It's OK, no one's actually going to read all this
Freedom From Evil Through Ethical Constraints
One aspect of commonsense moral thinking is the maxim “the ends don’t always justify the means”, which seems to advocate for ethical constraints on what actions are permissible. This is irreconcilable with purely consequentialist theories, as they will claim whatever outcome produces the best outcome is always morally required, regardless of the means used. The commonsense endorsement of constraints isn’t itself a good argument, because it begs the question. Instead, the best argument for ethical constraints is a reductio ad absurdum, which will show that any ethical theory without the right kind of constraints will be dangerous and undesirable. This will result in an agent-focused theory of constraints of the kind McNaughton and Rawling support, and rule out both pure consequentialist and victim-focused constraints as adequate moral theories.
A theory of constraints will postulate that some actions are ethically impermissible, even if they result in the best consequence. For example, if we were made to choose between killing one person or letting five people die, a constraint theory could, and likely would claim we cannot kill the one person. This would be so because our actions are constrained in some way that makes killing him ethically impermissible. Under a pure consequentialist theory (PC), assuming these people are equivalent in the ethically relevant respects, it would be best to have one dead person when all is said and done than five, regardless of how that is achieved. The two strongest constraint theories are the victim-focused view and the agent-focused view (VF and AF). VF holds that constraints arise from certain rights that people have, and these rights cannot be violated merely because the outcome would be better. Not all rights need be like this; all that VF claims is that some rights are strong enough to merit this kind of force. The term ‘right’ from here on it will exclusively refer to the latter kind of rights. The right to not be killed will be supposed as an example of this kind of right throughout the paper. The conclusion we can draw if we accept VF is that actions are constrained to the extent they violate the rights of others. AF, conversely, claims that constraints come not from a property the victim has (having certain rights) but on a duty of the agent to not infringe on the rights of the victim. The critical idea here is that every moral agent has a special duty to not himself violate the rights of others. Put in terms of constraint, an action would be constrained to the extent that it would involve the agent performing the action to violate the rights of another.
Since it is not immediately obvious why we should think there are moral constraints, it will be fruitful to examine the alternative: PC. If there are no moral constraints, then we will always be morally required to act in whatever way results in the best outcome. However, cunning evildoers will be able to leverage this system of ethics to have good agents meet their evil ends by making it the best outcome of all possible scenarios. Imagine if an evildoer captured five people, and will kill them unless we kill one person; further, the five people and the evildoer are too far away to affect. Assume that all the potential victims are the same in all ethically relevant ways, and there is some guarantee that the evildoer will do as he says. PC would require us to kill the person the evildoer wants dead, to prevent five other deaths. This would hold in all cases like this, leading to good agents frequently being made to pursue evildoers’ agendas. Then there are situations where an agent becoming a tool of evil achieves the best outcome. So, according to PC, there will be times when we are morally required to become the tools of evil. This is ridiculous, as no adequate system of ethics should be exploitable in such a way that we will be morally required to become tools of evil. Thus, constraints are needed in order that we not serve the agendas of evildoers.
The last premise in the above argument, “no adequate system of ethics should be exploitable in such a way that we will be morally required to become tools of evil”, arises from an intuitive understanding of what kind of thing morality is. This places it in the same category as the maxim we ruled out at the beginning of the paper, that of claims based on intuition, but it differs from it in two respects. Firstly, it is not question-begging: it by itself does not say anything about the existence of constraints. It is rather a claim about what morality is, and what functions it can and cannot serve. Secondly, it is extremely plausible. If we look at the alternative, namely, that an ethical system can morally require agents to be tools of evil, while not logically inconsistent, is extremely undesirable and undermines moral motivation. After all, if morality is the kind of thing that can turn someone into a tool of evil, what reason do we have to adhere to it? It seems almost preferable to disregard morality in that case, as it would be something that facilitates the machinations of evildoers. And so we are left with no reason to favor any ethical theory that makes good agents into tools of evil.
An adherent of PC might claim here that the best outcome might still be achieved by not killing the one person. True, five people die in its place, but this act discourages these evildoers from enacting these hostage scenarios, and will reduce the frequency of this kind of situation to the point where more lives are saved in the long run. It should be easy to see how this kind of numbers game fails to provide a real counter-argument. There’s nothing stopping an evildoer in this case from taking more hostages until the number of hostages exceeds the number of future saved lives. If never killing a person for an evildoer in this kind of situation causes the frequency of this situation to drop to one-tenth what it would be otherwise, an evildoer would just have to take more hostages until the deaths of the hostages outweighed even the long-term gains. Even if ten more people die as a result of serving the agenda of evildoers, if each one takes, say, fifty hostages, we will still be morally required to do kill the ten. Thus, good agents will once again become the tools of evil, and PC falls to the same objection as before.
