Retributivism claims that people should be treated as they deserve (Rachels 1997). However, how do we determine what one deserves? Should we be judged for factors which are out of our control, matters of luck? According to Kant (1784), “A good will is not good because of what it effects or accomplishes, because of its fitness to attain some proposed end, but only because of its volition, that is, it is good in itself (...)”. Therefore, morality should be immune to phenomena of luck. Nevertheless, we judge people every day for things outside of their control. Nagel (1976) writes: “where a significant aspect of what someone does depends on factors beyond his control, yet we continue to treat him in that respect as an object of moral judgment, it can be called moral luck”. Moral luck can take on many different forms, such as resultant luck or constitutive luck. Resultant luck is concerned with “how things turn out” (Wolf 2013), while constitutive luck is about “luck in who one is, or in the traits and disposition one has” (Nelkin 2019).

The problem of moral luck has often been discussed in terms of resultant luck and its consequences with regards to criminal law. For example, consider two criminals, equally wicked in their desire to kill but only one of them succeeds. Today, we often differentiate punishment for attempted versus successful murder. Is it just? Those who subscribe to Nagel’s view and hence believe in moral luck will claim that differential punishment is just. People can be justly punished for matters outside of their control. On the other hand, people who reject moral luck will claim that differential punishment is unjust: people should be equally blamed and punished, independently from bad or good resultant luck.

The two views seem irreconcilable. However, Lewis (1989) attempted to show that our current penal system is just (or at least not obviously unjust) while still rejecting resultant luck. In Lewis’ view, “it is not unjust to punish success more severely than failure when this practice is properly understood as a natural penal lottery” (Khoury 2018).

I argue in this essay that Lewis’ natural penal lottery might be clearly unfair in some cases.

My argument goes as follow:

Premise 1: One requirement for a penal lottery to be not obviously unfair is that equally blameworthy people should be subjected to the same odds of receiving harm.
Premise 2: If one rejects resultant luck, one will typically reject constitutive luck too.
Premise 3: Resultant luck is not independent from constitutive luck.
Premise 4: The combination of premises 2 and 3 implies that equally blameworthy people could be subjected to different odds of punishment in a natural penal lottery.
Conclusion: Lewis’ natural penal lottery is bound to be unfair in some cases.

I will first give an illustration of resultant luck and an overview of Lewis’ argument. Then, I will discuss each premise in order.

Case 1: Imagine two persons, A and B. They both drunk drive on a given day. Driver A hits and kills a child. Driver B, however, makes it home without any incident. For those who reject resultant luck, it seems difficult to explain why, ceteris paribus, we typically punish A more than B since both are guilty of the same act.

Lewis (1989) offers a possible explanation. By engaging in an illegal act, both A and B subject themselves to a penal lottery. If they lose the lottery, they will be harmed. If they win the lottery, no harm will be done to them. The odds and the harm are determined by the nature of the illegal act and not by the result of the act. Whether A or B receives harm or not is, on the contrary, decided by the result of the act. In other words, A and B both play a lottery, and are subjected to the exact same risk of harm. Then, A loses the lottery because his action results in the killing of the child. On the other hand, B wins the lottery because her action did not result in the killing of the child. Hence, the outcome of the natural penal lottery is not determined by drawing straws but by the result of the action itself. However, it manages to reject resultant luck because we recognize that A is no more guilty than B.

Now, is such a system just? Lewis (1989) argues that the natural penal lottery is not obviously unjust at least. The lottery might at first sight be violating the proportionality principle of retributive views on punishment, according to which the punishment should be proportional to the blameworthiness (Khoury 2018). A is in prison, B is not. The lack of proportion is evident. However, Lewis (1989) argues that criminals subject their victims to a risk of harm and therefore should be submitted to a risk of harm themselves. Both A and B drunk drive and subject all potential children on the road to a risk of being killed, but they also subject themselves to a risk of harm by entering the lottery. As the lottery outcome is determined by resultant luck, the proportionality principle is then satisfied since the criminals expose themselves to the same risk of harm to which they expose the children (assuming the punishment for killing a child is a death sentence). A corollary of the proportionality principle, and an important fairness requirement, is that equally blameworthy people should be punished equally (Lewis 1989), which implies they should be faced by the same odds, in the case of the natural penal lottery (premise 1).

Let’s now discuss the rest of the premises. Premise 2 stipulates that if one rejects resultant luck, one will typically also reject constitutive luck. Let’s define case 2. In case 2, A had a troubled upbringing which pre-disposed him to drink driving twice a week. On the other hand, B had a happier upbringing which leads her to drunk drive once a week only. I argue that if one tends to refuse attributing different levels of guilt to A and B in the original case, then the same person would probably agree that the guilt in case 2 is not the same for A and B. If one reject resultant luck, one asserts the rationale that we should blame people only for things which are under their control. The same rationale in the case of constitutive luck would give the following result: if we are to blame people only for things under their control, then we should not hold A responsible for having had a more difficult upbringing than B. Therefore, we should blame A less than B for drunk driving. The result holds since both constitutive luck and resultant luck are instances of moral luck, and one who rejects moral luck agrees that we should not judge people on matters of good or bad luck. Further, I take the rejection of constitutive luck to be even less controversial than the rejection of resultant luck. When one rejects resultant luck, one makes no difference between the case where one child is killed, and the case where no child is killed. Rejecting constitutive luck in case 2 is not as demanding. One only needs to grant that we should show more lenience towards blaming the person who had worse dispositions because of their upbringing.

