In his recent book, No Future Without Forgiveness, Archbishop Desmond Tutu evaluates the successes and failures of the South African Truth and Reconciliation Commission (TRC). The chair of the TRC, Tutu defends the Commission’s granting of amnesty to wrongdoers who revealed the truth about their pasts, and he lauds those victims who forgave their abusers. While recognizing that a country must reckon with its past evils rather than adopt "national amnesia," Tutu nevertheless rejects what he calls the "Nuremberg trial paradigm." He believes that victims should not press charges against those who violated their rights, and the state should not make the accused "run the gauntlet of the normal judicial process " and impose punishment on those found guilty. 

Tutu does at times concede that trials have two other aims, at least during South Africa’s transition: vindicating the rights of victims and generating truth about the past. Again and again, Tutu states that victims of past wrongs have the right—at least a constitutional right and perhaps also a moral one — to press criminal charges against and seek restitution from those that abused them. He also extols the "magnanimity" of individuals who, like former South African President Nelson Mandela, have not exercised this right but are willing to forgive and seek harmony (ubuntu) with their oppressors. These statements suggest that Tutu regards legal punishment not merely as a means to retribution but also as a way to affirm and promote the rights of victims.

But Tutu cannot have it both ways. He cannot both reject actual punishment and still defend the threat of punishment as efficacious in dispelling lies and generating truth. Hence, Tutu’s acceptance of a "threat to punish" practically commits him to a nonretributive and consequentialist role for punishment, since without occasionally making good on the threat to punish, such a threat loses credibility.

Tutu does not bring enough precision to the term "retribution." He seems, at points, simply to identify retribution with legal punishment. Instead, one must understand retribution as one important rationale or justification for and a constraint upon punishment.  Proponents of the retributive theory of punishment offer a variety of competing accounts, but all agree that any retributive theory minimally requires that punishment must be "backward looking in important respects." That is, justice requires that a crime is punishable as, in the words of lawyer and legal theorist Lawrence Crocker, "a matter of the criminal act, not the future consequences of conviction and punishment."  These future consequences might comprise such good things as deterrence of crime, rehabilitation of criminals, or promotion of reconciliation. For the proponent of retributivism, however, the infliction of suffering or harm, something normally prohibited, is justified because of — and in proportion to — what the criminal has done antecedently. Only those found guilty should be punished, and their punishment should fit (but be no more than) their crime.

Some supporters of the retributive theory of punishment, assert, moreover, that only (and perhaps all) wrongdoers deserve punishment, and the amount or kind of punishment they deserve must fit the wrong done. Harvard philosopher Robert Nozick explains desert in terms of both the degree of wrongness of the act and the criminal’s degree of responsibility for it.   Retribution as a justification for punishment requires that wrongdoers should get no more than (and perhaps no less than) their "just deserts."

The second premise in Tutu’s argument against punishment — that retribution is (nothing but) vengeance or revenge — is flawed as well. Given Nozick’s understanding of retribution as "punishment inflicted as deserved for a past wrong," is Tutu right to treat retribution and revenge or vengeance as equivalent? Both retribution and revenge share, as Nozick puts it, "a common structure." They inflict harm or deprivation for a reason. Retribution and vengeance harm those who in some sense have it coming to them. Following Nozick’s brief but suggestive analysis, I propose that there are at least six ways in which retribution differs from revenge.

Retribution addresses a wrong.  First, as Nozick observes, "retribution is done for a wrong, while revenge may be done for an injury or slight and need not be done for a wrong." I interpret Nozick to mean retribution metes out punishment for a crime or other wrongdoing while revenge may be exacted for what is merely a slight, an unintended injury, an innocent gaze, or shaming in front of one’s friends.

Retribution is constrained. Second, Nozick also correctly sees that in retribution there exists some "internal" upper limit to punishment while revenge is essentially unlimited. Lawrence Crocker concurs: "an absolutely central feature of criminal justice" is to place on each offense "an upper limit on the severity of just punishment." This limitation "is the soul of retributive justice." It is morally repugnant to punish the reluctant foot soldier as severely as the architects, chief implementers, or "middle management" of atrocities. Retribution provides both a sword to punish wrongdoers and a shield to protect them from more punishment than they deserve. In contrast to punishment, revenge is wild, "insatiable," unlimited. After killing his victims, an agent of revenge may mutilate them and incinerate their houses. As Nozick observes, if the avenger does restrain himself, it is done for "external" reasons having nothing to do with the rights or dignity of his victims. His rampage may cease, for instance, because he tires, runs out of victims, or intends to exact further vengeance the next day. 