The question left to answer is if we need constraints, as has been shown here, what is the nature of these constraints? In other words, between VF and AF, which is the better theory? They both constrain us against killing one to save five others, but consider a slightly different example. An evildoer wants to see someone kill someone else, and so kidnaps five people. He puts the five in front of you, and once again, they are identical in all ethically relevant ways. He has also supplied a gun with one bullet, and he tells you that if you kill one of the five, the rest go free; if you don’t kill one, he will kill them all. Again, you have a guarantee that he will abide by his word. Under VF, because who specifically does the violation is irrelevant, you have two choices: fire, and there will be one violation of a right. Don’t fire, and there will be five violations of that same right. Acting to limit the number of right violations within the same group seems to be mandated by VF because it has less of the same constraint on it than the only alternative, and so you must kill the one and serve the evildoer’s agenda. This means just like PC, VF can lead to good agents becoming tools of evil, and should be rejected on the same grounds as PC was. AF, alternatively, would morally require you to not shoot, on the grounds you have a special duty to the victims to not violate their rights. Even though the evildoer is going to do away with them all, you are required by AF to refrain from killing anyone yourself. Thus AF is the only form of constraint theory to avoid the objection that made us abandon VF and PC.
There is a concern that AF may commit us to some kind of similar trade-offs between victims when the same agent introduced the threat to the group as attempts to minimize it. If an agent throws a grenade at five people, and then wishes to minimize his wrongdoing upon realizing what he has done, it seems prima facie permitted, if not required, to throw one person on the grenade so that the four others may live. As McNaughton and Rawling explain, this is not the case; the act that infringed on their rights was the throwing of the grenade with the intent of harm. According to AF the constraints are on knowingly ignoring your moral duties, rather than the effects that causes. Throwing someone on there after the fact will not change the initial wrongdoing, it will only infringe on one person’s rights another time. Secondly, it is hard to see how this problem is actually problematic. Even if AF were formulated in a way that required this kind of trade-off, by following it, we would not turn into tools of evil. We could only arrive at this situation by going against AF to begin with, and as such, it is a poor attack against it.
We have seen that in order to avoid some extremely serious consequences, we have to accept that at times, maximizing the good has to be curtailed by constraints so morality cannot be abused by evildoers. And the only kind of constraint theory that can avoid all the scenarios where an evildoer could exploit is the agent-focused view.
One aspect of commonsense moral thinking is the maxim “the ends don’t always justify the means”, which seems to advocate for ethical constraints on what actions are permissible. This is irreconcilable with purely consequentialist theories, as they will claim whatever outcome produces the best outcome is always morally required, regardless of the means used. The commonsense endorsement of constraints isn’t itself a good argument, because it begs the question. Instead, the best argument for ethical constraints is a reductio ad absurdum, which will show that any ethical theory without the right kind of constraints will be dangerous and undesirable. This will result in an agent-focused theory of constraints of the kind McNaughton and Rawling support, and rule out both pure consequentialist and victim-focused constraints as adequate moral theories.
A theory of constraints will postulate that some actions are ethically impermissible, even if they result in the best consequence. For example, if we were made to choose between killing one person or letting five people die, a constraint theory could, and likely would claim we cannot kill the one person. This would be so because our actions are constrained in some way that makes killing him ethically impermissible. Under a pure consequentialist theory (PC), assuming these people are equivalent in the ethically relevant respects, it would be best to have one dead person when all is said and done than five, regardless of how that is achieved. The two strongest constraint theories are the victim-focused view and the agent-focused view (VF and AF). VF holds that constraints arise from certain rights that people have, and these rights cannot be violated merely because the outcome would be better. Not all rights need be like this; all that VF claims is that some rights are strong enough to merit this kind of force. The term ‘right’ from here on it will exclusively refer to the latter kind of rights. The right to not be killed will be supposed as an example of this kind of right throughout the paper. The conclusion we can draw if we accept VF is that actions are constrained to the extent they violate the rights of others. AF, conversely, claims that constraints come not from a property the victim has (having certain rights) but on a duty of the agent to not infringe on the rights of the victim. The critical idea here is that every moral agent has a special duty to not himself violate the rights of others. Put in terms of constraint, an action would be constrained to the extent that it would involve the agent performing the action to violate the rights of another.
