1. There are cases in which deliberately framing and punishing an innocent person would produce more long-term utility overall than would any available alternative, even when taking into account all of the potential risks involved in performing such an act.

2.  Act utilitarianism entails that, in such cases, it is morally permissible (indeed, obligatory) to deliberately punish an innocent person.

3. This implication is unacceptable.

In short, assume, at least for the sake of the argument, that the punishing the innocent objection is a decisive objection to the act-utilitarian solution to the problem of punishment.

Rawls’s argument: Rawls provides an argument that attempts to demonstrate that a rule-utilitarian approach to punishment can overcome the punishing the innocent objection.  The argument can be understood as consisting of two claims:

1. the rule content claim: from a utilitarian point of view, the best choice for the legislator to make is to enact a rule which prohibits judges from deliberately punishing innocent people.  The enactment of such a rule will have better long-term consequences than the enactment of rules which would sometimes allow judges to deliberately punish innocent people.

2. the rule following claim: from a utilitarian point of view, the fact that the best rule for legislators to choose would prohibit judges from deliberately punishing innocent people means that judges should always follow this rule and thus never deliberately punish innocent people.

If both of these claims are true, then the rule utilitarian can overcome the punishing the innocent objection.

Rawls’s defense of the rule content claim: Rawls’s argument in defense of the first claim involves comparing the institution of punishment, on which innocent people are never punished, with an institution he calls “telishment” on which “the officials set up by it have authority to arrange a trial for the condemnation of an innocent man whenever they are of the opinion that doing so would be in the best interests of society.  The discretion of officials is limited, however, by the rule that they may not condemn an innocent man to undergo such an ordeal unless there is, at the time, a wave of offenses similar to that with which they charge him and telish him for” (1955: 11).  He then argues that the long-term consequences of having a system of telishment would be worse than those of having punishment because of the various risks involved in giving judges such discretion, trying to keep such secrets, etc.

First objection: even if Rawls's claim is true of our society, we can imagine at least some circumstances in which a society would produce better long-term results by selecting telishment over punishment: a society, for example, in which it is very easy to keep such secrets, in which public officials feel such a strong sense of honor that they never abuse their public offices, etc.  Rawls’s argument entails that in this sort of society, the right thing to do would be to select telishment over punishment.  If we believe that even in such a society it would still be immoral to deliberately punish innocent people, then this will count as an objection to Rawls’s argument.

Second objection: even if we confine ourselves to the world as it is, Rawls’s argument fails because it ignores further options.  In particular, Rawls’s argument forces the legislator to choose between only these two possible rules (presented here as addressed to the judges)

(1) you may never punish an innocent person
(2) you may punish an innocent person in every case where you believe that doing so will best serve the interest of society

But this is too limited a choice.  In particular, it neglects other rules that would also allow judges at times to deliberately punish innocent people, but which would place stronger restrictions on when they may do so.  For example:

(3) you may never punish an innocent person except in cases where it is clear to you beyond a reasonable doubt that doing so will best serve the interest of society

A case can be made for preferring (3) over both (1) and (2) by considering other cases (e.g., giving soldiers discretion to disobey orders) where we think that some giving some discretion to break a rule is better than either an exceptionless rule or full discretion to break it every time it seems useful to do so.

Rawls’s defense of the rule following claim: Rawls’s argument here is designed to overcome the rule worship objection.  The rule worship objection maintains that even if it is right to select an exceptionless rule on consequentialist grounds, it will later be right to break that rule on consequentialist grounds, in those cases where breaking the rule produces better overall consequences than the alternative.  Rawls’s argument here turns on what he calls the “practice conception” of rules.  On this account, “rules are pictured as defining a practice.  Practices are set up for various reasons, but one of them is that in many areas of conduct each person’s deciding what to do on utilitarian grounds case by case leads to confusion, and that the attempt to coordinate behavior by trying to foresee how others will act is bound to fail. As an alternative one realizes that what is required is the establishment of a practice, the specification of a new form of activity; and from this one sees that a practice necessarily involves the abdication of full liberty to act on utilitaran and prudential grounds” (1955: 24).  That is, since the need for the practice arises precisely because of problems that are caused by people making decisions directly on a utilitarian basis, it follows that the nature of the practice that will satisfy this need will be one that itself prevents people from reasoning in this way.