While Minow’s view deserves serious consideration, Nozick, I think, gives us a picture of vengeance—and its fundamental difference from retribution — that better matches our experience. Precisely because the agent of revenge is insatiable, limited neither by prudence nor by what the wrongdoer deserves, revenge is not something admirable that goes wrong. The person seeking revenge thirsts for injury that knows no (internal) bounds, has no principles to limit penalties.   Retribution, by contrast, seeks not to tame vengeance but to excise it altogether. Retribution insists that the response not be greater than the offense; vengeance insists that it be no less and if possible more. Minow attempts to navigate "between vengeance and forgiveness," but she does so in a way that makes too many concessions to vengeance. She fails to see unequivocally that retribution has essential limits. Vengeance has no place in the courtroom or, in fact, in any venue, public or private.

Retribution is impersonal. Third, vengeance is personal in the sense that the avenger retaliates for something done antecedently to her or her group. In contrast, as Nozick notes, "the agent of retribution need have no special or personal tie to the victim of the wrong for which he exacts retribution." Retribution demands impartiality and rejects personal bias while partiality and personal animus motivate the "thirst for revenge."

The figure of Justice blindfolded (so as to remove any prejudicial relation to the perpetrator or victim) embodies the commonplace that justice requires impartiality. Justice is blind — that is, impartial — in the sense that she cannot distinguish between people on the basis of familiarity or personal ties. This not to say, however, that justice is impersonal in the sense that she neglects to consider an individual’s traits or conduct relevant to the case. Oddly, Tutu suggests that the impartiality or neutrality of the state detracts from its ability to deal with the crimes of apartheid. He defends the TRC because it is able to take personal factors into account. He writes:

Although justice eliminates bias from judicial proceedings, it may be fair only if it takes certain personal factors into account. Because Tutu confuses the impersonality or neutrality of the law with an indifference to the personal or unique aspects of a case, Tutu insists that judicial processes and penalties give little regard to "real victims" or their oppressors.

Retribution takes no satisfaction. A fourth distinction between retribution and revenge concerns the "emotional tone" that accompanies — or the feelings that motivate — the infliction of harm. Agents of revenge, claims Nozick, get pleasure, or we might say "satisfaction," from their victim’s suffering. Agents of retribution may either have no emotional response at all or take "pleasure at justice being done." (Adding to Nozick’s account and drawing on the work of political theorists Jeffrie Murphy and Jean Hampton, I should add that a "thirst for justice" may—but need not— arise from moral outrage over and hatred of wrongdoing.)

Retribution is principled. Fifth, Nozick claims that what he calls "generality" is essential to retribution but may be absent from revenge. By this term, Nozick means that agents of retribution who inflict deserved punishment for a wrong are "committed to (the existence of some) general principles (prima facie) mandating punishment in other similar circumstances."

Retribution rejects collective guilt. Nozick, I believe, helpfully captures the main contrasts between retribution and revenge. To these, I add a sixth distinction.  Mere membership in an opposing or offending group may be the occasion of revenge, but not of retribution. Retributive justice differs from vengeance, in other words, because it extends only to individuals and not to the groups to which they belong. In response to a real or perceived injury, members of one ethnic group might, for instance, take revenge on members of another ethnic group. However, the state or international criminal court could properly mete out retribution only to those individuals found guilty of rights abuses, not to all members of the offending ethnic group. Since collective guilt has no place in an understanding of retributive justice, revenge and retribution should not be conceived as equivalent. Tutu makes precisely this mistake.

Following the Hegelian dictum "first distinguish, then unite," Nozick promptly concedes, as he should, that vengeance and retribution can come together in various ways. Particular judicial and penal institutions may combine elements of retribution and of revenge. The Nuremberg trials, arguably, were retributive in finding guilty and punishing some Nazi leaders, punishing some more than others, and acquitting those whom it found not guilty as charged. But Tutu is right to say that the Nuremberg precedent was contaminated, compromised by revenge or "victor’s justice." As he notes, Nuremberg used exclusively allied judges and failed to put any allied officers in the dock. However, Tutu neglects to affirm the achievements of Nuremberg: it vindicated the notion of individual responsibility for crimes against humanity and defeated the excuse that one was "merely following orders." One reason that Nuremberg is an ambiguous legacy is that it had both good (retributive) and bad (vengeful) elements. In no case can one accept Tutu’s second premise that retribution is nothing but vengeance.