Since it is not immediately obvious why we should think there are moral constraints, it will be fruitful to examine the alternative: PC. If there are no moral constraints, then we will always be morally required to act in whatever way results in the best outcome. However, cunning evildoers will be able to leverage this system of ethics to have good agents meet their evil ends by making it the best outcome of all possible scenarios. Imagine if an evildoer captured five people, and will kill them unless we kill one person; further, the five people and the evildoer are too far away to affect. Assume that all the potential victims are the same in all ethically relevant ways, and there is some guarantee that the evildoer will do as he says. PC would require us to kill the person the evildoer wants dead, to prevent five other deaths. This would hold in all cases like this, leading to good agents frequently being made to pursue evildoers’ agendas. Then there are situations where an agent becoming a tool of evil achieves the best outcome. So, according to PC, there will be times when we are morally required to become the tools of evil. This is ridiculous, as no adequate system of ethics should be exploitable in such a way that we will be morally required to become tools of evil. Thus, constraints are needed in order that we not serve the agendas of evildoers.
The last premise in the above argument, “no adequate system of ethics should be exploitable in such a way that we will be morally required to become tools of evil”, arises from an intuitive understanding of what kind of thing morality is. This places it in the same category as the maxim we ruled out at the beginning of the paper, that of claims based on intuition, but it differs from it in two respects. Firstly, it is not question-begging: it by itself does not say anything about the existence of constraints. It is rather a claim about what morality is, and what functions it can and cannot serve. Secondly, it is extremely plausible. If we look at the alternative, namely, that an ethical system can morally require agents to be tools of evil, while not logically inconsistent, is extremely undesirable and undermines moral motivation. After all, if morality is the kind of thing that can turn someone into a tool of evil, what reason do we have to adhere to it? It seems almost preferable to disregard morality in that case, as it would be something that facilitates the machinations of evildoers. And so we are left with no reason to favor any ethical theory that makes good agents into tools of evil.
An adherent of PC might claim here that the best outcome might still be achieved by not killing the one person. True, five people die in its place, but this act discourages these evildoers from enacting these hostage scenarios, and will reduce the frequency of this kind of situation to the point where more lives are saved in the long run. It should be easy to see how this kind of numbers game fails to provide a real counter-argument. There’s nothing stopping an evildoer in this case from taking more hostages until the number of hostages exceeds the number of future saved lives. If never killing a person for an evildoer in this kind of situation causes the frequency of this situation to drop to one-tenth what it would be otherwise, an evildoer would just have to take more hostages until the deaths of the hostages outweighed even the long-term gains. Even if ten more people die as a result of serving the agenda of evildoers, if each one takes, say, fifty hostages, we will still be morally required to do kill the ten. Thus, good agents will once again become the tools of evil, and PC falls to the same objection as before.
The question left to answer is if we need constraints, as has been shown here, what is the nature of these constraints? In other words, between VF and AF, which is the better theory? They both constrain us against killing one to save five others, but consider a slightly different example. An evildoer wants to see someone kill someone else, and so kidnaps five people. He puts the five in front of you, and once again, they are identical in all ethically relevant ways. He has also supplied a gun with one bullet, and he tells you that if you kill one of the five, the rest go free; if you don’t kill one, he will kill them all. Again, you have a guarantee that he will abide by his word. Under VF, because who specifically does the violation is irrelevant, you have two choices: fire, and there will be one violation of a right. Don’t fire, and there will be five violations of that same right. Acting to limit the number of right violations within the same group seems to be mandated by VF because it has less of the same constraint on it than the only alternative, and so you must kill the one and serve the evildoer’s agenda. This means just like PC, VF can lead to good agents becoming tools of evil, and should be rejected on the same grounds as PC was. AF, alternatively, would morally require you to not shoot, on the grounds you have a special duty to the victims to not violate their rights. Even though the evildoer is going to do away with them all, you are required by AF to refrain from killing anyone yourself. Thus AF is the only form of constraint theory to avoid the objection that made us abandon VF and PC.
There is a concern that AF may commit us to some kind of similar trade-offs between victims when the same agent introduced the threat to the group as attempts to minimize it. If an agent throws a grenade at five people, and then wishes to minimize his wrongdoing upon realizing what he has done, it seems prima facie permitted, if not required, to throw one person on the grenade so that the four others may live. As McNaughton and Rawling explain, this is not the case; the act that infringed on their rights was the throwing of the grenade with the intent of harm. According to AF the constraints are on knowingly ignoring your moral duties, rather than the effects that causes. Throwing someone on there after the fact will not change the initial wrongdoing, it will only infringe on one person’s rights another time. Secondly, it is hard to see how this problem is actually problematic. Even if AF were formulated in a way that required this kind of trade-off, by following it, we would not turn into tools of evil. We could only arrive at this situation by going against AF to begin with, and as such, it is a poor attack against it.
We have seen that in order to avoid some extremely serious consequences, we have to accept that at times, maximizing the good has to be curtailed by constraints so morality cannot be abused by evildoers. And the only kind of constraint theory that can avoid all the scenarios where an evildoer could exploit is the agent-focused view.