I am grateful to Larry Solum for recommending my new draft paper, Punishment and Moral Risk, and for his characteristically thoughtful commentary. Larry makes four points, and I'd like to respond to each:
Solum's First Point: Kolber's analysis assumes that the relevant retributivist beliefs are independent of one another, but it is not clear that this is the case. At an abstract level, the various retributivist beliefs may be part of a "web of belief" to use Quine's phrase--and hence mutually dependent in almost all cases to at least some degree. Ron Allen has made a similar point in the context of evidence law and the burdens of persuasion. More concretely, retributivist intuitions about free will and about the moral standards for adequate proportionality may rest on overlapping premises.
Reply: In fact, I don't assume that the relevant retributivist beliefs are independent of each other. Rather, I ask readers to offer their levels of confidence as to each retributivist commitment while "[t]aking the prior numbered propositions as given" (draft p.13 et seq.). This is another way of addressing Larry's concern. So long as reader's follow the instructions, we should be able to multiply the relevant probabilities. More substantively, I accept Larry's point that I should offer some discussion of the matter. What I would say is something like this: Sure, there will likely be some interdependence in retributivist commitments to these propositions. But there's also a lot of independence. For example, many of the people who think that retributivism is barbaric nevertheless believe in free will. And certainly our beliefs about the factual guilt of a particular defendant will be largely independent of our beliefs about either of these philosophical issues. So, I welcome Larry's suggestion to clarify the matter in the next draft, but I don't believe his point detracts much from my thesis. (To test this, I invite readers to give high probabilities to propositions that they believe are highly dependent on prior propositions.)
Solum's Second Point: Kolber's argument that consequentialism does not share the "problem" he identifies may be based on a double standard. Thus, consequentialist theories of punishment could be thought to depend on: 1) the truth of consequentialism as a moral theory despite the fact a substantial share (perhaps a majority) of moral philosophers reject it, 2) the belief that deterrence, incapacitation, or rehabilitation is actually efficacious, when there is substantial evidence that each of these functions fails in practice, and 3) the belief that actual punishments can be calibrated to achieve the benefits of these consequentialist justifications for punishment.
Reply: I do claim that my criticism strikes more sharply at retributivists than consequentialists. Consequentialism has plenty of uncertainty but it's largely of a different sort. As I write in the draft: "The reason that I treat consequentialism and retributivism quite differently from an epistemic perspective is that the path to resolving consequentialism’s empirical uncertainty is much clearer than the path to resolving retributivism’s moral uncertainty. At least in principle, we know how to gather evidence and set up experiments to estimate how punishment policies will affect deterrence, incapacitation, and rehabilitation. The means of resolving age-old debates about free will and proportionality, however, are highly disputed and have been for centuries." (draft p. 32).
Let me address each point more specifically, though. In #1, it's true that lots of people are not consequentialists. But consequentialists are! So they will likely have a high degree of confidence in their own beliefs. They should not be overconfident, of course. But one can be a reasonable consequentialist with high confidence in consequentialism. As for #2, the consequentialist doesn't have to believe that deterrence, incapacitation, and rehabiltation are particularly efficacious. They just believe that those are the sorts of things that justify punishment. If they're not efficacious, then the consequentialist will deem little or no punishment justified. That being said, it's hard to doubt that incapacitation often reduces the harms that some very dangerous people cause and that punishment provides at least something of a deterrent effect (see, e.g., The Purge). Finally, the key point as to #2 and #3 is that both issues are largely about empirical uncertainty, and, to the extent that there is empirical uncertainty for consequentialists, there is uncertainty in both acting and in failing to act. True, we don't know for sure if adding a year in prison for some particular offense will generate net benefits. But we may also be unsure if failing to add a year will cause unnecessary risk of harm to victims. The good consequentialist will have to balance these. It's a difficult task. But the challenge is one we know how to address in principle (relative to answering questions about free will and the like). The asymmetry in the way typical consequentialists and typical retributivists treat the act/inaction distinction is indeed critical to my view and leads us to Larry's next point.
Solum's Third Point: In addition, Kolber fails to consider the moral risk of failing to punish from a retributivist perspective. Many retributivists believe that there is a moral requirement to impose deserved punishment and that failure to punish is a moral wrong. One can easily invert Kolber's argument and show that retributivists cannot justify a failure to punish. I am not suggesting that the inverted argument is correct; rather, the point is that the structure of the argument is suspect.
Reply: Almost all retributivists find it substantially worse to punish the undeserving than to fail to punish the deserving. But I agree with Larry that this need not be the case. That's why I state that my view only applies to versions of retributivism that "share the Blackstonian value that it is substantially worse to punish the undeserving than to fail to punish the deserving." (draft p.5-6).
Solum's Fourth Point: Finally, I am dubious that retributivists themselves believe that punishments must be precisely proportionate. The relevant standard for a complex legal system is much more likely to be "rough proportionality."
Reply: Many retributivists would say that there is a firm deontological prohibition against knowingly or recklessly punishing someone in excess of desert. Larry is probably right that many would frame all of this in terms of rough proportionality. I am dubious of that approach. Why is it strictly forbidden to give an innocent person one day in prison, but it's okay if, say, we take a significant risk that a murderer will spend a year in prison that exceeds his desert? But I'd be fine to adjust the relevant proposition to speak of rough proportionality. I don't think it will affect my thesis, as I discuss nine retributivist propositions required to punish in a particular case and tweaking the probability of one them will likely have only a modest effect. There are, of course, many versions of retributivism, and I say in the paper (draft p.5) that though "I surely do not address every version of retributivism, the thrust of my argument, with modest adjustments, applies to a broad range of retributivist views." The more pure and the more traditional one's retributivism, the stronger I take my argument to be. But I think it applies to lots of versions of retributivism with adjustments along the way.
Let me end by thanking Larry again for sharing his insightful comments and for offering great suggestions as to matters that I can clarify in the next draft!
