the intern has no duty to comply. Should he try to rape her, she may kill him in self-defense, subject to the conditions described in Section II.
Or, suppose a police officer, following the Fugitive Slave Act, arrests an escaped slave in Antebellum America. Suppose I shoot the police officer in order to free the slave. Even if we suppose that the U.S. Federal Government in the 1850s was legitimate and authoritative overall, it seems deeply implausible to hold that citizens had a duty to let it enforce slavery. Until I see a compelling argument for a theory that says otherwise, I would regard it as a reductio of any purported theory of authority that it implies I must let police officers enforce slavery.
Or, suppose the United States conducts a referendum on whether it will nuke the tiny island nation of Tuvalu. Suppose all eligible American voters vote. Suppose each of them, except me, votes to nuke Tuvalu. Suppose I know that there are no good grounds to nuke Tuvalu. (I can imagine good grounds to do so, but they are incredibly improbable.) In this case, it seems obvious that I have no duty to defer to my government leaders or my fellow citizens when they attempt to nuke Tuvalu. Now, perhaps someone will one day prove that, according to the Correct Theory of Government Authority, democratic governments specifically have authority to nuke other countries. But at least until I see the compelling argument, I would regard it as a reductio of any purported theory of authority that it implied this.
F. Estlund on the authority of unjust orders
Thus far, I’ve laid out a general challenge to anyone two tries to justify the Special Immunity Thesis on the basis of authority: One has to prove not that governments have some authority, but that they specifically have the authority to commit severe injustice, the kinds of injustice we would be allowed to kill civilians to prevent.
In this section, I’ll examine David Estlund’s attempt to meet this challenge. Estlund holds that we can have a duty to follow or defer to unjust orders if the orders were produced by the right people in the right way. Suppose that the United States conducts a referendum on whether to bomb Tuvalu using conventional weapons, but in this case, it makes a series of “honest mistakes.” Suppose Americans mistakenly but sincerely believe that bombing Tuvalu is justified according to correct theory of just war. Suppose that if their beliefs were correct, it would be justifiable to bomb Tuvalu. Suppose they decide to bomb Tuvalu only after employing a deliberative forum that perfectly fulfills Jürgen Habermas’s or Joshua Cohen’s ideal procedures for democratic deliberation. But now suppose I happen to know that their beliefs are mistaken and that war against Tuvalu is not justified. Suppose it’s my job to launch the bombs. Must I bomb Tuvalu, knowing full well that in doing so, I’ll kill innocent people who are not liable to be killed?
I think it is obvious that I should not do so. However, Estlund contends otherwise. He claims that when the following conditions are met, a person not only may, but usually must follow orders, even orders to commit unjust acts:
-
1. The act one is ordered to do is a token of a type of act that could in principle be justified.
-
2. The decision process used is one that is publicly justifiable to all reasonable people.
-
3. The order results from a reliable and fair decision process — a process that usually tracks the truth as well as any other process that meets condition 2.
-
4. Those who issue the order sincerely and in good faith believe that the order is justified. Footnote 23
Estlund takes these conditions to establish both the legitimacy of acting on unjust orders, and the authority of those orders. He thinks only democracies could meet these four conditions, though he does not argue that any real democracies in fact do meet them.
Even if Estlund were correct that such conditions could establish the authority of some unjust orders and actions, this does not yet justify the Special Immunity Thesis. After all, as Estlund seems to admit, the conditions under which a democracy would have authority are highly ideal, and it is not clear that any actual democracy meets his conditions. Footnote 24 In his book Democratic Authority, he never tries to show that any actual democracy has authority; he just tries to outline conditions under which a democracy could have authority. So, for Estlund to claim that some democratic government agents enjoy special immunity, he would have to show first that some existing democratic government is sufficiently just and fair to qualify as authoritative, and second that this government has actually issued an unjust order while meeting his four criteria above.
This issue aside, should we grant Estlund his theory of the authority of unjust orders? For the sake of argument, I will assume on Estlund’s behalf that he has succeeded in showing that some extant democracies do have a general kind of authority. Even if so, why believe that they could have the specific authority to commit severe injustice? Estlund has two strands of argument on behalf of this claim.
