Criticism of Combined Views

Goldman believes that mixed theories of punishment like Hart’s are internally contradictory. Mixed theories hold that there are two necessary conditions that must be satisfied for punishment to be justified:

  1. Punishment cannot exceed what Goldman calls the equivalence limitation.
  2. Punishment must have good future effects, such as deterring crime.

Goldman argues that these conditions cannot both be met because the level of punishment that would deter crime exceeds the equivalence limitation.

What is the paradox, exactly?

One way of taking Goldman’s point is that it shows that punishment cannot be justified: there are two necessary conditions for punishment to be justified, those conditions cannot both be met, therefore, punishment cannot be justified.

However, I think he has something more disturbing on his mind.

  1. The state must do what is necessary to protect the public.
  2. Punishment at a level that deters crime is necessary to protect the public.
  3. Therefore, the state must punish at a level that deters crime.
  4. Punishment must not exceed the equivalence level of harm.
  5. Punishment deters crime only when it exceeds the equivalence level of harm.
  6. Therefore, the state must not punish at a level that deters crime.

That’s a real paradox: the state is both required to punish and required not to punish. Zounds!

The equivalence limitation

Criminals forfeit their rights. That is why society is permitted to do things to them that it would otherwise not be permitted to do, like locking them up. Which rights do criminals forfeit? Goldman claims they forfeit rights that are equivalent to those that they violated when they committed their crimes.

if we ask which rights are forfeited in violating rights of others, it is plausible to answer just those rights that one violates (or an equivalent set). One continues to enjoy rights only as long as one respects those rights in others: violation constitutes forfeiture. But one retains those rights which one has continued to respect in others. Since deprivation of those particular rights violated is often impracticable, we are justified in depriving a wrongdoer of some equivalent set, or in inflicting harm equivalent to that which would be suffered in losing those same rights (for example, rights to fifty dollars of one's own and not to suffer the trauma of being a victim of theft). (Goldman 1979, 46)

The innocent forfeit no rights, so punishing them at any level is unjust. The guilty forfeit some rights. Petty thieves forfeit less important rights than murderers do. Punishing a petty thief like a murderer violates the thief’s rights, much as punishing an innocent person would. That is the idea behind the equivalence limitation (Goldman 1979, 51).

As Noah noted, Goldman’s way of thinking about punishment puts excessive punishment on the same scale as punishing the innocent. It’s very clever, I think.

That said, Antonio and I were suspicious of the idea that criminals should be treated equally with their victims. Criminals are guilty and victims are innocent. So why isn’t it OK to treat the criminals worse than the victims are treated? Of course, as Peter pointed out, the state is treating the criminals worse: it does not do anything at all to the victims, but it is harming the criminals by punishing them. That is treating them worse! I still have the feeling that it should be OK if the criminals wind up worse off than their victims did, but I do not have a snappy reply to Peter’s point. Maybe one of you can come up with one.

Jerry asked whether we should measure equivalence on an objective scale or a subjective one. For instance, should the penalty for a property crime depend on how wealthy the victim and criminal are? A poor victim will miss the money a lot more than a rich victim will; conversely, a poor criminal will suffer more from a fine than a rich one will. Goldman’s own response to Jerry’s point is that the measure of equivalence will be objective. Two losses are equivalent, according to him, if an average person would be indifferent between suffering one rather than the other.

One right or set of rights is equivalent to another for these purposes when an average preference scale registers indifference between the loss of either the one or the other. There are problems facing construction of the proper preference scale—for example, the loss of fifty dollars to which one has a right will mean more to a poor person than to a rich person—but these are problems facing any moral theory concerned with distribution. We also need to adjust our concept of deserved punishment to focus upon intention rather than actual harm, and to allow for excuses. I leave these complications to pursue our main topic. (Goldman 1979, 46)

I think that Goldman wishes to postpone settling this for another time. This article is about showing that the idea has merit; if it succeeds, we can come back to iron out the details. You need to be careful with that sort of move, as the details sometimes matter more than an author cares to admit. That said, this strikes me as a reasonable point.

Finally, Austin has been reminding us that there are lots of costs to being convicted of a crime in addition to the official sentence. For instance, it is very difficult for someone convicted of a crime to get a job, both because of the person’s criminal record and also because time that could have been spent developing economically useful experience was lost. Should we add these informal costs to the formal sentence when determining whether a given punishment meets the equivalence limitation? If we leave them out, we might feel a little bit better about deterrence on the grounds that the punishments that formally respect the equivalence limitation are really more costly to the criminal than they seem to be.


Patrick had a dandy way of explaining why punishments that respect the equivalence limitation will not deter crime. (I haven’t been able to make the point this well in over two years of trying: I bow to Patrick here.)

Suppose that Bob steals $10 each from 10 people. He has $100 more than he did before committing his crimes. Now, suppose Bob is convicted for two of these thefts. He has to give the money back, so he loses $20 of out the $100. That leaves him $80 ahead. Then he is punished, meaning he has to suffer an equivalent loss to the one he imposed on others. That is another $20. So he loses $40 out of the $100 he stole. But that still leaves him $60 ahead. Crime pays in this case.

Maybe the penalty should exceed the cash loss. As Abby pointed out, being the victim of a crime is always irritating and sometimes deeply upsetting. That should count and so the punishment should probably be higher than $20 (plus the returned $20, of course). But I think the point remains the same. So long as Bob racks up lots of other small crimes, he is going to come out ahead.

We could punish him at greater than the equivalence level or we could punish him for the crimes that we suspect he has committed even if we cannot prove it. And that is Goldman’s point: we cannot both deter crime and respect the equivalence limitation at the same time.

Key concepts

  1. The paradox of punishment, according to Goldman.
  2. The equivalence limitation on punishment.


I like to make outlines identifying the broad chunks in an article. It helps to keep me oriented. Here’s how I see the Goldman article breaking down (page numbers in parentheses).

  1. Target: views that hold punishment must be both deserved and beneficial. (42)

  2. The version of retributivism that he likes: criminals may be punished because they forfeit their rights against being harmed. The most important part of this is what he calls the equivalence limitation. It holds that the rights criminals forfeit is equivalent to the rights violated by their crimes. It follows that punishment cannot inflict harm on the criminal that exceeds the harm the criminal inflicted on others (see p. 46). (43–47).

  3. The paradox. In order to deter crime, we have to punish in excess of what is permitted by the equivalence limitation: we have to inflict more harm on the criminal than the criminal inflicted on others. This is the central part of his argument and it is the most important section. (47–49)

    Everything else in the article will be concerned with objections to this part. The objections will either try to show that the level of punishment needed to deter crime is acceptable (meaning the equivalence limitation is wrong) or that it is possible to have deterrence in a system of punishment that respects the equivalence limitation. Goldman answers each one as it comes up.

  4. Objection: punishment beyond the equivalence limitation is OK if it is needed for deterrence, preventing the violation of other people’s rights. (52–54)

  5. Objection: the guilty voluntarily accept punishment. That means it doesn’t have to be equivalent. (54–55)

  6. Objection: excessive punishment is the lesser of two evils, where the other evil would be vigilante justice. (56–57)

  7. Alternative: increase the probability of punishment. That would deter crime even if the punishment itself satisfied the equivalence limitation. (58)

  8. Alternative: improve social conditions so there is less incentive for crime. (58)


Goldman, Alan H. 1979. “The Paradox of Punishment.” Philosophy & Public Affairs 9 (1): 42–58.