Skip to main content Accessibility help

On the Common Saying that it is Better that Ten Guilty Persons Escape than that One Innocent Suffer: Pro and Con

  • Jeffrey Reiman (a1) and Ernest Van Den Haag


In Zadig, published in 1748, Voltaire wrote of “the great principle that it is better to run the risk of sparing the guilty than to condemn the innocent.” At about the same time, Blackstone noted approvingly that “the law holds that it is better that ten guilty persons escape, than that one innocent suffer.” In 1824, Thomas Fielding cited the principle as an Italian proverb and a maxim of English law. John Stuart Mill endorsed it in an address to Parliament in 1868. General acceptance of this maxim continues into our own period, yet it is difficult (and for us so far, impossible) to find systematic attempts to defend the maxim. It is treated as a truism in no need of defense. But the principle within it is not at all obvious; and since it undergirds many of our criminal justice policies, we should be sure that it is justifiable. First, however, we must clarify what the principle means.



Hide All

1 Marie Arouet de Voltaire, Jean Francois, Zadig, or Fate, in Candide and Other Stories London: Dent & Sons, 1962), p. 20.

2 William Blackstone, Commentaries on the Laws of England, 21st ed. (1765; London: Sweet, Maxwell, Stevens & Norton, 1844), bk. IV, ch. 27, p. 358. In his History of the Pleas of the Crown (1694), Sir Matthew Hale put the ratio at 5:1; Fletcher, George (Yale Law Journal 78 [1968], p. 881) uses 20:1. There has been some inflation since Sir Matthew.

3 “Thomas Fielding, Proverbs of All Nations, p. 59 (1824). Citing an Italian proverb, which is also a maxim of English law, Fielding says it originated in Italy, and that Dr. Paley was against it, while Blackstone and Romilly approved of it” (The Macmillan Book of Proverbs, Maxims, and Famous Phrases, ed. Stevenson, B. [New York: Macmillan, 1948], p. 1249).

4 Hansard, “Parliamentary Debate on Capital Punishment Within Prisons Bill,” April 21, 1868.

5 For example, in the widely-read essay “Persons and Punishment,” Herbert Morris describes an ideal criminal justice system (not far from an idealized version of our own) as one in which it is thought “a greater evil for society to interfere unjustifiably with an individual by depriving him of some good than for the society to fail to punish those that have unjustifiably interfered” (Morris, , “Persons and Punishment,” The Monist 52, no. 4 [October 1968], p. 479).

6 Kant, Immanuel, The Metaphysical Elements of Justice, Part I of the Metaphysics of Morals, tr. Ladd, J. 1797; (Indianapolis: Bobbs-Merrill, 1965), p. 102.

7 Hegel, G.W.F., The Philosophy of Right, tr. Knox, T.M. (1821; Oxford: Clarendon Press, 1962), p. 69.

8 “It should be noted that deontological theories are defined as non-teleological ones, not as views that characterize the Tightness of institutions and acts independently from their consequences. All ethical doctrines worth our attention take consequences into account in judging lightness. One which did not would simply be irrational, crazy” (Rawls, John, A Theory of Justice [Cambridge: Harvard University Press, 1971], p. 30).

9 One exception to this might be a theological deontological ethics that took its duties strictly from God's commands. We can exclude this possibility since, for reasons mentioned in the previous section, we are limited to secular considerations.

10 It is because Kant did not take this step that his theory is pilloried for not accounting for the permissibility of lying to prevent murder. But there is nothing in Kant's theory that rules this step out Kant's test of universalization of maxims can be applied to maxims that state priority relations between actions just as easily as it can be applied to maxims that refer directly to specific actions. That I cannot universalize the maxim that one may lie when it is convenient does not imply that I cannot universalize the maxim that one may lie when (and only when) necessary to prevent some other immoral (i.e., nonuniversalizable) action, such as murder.