First, Estlund relies upon casuistical reasoning. He describes a few cases in which a person — Jason the Jailor — is ordered to commit relatively minor injustices. Estlund expects his readers to agree that Jason must obey the orders. He then tries to show that if so, his readers must also agree, on pain of inconsistency, that Jason could be obligated to follow orders to commit relatively severe injustices.
Estlund’s least controversial case, in his own eyes, goes as follows. Suppose that a fair and reliable jury in a fair and impartial legal system convicts a defendant, but that the jury or judge makes an honest moral mistake. Estlund writes,
Jason the Jailer realizes the defendant is guilty, but knows that the sentence of 20 years in jail is excessive. Suppose the crime is embezzling $1000. Jason sees that anything more than 5 years is morally indefensible. Jason is legally ordered to keep the prisoner for 20 years, but suppose he could easily let him escape after 5. Is Jason permitted to carry out the full punishment?
It seems clear to me that he is. Footnote 25
Estlund goes on to argue that Jason not only may carry out the orders (that is, that he has legitimacy to do so), but that, unless following the orders imposes some severe psychological burden on Jason, he must do so (that is, that the orders are authoritative). Note that in the case above, Estlund is not saying that Jason merely judges that the punishment is too severe. Rather, he wants us to understand that Jason is right and the order is wrong. There is a procedure-independent moral fact that anything more than five years is wrong.
Estlund builds his argument on this kind of case. If you are committed to saying that Jason has a duty to carry out the full punishment here, then, Estlund thinks, you will have to admit that executioners should kill convicts they know to be innocent and that soldiers should follow unjust orders.
I am unmoved by Estlund’s example. It seems clear to me not only that Jason may allow the prisoner to escape, but also, if he’s not under duress, he must allow the prisoner to escape. Estlund also describes a case in which a person is wrongly convicted by a fair and impartial jury trial. The defendant is in fact innocent, but the jury makes an honest, blameless mistake, and finds him guilty. Jason the Jailor knows that the defendant is innocent, but, despite knowing this, he cannot prove it to others. Estlund asks, may Jason let the defendant escape? Estlund thinks that it is obvious that Jason may not, but I think it’s obvious that he may. I don’t have the moral intuitions that Estlund wants to build upon.
Strictly speaking, that I, a reasonable person, don’t share Estlund’s intuitions is a problem for him rather than for me. After all, according to Estlund’s own underlying philosophical commitments as a public reason liberal, he needs to justify coercion to all reasonable people. He cannot build a theory on controversial intuitions and judgments that reasonable people reject. Also, it is a part of commonsense moral thinking, by default, that we are presumed not to have a duty to follow unjust orders. This duty must be justified.
Estlund’s other strategy is to argue that disobedience involves some kind of epistemic or moral immodesty. He thinks it can be wrong for you to hold that you have superior judgment to others or to certain collective decision processes, or that it can be wrong for you to just install yourself as the “boss.” For Jason the Jailor to refuse to obey the order would be, in effect, to assert that his judgment is superior to the court’s. Or in effect it would be to assert that he, Jason, is the boss over the court. Estlund adds that even if Jason did have superior judgment, there is no way that Jason could prove to all reasonable people that his judgment was in fact superior. Footnote 26
I find this strand of argument strange. Jason the Jailor might agree that the legal system is reliable overall and the system tends to be more reliable than he. But, he might also hold that in this particular case, it got the wrong answer. And, by Estlund’s hypothesis, indeed it did, and Jason knows this. What justifies Jason in letting the convict escape is not that Jason has some specific moral right to always abide by his own judgment — of course, he does not — but that Jason got the right answer while following correct reasoning procedures. Talking about who is authoritative misses the point. Jason can let the defendant go not because his judgment that the defendant is innocent is his, but because his judgment is correct. Estlund might say to Jason the Jailor, “You might be right that the defendant is innocent, but who made you boss?” Jason might respond, “In this case, no one is the boss. The jury justifiedly but mistakenly believes that the defendant is guilty, so it has legitimacy, but not authority, to jail him. But I know that the defendant is innocent, so I must not obey. The reason I can let the convict escape is not because I’m the boss, but because the jury fails to be my boss.”