What of Tutu’s third premise that vengeance is morally wrong? When I shift the focus from vengeance to the agent of revenge, I accept Tutu’s premise. Unlike the agent of retribution, the agent of revenge does wrong, or at least he is morally blameworthy. He retaliates and inflicts an injury without regard to what the person impartially deserves. If the penalty happens to fit the crime, it is by luck; the agent of revenge is still blameworthy since he gave no consideration to desert, impartiality, or generality. If, as is more likely given the limitless nature of revenge, the penalty is more excessive than the crime, the agent of revenge is not only culpable but also his act is morally wrong. Nonetheless, Tutu’s overall argument against vengeance is unsound since two of its premises are not acceptable.

Tutu proposes a second moral argument against the "Nuremberg trial paradigm" for South Africa’s transition and others like it. Tutu rejects retributive justice on the grounds that it prevents or impedes reconciliation. He understands reconciliation as "restorative justice," the highest if not the only goal in South Africa’s reckoning with past wrongs. Tutu defends amnesty and forgiveness as the best means to promote reconciliation. What does Tutu mean by the vague and not infrequently contested term "reconciliation" and its synonym "restorative justice"? Tutu explicitly defines restorative justice (in contrast to retributive justice) as reconciliation of broken relationships between perpetrators and victims:

Although Tutu in this passage uncharacteristically leaves room for punishment, he understands the "central concern" of restorative justice as the reconciliation of the wrongdoer with his victim and with the society he has injured. The wrongdoing has "ruptured" earlier relationships or failed to realize the ideal of "ubuntu."  Ubuntu, a term from the Ngunui group of languages, refers to a kind of "social harmony" in which people are friendly, hospitable, magnanimous, compassionate, open, and nonenvious. Although Tutu recognizes the difficulty of translating the concept, it seems to combine the Western ideal of mutual beneficence, the disposition to be kind to others, with the ideal of community solidarity.

Tutu regards "social harmony" or "communal harmony" as the summum bonum, or highest good. He concedes that South Africa must in some way "balance" a plurality of important values – "justice, accountability, stability, peace, and reconciliation." Whatever "subverts" or corrodes social harmony, how-ever, "is to be avoided like the plague." Presumably, whatever maximizes social harmony is morally commendable and even obligatory.

Tutu may believe that ubuntu presents so lofty an ideal that no one would question its justification or importance. In any case, he offers little argument for its significance or supremacy. He does seek to support it by calling attention to its African origins. He also remarks that, while altruistic, ubuntu is also "the best form of self-interest," for each individual benefits when the community benefits.

As it stands, neither defense is persuasive. The moral disvalue of apartheid, also a South African concept, has nothing to do with its origins. Similarly, the geographical origin of ubuntu does not ensure its reasonableness. Further, although individuals often benefit from harmonious community relationships, the community also at times demands excessive sacrifices from individuals. Moreover, dissent or moral outrage may be justified even though it disrupts friendliness and social harmony.

Tutu offers practical objections—as well as moral ones—to seeking retributive justice against former oppressors. He does not consider the practicability of ubuntu, however, as a goal of social policy. He does not discuss, for example, what to do with those whose hearts cannot be purged of resentment or vengeance. Nor does he explain how society can test citizens for purity of mind and heart—how it can determine who has succeeded and who has failed to assist society toward this supreme good.

This analysis of alternative conceptions of reconciliation not only shows that Tutu’s ideal is unrealistic  but also that it pays insufficient attention to individual- freedom, including the freedom to withhold forgiveness. In making social harmony the supreme good, Tutu unfortunately subordinates—without argument—other important values, such as truth, compensation, democracy, and individual accountability. In some contexts, social harmony—if it respects personal freedom and democratic deliberation— should have priority. In other contexts, society may pursue other equally important values, for example, justice, which might require a society to indict, try, sentence, and punish individuals who violated human rights. If social harmony is judged to have priority over other values, that judgment should emerge not from a cultural, theological, or philosophical theory but from the deliberation and democratic determination of citizens.