Rawls then makes two important points about rules so understood.  The first is that, on this conception of rules, the “practice is logically prior to particular cases: unless there is the practice the terms referring to actions specificed by it lack a sense” (25).  Rawls illustrates this point by appealing to the case of baseball.  A person standing on a baseball diamond, for example, can run from the third base to home plate and step on the plate.  He can do this all by himself.  But this, in itself, is not the same thing as scoring a run.  An act, that is, can be the act of simply running around the field, and it can be this fully independently of any rules determining the content of a game.  But an act simply cannot be an act of  “scoring a run” unless there is already a set of rules that determines what counts as scoring a run.

The second point follows from the first: “If one wants to do an action which a certain practice specifices then there is no way to do it except to follow the rules which define it.  Therefore, it doesn’t make sense for a person to raise the question whether or not a rule of a practice correctly applies to his case where the action he contemplates is a form of action defined by a practice” (26).  For example: it is a rule of baseball that you can’t step outside the path in trying to get from one base to another.  If you do this, you are called out.  And so, on the conception of rules that Rawls is advancing, it would make no sense for you to say “but in this case wouldn’t it be better to run outside the lines”.  The only way for you to do an act that really counts as the act of scoring a run is to do the act in accordance with the rule.  This, again, is because the rule is logically prior to the act.  The act, understood as a real baseball act, could not exist without it.

This second point has important implications: “if a person is engaged in a practice [e.g., the judge acting qua judge], and if he is asked why he does what he does, or if he is asked to defend what he does, then his explanation, or defense, lies in referring the questioner to the practice.  He cannot say of his action, if it is an action specified by a practice, that he does it rather than some other because he thinks it is best on the whole. ..... One doesn’t so much justify one’s particular action as explain, or show, that it is in accordance with the practice.  The reason for this is that it is only against the stage-setting of the practice that one’s particular action is described as it is.  Only by reference to the practice can on say what one is doing.  To explain or to defend one’s own action, as a particular action, one fits it into the practice which defines it” (1955: 27).

And this, in turn, is meant to overcome the rule worship problem: “On the practice conception, if one holds an office defined by a pracitce then questions regarding one’s actions in this office are settled by reference to the rules which define the practice” (28).  The result is this: suppose you are a judge and the rules accepted on consequentialist grounds are punishment rather than telishment.  You have before you a particular innocent man whom it would be very useful nonetheless to send to jail.  You set him free.  You are asked why.  Your answer must be because you are a judge and the acts of a judge can only be assessed from within this framework.  And so, according to Rawls, the rule worship problem can be overcome.

Objection: Rawls’s argument does seem to establish that the judge, qua judge, lacks the authority to subject his decisions directly to a utilitarian standard of evaluation.  The judge, in short, should correctly note that his office does not give him the authority to punish innocents when this benefits society.  He has the legal authority to sentence people, but if he wants to do an act that really counts as an act of sentencing , then he must follow the rules.  Otherwise, his act simply can’t be an act of sentencing.  What makes acts acts of sentencing in the first place is precisely their taking place within the practice defined by the rules which, in this case, prohibit the deliberate punishment of the innocent. Just as if the baseball playing wants to do an act of running the bases, then he must stay within the lines.

But then why not say this: the judge in such a position should decide not to do an act of sentencing but should instead do an act of having an innocent person locked up in the sort of facility that guilty people are locked up in when they are sentenced.  We can't call this sentencing, but could call it, for example, "tenencing".   He wouldn’t be acting as a good judge, but he would be acting as a good person.  After all, suppose that you are on third base trying to get to home plate.  As you start to run, you see a young child in the front of the stands, choking on a hot dog.  No one around him seems to know what to do, but you know the Heimlich maneuver.  So, you run out of the path and save his life.  Clearly, qua ballplayer, you did the wrong thing.  But it would be crazy to say that, all things considered, you did the wrong thing.  Your act can only be described as the act of abandoning the practice, but it still seems the right thing to do.