Estlund’s worry about Jason taking himself to be the boss might expose some tension in Estlund’s thought. Estlund wants to build his theory of democratic authority within the tradition of public reason liberalism. Public reason liberalism’s fundamental idea is that coercion is presumed unjust, illegitimate, and non-authoritative unless it is justified in a suitably public way to all reasonable people. Footnote 27 On a public reason liberal account, there is a massive asymmetry in what it takes to justify coercion versus what it takes to invalidate it, or in what takes to justify authority versus what it takes to invalidate it. Every reasonable person has a special power to block coercion and invalidate purported authority. For a reasonable person to reject government coercion need not mean that he takes himself to be the boss or to have authority, but that he takes the government to fail to be his boss or to have authority. When Jason refuses to jail the defendant, he is not coercing anyone, and so his refusal to follow orders does not need to be publicly justified.
Moreover, it seems that the citizen of a just society would want Jason to let the innocent man go free. Consider a parallel situation. Suppose I believe that the correct moral principles imply that my son Aiden should do X. But suppose I am mistaken — in fact, the moral principles imply he should do Y. Now, when I tell him to do X, I want him to obey me. But, being a decent human being, I also want even more for him to do what’s in fact right. In this situation, I thus have a de re desire that Aiden do wrong, but an even stronger de dicto desire that he do what’s right. As a moral person, I prefer (i) that my son do what’s right even though I think it’s wrong and ordered him not to, over (ii) that my son do what’s wrong because I think it’s right and ordered him to do so. To prefer (ii) to (i) would be, well, rather vile. For me to prefer (ii) to (i) would mean I am more concerned with being the boss than I am with doing what’s right.
So, similarly, a just democratic society would prefer (i) that its citizens do what’s in fact right, even if the democracy mistakenly ordered the citizens to do something wrong, over (ii) that its citizens do what’s wrong because the democracy mistakenly ordered them to do so. Only a population of vicious people would prefer (ii) to (i).
IV. Anti-Vigilantism
One might try to justify the Special Immunity Thesis by claiming that killing democratic officials is an instance of impermissible vigilante justice. Estlund says, “Vigilante justice is commonly assumed to be wrong once there is an adequate public justice system.” Footnote 28 He adds: “. . . when there is a system that serves the purposes of judgment and punishment without private punishment, then private punishment is morally wrong.” Footnote 29
But one problem with invoking anti-vigilantism here is that it seems to conflate A) killing wrongdoers to stop them from harming others with B) punishment. Anti-vigilante arguments are supposed to show that under certain conditions, private citizens must transfer their natural right to punish to the State. However, the question of permissible defensive killing is not about the right to punish. None of the cases I described above concern punishment. If I claim Ann may kill the hawkish president, I do not claim she has the right to punish him. Instead, I am just saying she may kill him to stop him from causing a nuclear holocaust. (I remain agnostic, for the purposes of this essay, as to whether vigilante punishment is acceptable.)
Another flaw with the anti-vigilantism objection is that it does not offer any grounds for distinguishing between cases A–C and D–F. The anti-vigilante principle forbids us from taking justice into our own hands. It tells us we must instead let the government take care of problems. However, this applies equally well to Shooter in the Park as it does to Minivan Shooter. In both cases, Ann is a “vigilante” — she stops the wrongdoers herself rather than waiting for the police to stop the wrongdoers. So, if the anti-vigilante principle explains why we must not kill democratic officials in D–F, it does so only while also implying that we cannot kill civilians in A–C. What the defender of the Special Immunity Thesis needs is some version of the anti-vigilante principle that allows killing in A–C but not D–F.
Furthermore, if an anti-vigilante principle always forbids violence, even in cases A–C, then the principle seems implausible. When there is an effective and impartial method for administering justice, perhaps one should tend to defer to that system. Yet in each of A–C, Ann is in an emergency situation. In these cases, waiting for the government to fix the problem means innocent people die, period. The anti-vigilante principle is plausible when governments A) are more effective at taking care of the problem, and B) are actually taking care of the problem (or could easily be induced to do so). In those cases, perhaps citizens should let the State handle the problem. However, if those conditions are not met, then there is no obvious reason to hold that citizens must defer to the State.