Premise 3 claims resultant luck depends on constitutive luck. To see this, let’s first assign a probabilistic assessment of resultant luck in case 1. Let’s assume that the probability for each driver to hit the child is 1/10. Now, let’s assign a probabilistic assessment to the killing of the child in case 2. We keep assuming that, when one drunk drives, the probability of hitting a child is 1/10. Further, because of his constitutive bad luck, A is obviously more likely than B to drunk drive. The probability for A to drunk drive on a given day is 2/7, and the probability for B to drunk drive on a given day is 1/7. The odds of receiving harm is 1/10 for both drivers in case 1. However, in case 2, A faces the odds 2/70, and B faces the odds 1/70. The point here is not to demonstrate already that the natural penal lottery is unfair. It is obvious that, because of their difference in constitutive luck, A and B are not equally to blame and hence we cannot compare and discuss their odds in the lottery straight away. However, the point is to show that constitutive luck and resultant luck cannot be considered as independent concepts, as the earlier will influence the later. As one experiences bad constitutive luck, resultant luck tends to worsen.

According to premise 4, premise 2 and 3 must imply that equally blameworthy people might be subjected to different odds in a natural penal lottery. Let’s consider case 3. A drunk drives once per week, half because of constitutive bad luck, half because of his own will. Since we reject constitutive bad luck, we agree that A is not to blame for “half of his drunk driving”. B also drunk drives once a week but is fully to blame for it (B does not have constitutive bad luck). The summary of case 3 is illustrated in the table below:

Table 1: Summary of Case 3.
Table 1: Summary of Case 3.

A and B are subjected to the same odds in the penal lottery. However, A and B are not equally blameworthy. We want to see whether the odds are the same for all participants of the natural penal lottery when they are equally blameworthy. To do so, we can adjust blameworthiness through changing the probability of drunk driving. We create a hypothetical situation, case 3'. Let’s say now that the probability for A of drunk driving is 2/7. We established before that his constitutive bad luck is such that he is only half to blame for drunk driving once per week. Therefore, if A drunk drives twice per week, he will be to blame for half of those drunk drives. Instead of halving the blame in each drunk drives, one can consider that the first drunk drive of the week is the one A is fully blameable for, and the second drunk drive of the week is due to bad constitutive luck. One can now compare A’s situation with B’s:

Table 1: Summary of Case 3'.
Table 1: Summary of Case 3'.

The table shows that A and B, despite being equally to blame for drunk driving, do not face the same odds of receiving harm. Hence, by premise 1, the lottery cannot be considered to not be obviously unfair anymore.

To sum it up, it might at first seem irrelevant to discuss constitutive luck as it creates a difference in blameworthiness among participants of the lottery. Therefore, their odds of receiving harm need not be equal. However, one can of course account for such a difference in blameworthiness by creating hypothetical cases: “If A was to be blamed equally than B, how much should his odds in the lottery change by?” In such hypothetical cases, we observe that the odds faced by A are less advantageous than those faced by B. Hence, there are instances where a natural penal lottery can be unfair. More specifically, when differences in constitutive luck are not taken in account, the lottery is bound to be unfair. It is important to note that those instances are not special cases or extensions of the natural penal lottery which arise when one considers both constitutive luck and resultant luck, instead of only resultant luck. Indeed, resultant luck itself is directly dependent on constitutive luck, and cannot be considered independently. One would therefore be unable to give a satisfying answer to the problem of resultant luck if we ignore that fact.


  1. Chinnery, A. 2015. On Moral Luck and Nonideal Moral Education. Educational Theory 65(2): pp.169-181. DOI:
  2. Enoch D. 2010. Moral Luck and the Law. Philosophy Compass. 5: pp.42-54. DOI:
  3. Khoury, AC. 2018. Criminal Attempts and the Penal Lottery. Australasian Journal of Philosophy. 96(4): pp.779-792. DOI: https://10.1080/00048402.2017.1422768.
  4. Kant, I. Gregor, MJ. and Timmermann, J. (eds. and transl.) 1784 [2012]. Groundwork of the Metaphysics of Morals Cambridge: Cambridge University Press.
  5. Lewis, D. 1989. The Punishment that Leaves Something to Chance. Philosophy & Public Affairs. 18(1): pp.53-67.
  6. Nelkin, DK. 2019. Moral Luck. 26 January 2004. Available at [Last accessed 31 January 2020].
  7. Rachels, J. 1997. Punishment and Desert. In: Hugh LaFollette (ed.) Ethics in Practice. Oxford: Blackwell. pp.470–479.
  8. Williams, B.A.O. and Nagel, T. 1976. Moral Luck. Proceedings of the Aristotelian Society, Supplementary Volumes. 50: pp.115-135+137-151. DOI:
  9. Wolf, S. 2013. The Moral of Moral Luck. Philosophic Exchange. 31(1): Article 1.