Rule-consequentialism holds that people should always act in accordance with rules that will promote the best results overall if generally followed even if, in the particular case, the good-maximizing alternative is to violate the rules. Act-consequentialism, by contrast, holds that people should always act to maximize the good. Rule-consequentialism is a species of consequentialism because its rules are justified on the grounds that they maximize good consequences; it is different from act-consequentialism in that the two theories direct actors to do different things in the same circumstances.
John Rawls supplies a rule-consequentialist justification for punishment that is meant to avoid Boonin’s punishing the innocent objection. It has two lemmas. The first lemma is that the rule that will promote the best results overall — the “ideal punishment rule” — is a rule that only authorizes punishment of offenders. The ideal legislature will thus draw a line between offenders and non-offenders. The second lemma is that the “ideal judge” will invariably follow the ideal legislature’s rule; in light of the fact that the ideal punishment rule does not authorize the punishment of innocent people, the fact that sentencing a non-offender in a particular case will do more to promote the good than not sentencing him does not license a judge to punish the non-offender.
The First Lemma
Rawls supports the first lemma — that the ideal punishment rule will only authorize the punishment of offenders — by imagining an alternative practice, “telishment,” that allows a judge to “telish” an innocent person if “there is, at the time, a wave of offenses similar to that with which they charge him and telish him for” (that is: if the conditions are ripe for deterrence . . . assuming deterrence works) (Boonin, 65; quoting Rawls). It could well be that punishment plus the wise application of telishment would promote more good consequences than just punishment, but Rawls thinks that an ideal legislature still will not enact a rule allowing for punishment + telishment: “Once one realizes that one is involved in setting up an institution, one sees that the hazards [of telishment] are very great” for, basically, two reasons: (1) the risk that judges telishing under conditions of secrecy will abuse their discretion is too high, and (2) telishment will undermine the legitimacy of punishment in the eyes of the citizenry.
Boonin disagrees. He challenges the first lemma — that the ideal punishment rule authorizes only punishment because it does not authorize punishment + telishment — in two ways.
First, he imagines a practice superior to telishment, which he calls “u-telishment.” U-telishment authorizes punishing the innocent in cases where the deterrent effect is likely to be greater than the cases where telishment applies (e.g. vicarious punishment of children for crimes committed on their behalf by parents) and which has an effective set of procedural safeguards cabining the judges’ discretion (e.g. high standards of proof, judicial review, a culture of professionalism among the judiciary, etc.). Boonin argues that, when we focus on these reasonably implementable technologies, we see that the ideal punishment rule authorizes punishment + u-telishment.
Second, he argues that even if the ideal punishment rule only authorizes punishment, this fact is contingent and so not the kind of reason for which we are properly inclined to reject u-punishment. (That is: we think it is wrong and regrettable to punish the innocent, but the consequentialist thinks it is a shame that we do not have the ability to effectively punish the innocent in cases where it will promote the best results overall.)
The Second Lemma
The second lemma is that the ideal judge will follow the ideal punishment rule even when, in the particular case, more good consequences will flow from violating it. This lemma has to overcome what Boonin terms, following others, the rule worship problem: “if one now finds that one can produce more overall good by breaking a rule that it made the most sense on utilitarian grounds to adopt at an earlier point in time, then it should seem clear that a utilitarian must urge you now to break the rule even though accepting the rule was initially justified on good utilitarian grounds” (Boonin, 70).*
Rawls’s response, which is also a defense of the second lemma, is that the rule worship problem is only a problem on one conception of what a “rule” is. If a rule is “a useful generalization of lessons that have been learned from the past” such that it “gains its authority from the fact that it correctly summarizes the consequences of [violating the rule] in the past” then the rule worship problem defeats rule-consequentialism; “[i]f you come across a new situation in which the generalization embodied in the rule is not true, then the fact that following the rule was the right thing to do in previous situations will provide no reason for you not to break the rule in this new situation” (Boonin, 70-1).
But there are two conceptions of rules, and the rule worship problem does not threaten the other conception. On the second “practice conception,” rules define a practice (such as punishment), which is, in turn, set up for much the same reasons that the ideal punishment rule is supposed to authorize only punishment as opposed to punishment + telishment: “in many areas of conduct each person’s deciding what to do on utilitarian grounds case by case leads to confusion . . . [a]s an alternative one realizes that what is required is the establishment of a practice” (Boonin, 71; quoting Rawls). But a practice, argues Rawls, “necessarily involves the abdication of full liberty to act on utilitarian and prudential grounds” (Boonin, 71).