Suppose I witness a man trying to rape a woman. There are police nearby. I implore them to intervene. They respond, “Sorry, we’re on break. Check back with us later.” I call 911. The operator says, “All the cops are busy. We can be there in 45 minutes.” If the anti-vigilante principle claims that even then, I may not defend the woman, the principle seems false. It certainly conflicts with the commonsense moral principles captured in common law.
V. Peaceful Alternatives
Consider the following argument:
Democracies provide peaceful and effective avenues for citizens to remove unjust leaders and to bring unjust leaders to justice. There are peaceful and effective legal procedures for citizens to stop their governments from committing injustice. Citizens should use these procedures instead of using violence against their leaders or against government agents.
This might seem like an objection to the Moral Parity Thesis and an argument in support of the Special Immunity Thesis. But it’s not.
The argument does not really give us grounds to distinguish between killing wrongdoing civilians or wrongdoing democratic agents. Rather, it is at most an elaboration of the necessity proviso on permissible defensive killing. Recall that according both to McMahan’s theory of permissible killing and the common law, defensive killing is permissible only when it is reasonably believed necessary to protect oneself or others from a grave threat. If a nonlethal alternative is known to be just as effective in preventing that threat, we may not kill the wrongdoers. But this point applies just as much to killing civilians as it does to killing democratic officials.
It might be, as an empirical generalization, that we are more likely to have peaceful means of stopping democratic government’s wrongdoing than we are to stop civilian wrongdoing. If so, it might be, as an empirical generalization, that the conditions under which it is permissible to kill a wrongdoer are less likely to obtain when the wrongdoer is a government agent than when he is a private civilian. But this remains compatible with the Moral Parity Thesis, because it allows that the conditions under which it is permissible to kill wrongdoers are the same.
VI. Good Faith and Following Orders
One might object that a key difference between the civilian and government agent cases has to do with the wrongdoers’ intentions. In Terrorist, Cobra Commander has evil motivations. In Hawk , we might presume that president tries in good faith to serve the nation’s interests; he just has false ideas about how to do so.
If this argument succeeds at all in justifying Special Immunity, it succeeds only when government agents in fact have good intentions. It would still allow that it is permissible to kill government agents, acting ex officio, provided they were not acting in good faith.
This argument has the same fatal flaws as the previous two. It fails to provide an in principle difference between cases involving civilians and government agents. Suppose for the sake of argument that killing a wrongdoer is permissible only if the wrongdoer has bad intentions. Even if so, civilian wrongdoers might also have good intentions. (In Health Nut, the wrongdoers intend to help people.) The good faith objection does not therefore show that the conditions under which it is permissible to kill government agents are more stringent than the conditions under which it is permissible to kill civilians. Rather, at most it allows that the conditions are the same, but are less likely to obtain among democratic government agents than among civilians. So this good faith objection, if sound at all, is compatible with the Moral Parity Thesis and does not support the Special Immunity Thesis.
Most philosophers who have written about the morality of killing think that acting in good faith does make a moral difference. The moral principles that govern killing a well-intentioned agent who acts in good faith, but who makes a grievous moral error, are different than those that govern the killing of an agent with bad intentions. So, for instance, suppose we modify Shooter in the Park somewhat. Imagine there are five shooters with bad intentions and only one child. It seems justifiable to kill all five of them to save the one child. We don’t have to weigh one life against five, because the shooters are culpable while the child is innocent.
In contrast, suppose there are five gunmen and only one child, but suppose they have good intentions. Suppose the gunmen sincerely believe the child is infected with a zombie apocalypse virus, as in 28 Days Later or I Am Legend. Suppose the gunmen have had such a bizarre set of experiences that their belief is justified, though false. The gunmen justifiedly but falsely believe that the child will soon turn into a zombie, and that killing the child is necessary to save the world from a zombie apocalypse. In this case, one might think that the gunmen are “innocent aggressors,” and that saving the child’s life must be balanced against saving theirs. Footnote 30 This seems right. But, again, at most this would show that there is a statistical difference between government agents and civilian actors. It might turn out empirically that when government agents and private civilians commit the same injustices, the former are more likely than the latter to be innocent aggressors acting in good faith. But this allows that the moral principles governing killing them are the same.
A closely related argument holds that the key difference between many civilian and governmental wrongdoers is that the latter are, at least in many cases, following orders. This explains why the former are liable to be killed when the latter are not. However, on closer inspection, this cannot explain the difference in liability between the two.
After all, civilian wrongdoers who are liable to be killed might be following orders. Suppose Kingpin orders his goons to kill Frank Castle’s family. Castle is justified in killing the goons in defense of others. Or, in case B above (Health Nut), Ann is justified in killing the henchman, even though the henchman is following orders.
What matters here is not whether the civilian or the governmental agent is following orders, but whether the orders are just. If a colonel orders soldiers to massacre the residents of My Lai, anyone could still justifiably kill the soldiers to stop the massacre. On the other hand, if a police captain orders his officers to arrest a murderer, the murderer should not resist, because the murderer should in fact be arrested and tried.
One might finally try to say that “following orders” makes a difference because those who follow the orders are sometimes under threat of duress. For instance, a soldier or a cop who fails to follow an unjust order might suffer a penalty, including, in some cases, execution. As Jeff McMahan concludes, a person who is a threat to others but is acting under duress might be, to some extent, an “Excused Threat.” Footnote 31 The idea here is that acting under duress relieves a person, to some extent, of his culpability in doing the wrongful action. However, McMahan says, Excused Threats are still liable to be killed under the same conditions as Unexcused Threats.
To see why, consider case A again, Shooter in the Park. Suppose Ann learned that the only reason the shooter was shooting the children was because a criminal had threatened to kill the shooter’s son unless he in turn killed ten other innocent children. Here, the shooter might not be blameworthy for trying to kill the children in the park, but Ann is still justified in shooting him in order to protect them. So it also goes with governmental cases. Suppose a democratic government conscripts soldiers and orders them, on pain of execution, to invade a country and kill its innocent citizens and soldiers. The innocent citizens and soldiers — or you, or I — may still kill the invading soldiers, though many of the soldiers qualify as Excused Threats.
VII. Retaliation, Fall-Out, and Extortion
We are still looking for reasons why it would be impermissible to kill government officials in cases D–F even though it is permissible to kill civilians in analogous cases A–C. One such objection goes as follows:
If citizens believed they were at liberty to kill democratic officials (under the principles described above), then this would cause dangerous instability and fallout. If civilians kill a bad cop, the other cops are likely to retaliate by harming other innocent people or curbing their rights. If civilians kill an evil president, future presidents or congress are likely to retaliate by harming other people or further violating their rights. Therefore, it is wrong to assassinate democratic officials.
The idea here is that morality is a strategic game. What I am permitted to do might depend on how others will respond to what I do. Perhaps what would otherwise have been a permissible action might be rendered impermissible if others will perform wrongful actions in response to it. That is, perhaps the threat of extortion might change my moral duties.
Presumably, there is no moral duty to choose a red over a blue toothbrush. However, suppose a terrorist threatens to nuke Washington DC unless I choose blue. Must I then choose blue?
Consider a variation on minivan shooter. Suppose Ann is about to kill the cop who is shooting at the children. However, just as she does so, another cop yells to her, “We cops stick together. If you shoot him, I’ll kill two minivans full of innocent kids. That’s not a threat; that’s a promise.” Is it still permissible for Ann to save the kids in the first van, or must she submit to the cops’ threat?
These are hard questions. How we are required to respond to extortion is bound to be controversial. Fortunately, though, I don’t need to answer these questions here. Even if we assume for the sake of argument that (i) extortion or retaliation do indeed render impermissible what would otherwise would have been permissible acts, and that (ii) democratic governments are likely to use extortion and retaliation to prevent civilians from killing wrongdoing government agents, this still does not justify the Special Immunity Thesis. Instead, it is compatible with the Moral Parity Thesis.