A practice necessarily involves the abdication of full liberty because there is no way to do an act that is defined by a practice “except by following the rules which define it.” To illustrate, Rawls gives the example of running the bases in baseball. If, all by yourself, you hit a ball and run around a baseball diamond you have not performed the action of scoring a run even though you have done the same exact thing that would count as scoring a run if you were in the midst of a baseball game. The action is defined by the practice; “the rule is logically prior to the act” (Boonin, 71-2).
The same holds, Rawls argues, for punishment. Punishment is a practice defined by rules. Somebody engaged in a practice “cannot say of his action . . . that he does it rather than some other because he thinks it is best on the whole” for “it is only against the stage-setting of the practice that one’s particular action is described as it is.” Put another way, a sentencing judge has no choice but to punish by following the rules of punishment because an action not in accordance with the rules of punishment is not punishment at all. (Similarly, a baseball player has no choice but to score a run by following the rules of the game because running around the bases without following those rules is not scoring a run at all.)
One thing to notice about this (pretty shitty) argument is that, if successful, it shows that a sentencing judge cannot do anything but act in accordance with the rule defining the practice of punishment irrespective of whether that rule is justified on rule-consequentialist grounds. The argument is simply that somebody engaged in a practice cannot, as a matter of necessity, perform the acts defined by the practice without following the rules of the practice. Thus, if the rules of punishment in a society call for harsher punishments for black people, it would appear that, by Rawls’s lights, a sentencing judge must impose harsher punishments on black people; it is not possible to impose a more lenient punishment on a black person because that is not an act authorized by the relevant practice.
Perhaps the idea is that a judge should decide to engage in a practice because it is justified on consequentialist grounds (it is the best possible practice). The judge could decide not to engage in the practice — he could use his office to disobey the rules of punishment (and thus, by Rawls’s lights, to not punish at all**) — but once he is engaged in the practice, the judge can only punish in accordance with it. In short: consequentialism explains why the judge should engage in the practice defined by the ideal punishment rule; the practice conception of rules explains why the judge can only punish in accordance with the ideal punishment rule. Upshot: the judge should should only punish in accordance with the ideal punishment rule.
The problem with this reply, as Boonin argues, is that it replaces rule worship with practice worship. It is simply false that consequentialism explains why the judge should engage in the practice defined by the ideal punishment rule. Consequentialism entails that the judge should sometimes engage in the practice defined by the ideal punishment rule — when doing so promotes the best consequences overall (including, presumably, by ensuring that the practice is not undermined) — but when punishing an innocent person will in fact promote the best results overall then the judge should disengage from the practice and punish the innocent person. Rawls’s response — that it is impossible for the judge to punish the innocent person precisely because the ideal punishment rule defines the practice of punishment — can either be denied or dismissed. It can be denied by rejecting Rawls’s wooden conception of how a practice such as punishment is defined. It can be dismissed, as Boonin observes, by conceding that the judge who “punishes” the innocent person is not “punishing” but (say) “telishing.” But telishing is still justified by the considerations that Rawls helps himself to in crafting the rule-consequentialist solution to the problem of punishment (namely: the idea that actions are to be evaluated by their consequences), and so Rawls’s putative justification for legal punishment justifies more than legal punishment: it justifies telishing the innocent.
Note that, if successful, Boonin’s reply to the Second Lemma amounts to a confutation of rule-consequentialism more generally. Note, also, that Boonin presupposes that a judge can know (or, at least, justifiably believe) that his violating the practice at a particular time will promote better results than his following it. This presupposition is at odds with a mood around here.
* The rule worship problem is thus an argument that rule-consequentialism is self-defeating.
** It is doubtful that the rules defining the practice of punishment are such that a judge’s violating them e.g. in sending an innocent man to jail would not count as an act of punishment at all. Punishment, like most emergent practices, is a family-resemblance concept and so there is no one set of necessary conditions for an act to count as an act of punishment. Because it is harder to argue about family-resemblance concepts — they are vague — philosophers of a certain stripe (Boonin’s) tend not to like them.