For the sake of argument, suppose it is impermissible for you to kill a wrongdoer if there is a real threat that others will respond by committing even greater harm or injustice. This provides us with no in-principle reason to treat democratic government officials differently from civilians. After all, civilians can and often do respond to what otherwise would have been justifiable violent self-defense or defense of others by threatening to cause even more harm. A bully on the playground might threaten to beat up two other kids if you stick up for your friend. The Mafia can and does tell people that they will kill even more people if their victims start to defend themselves. The Joker might threaten to bomb Gotham City if Batman tries to rescue Commissioner Gordon.
It may turn out, empirically, that democratic governments are unusually willing and able to use extortion to prevent us from defending ourselves against their wrongdoing agents. If so, it may thus turn out, empirically, that the conditions under which it is permissible to kill a wrongdoer are less likely to obtain when the wrongdoer is a government agent than when he is a private civilian. But this remains compatible with the Moral Parity Thesis, because it allows that the conditions under which it is permissible to kill wrongdoers are the same. In both cases, we are allowed to kill wrongdoers under certain conditions, one of those conditions being that killing the wrongdoers will not incite other wrongdoers to commit even greater harm or injustice.
So far, I have assumed for the sake of argument that we are required to surrender to credible threats of extortion. But that is not obviously true. It is not obvious that what would have been a permissible action becomes wrong just because someone else threatens to react badly to it. Suppose I kill the would-be Tuvalu-nuking president, even though I know my fellow citizens will react by rioting. During the riots, they kill ten thousand innocent Americans (more than the population of Tuvalu.) It is at least not obvious that this makes the assassination wrong when it otherwise would have been right. After all, when I kill the president, my fellow citizens are obligated not to riot in response. They should instead apologize for their despicable support of war and praise my heroism.
As I noted in the introduction, many believe it is justifiable to assassinate totalitarian dictators, such as Stalin or Hitler. However, killing a totalitarian dictator or a criminal mastermind seems more likely to endanger innocent third parties than killing a democratic official. Fanni Kaplan tried but failed to assassinate Lenin in 1918. Lenin and his government responded with the Red Terror. Even if Kaplan had killed Lenin, there was a good chance Lenin would have been succeeded by someone worse or at least equally bad. Totalitarian communist regimes do not value individual human life. After a successful assassination, newly installed dictators are likely to terrorize citizens into submission.
Compare this to the United States and other democracies. Four U.S. presidents have been assassinated, and many more have been targets. Thirteen congresspersons have been assassinated, and a few others have been targets. None of these events resulted in humanitarian disasters or terror purges. The United States has committed a great many atrocities, but not in response to assassination. Assassinating Lincoln got us Andrew Johnson. The attempt to assassinate Reagan just got us stronger gun control laws. The attempted assassination of Gabriel Giffords resulted in public figures pledging (insincerely, it turned out) to use less aggressive political rhetoric. When the IRA assassinated MP Ian Gow in 1990, the British did not respond by killing innocent Irish citizens. When Swedish Prime Minister Palme was assassinated in 1986, the government convicted a suspect of the murder, but the conviction was overturned on appeal. And so on. Compared to other forms of government, democracies tend to be more concerned with their citizens’ welfare. For this very reason, assassination in democracies will tend to be quite safe — democracies do not respond by crushing their citizens. Political scientists who study this issue empirically tend to find that democracies handle assassinations well, and the fallout from assassination is minor. Footnote 32
VIII. Conclusion
Democratic states tend to be winners from a civil rights or humanitarian point of view. Yet democratic leaders and agents frequently commit severe injustices that go far beyond what could plausibly been regarded as their authoritative scope of power. In some cases, these agents pose severe threats to innocent people. In similar circumstances, when civilians pose such threats, we are morally permitted to kill those civilians. If so, then we should presumptively conclude that in such circumstances, we are also permitted to kill democratic leaders, officials, and agents. This presumption would be defeated if we found some a plausible reason to treat democratic officials differently. After much searching, we have found no such principle.