11 Though I do not want to assert generally that deontological theories view duties as owed to specific persons, there is nothing non-deontological about treating duties this way. For example, retributivists often hold that the victim of an offense has the right to pardon his offender and thus spare him from the punishment he deserves. (See, for example, Kant, , Metaphysical Elements of Justice, pp. 107–8.) This makes sense on the assumption that the duty to punish offenders is owed to victims. This is Hegel's view, too. See Philosophy of Rights, tr. Knox, T. (London: Oxford University Press, 1969), sec. 100.

12 For example, see Morris, , “Persons and Punishment,” pp. 475–76.

13 The legal action normally taken against a convicted defendent will be called punishment, regardless of whether he actually is guilty or not.

14 Although conceding that conviction of the guilty is socially more important, one may hold that acquittal of the innocent remains morally more important. Some philosophies find a difference between what is morally right and what is socially important. However, for a policymaker, this may be a distinction without a difference.

15 So does what is admissible as proof and what is not, although the criterion of admissibility is not designed for the quantitative result. (Standards of admissibility are ignored here, for most of the effects do not differ from those of standards of proof.)

16 Although Reiman explicitly disavows “claiming that deontological duties are required because owed to these parties,” unless the duties are indeed owed to these groups, there would be no point in “comparing the losses to those parties of failures in those duties” and inferring, as Reiman does, that if failure in the duty not to convict innocents causes a greater loss to them than failure to convict the guilty does, it is morally worse to punish innocents than not to punish the guilty.

17 Reiman, Jeffrey and I made our disagreement on this point explicit in “Justice, Civilization, and the Death Penalty: Answering van den Haag,” and “Refuting Reiman and Nathanson,” Philosophy & Public Affairs 14, no. 2 (Spring 1985), pp. 115–48 and 165–76.

18 Historically, retributivism (despite Blackstone) has stressed punishing the guilty more than acquitting innocents.

19 Reiman's distinction between “immediate” and other consequences is too inchoate. Is the distinction temporal? Spatial? Does it depend on the density of relationships? The vagueness of the distinction would make it hard to distinguish retributivism from utilitarianism (which considers non-immediate consequences as well). Fortunately, that distinction does not actually depend on distinguishing immediate consequences from other ones. Suppose by executing Roger we save a hundred other innocents. Utilitarians certainly would. Deontologists would not, but not because the death of Roger is more immediate than the death of the hundred. They would oppose the swap because of the categorical nature of the prohibition against intentionally killing innocents, and because of their refusal to aggregate or offset. Although deontologists recognize more or less important duties (and, thereby, priorities), they do not justify failure to carry out a duty for the sake of some beneficial effect. Kant did not believe that the prevention of a murder could justify lying. Reiman may be right: Kant's view is implausible, an unintentional reductto ad absurdum. Yet, however good an argument against retributivism, its implausibility does not justify Reiman's addition of utilitarian reasoning while keeping the label of retributivism.

20 I agree with Reiman that the guilty will favor the punishment constraint. The reduction of the probability of punishment would not disturb them, even though it should, according to Kant and Hegel. If the obligation to punish is to them, they would release us from it. The beneficiary can release the makers from a promise – although not from a threat. If punishment were owed to criminals, it would be a promise, not a threat. But, to reiterate, for retributionists punishment is primarily neither a promise, nor a threat, but an obligation to do justice – pereat mundus.

21 For discussion of that ingenious calculus, see “Utilitarianism and the Punishment Constraint,” below.

22 This objection is far more limited (to unqualified utilitarianism rather than general consequentialism) than is usually realized. See my “Punishment as a Device for Controlling the Crime Rate,” 33 Rutgers Law Review 3 (Spring 1981).

23 I deal below with Reiman's belief that innocents, qua innocents, suffer more when punished.

24 The effects of indiscriminate political terrorism in discouraging opposition, defined as crime, are beyond the scope of this discussion (and of Bentham's).

25 Low standards need not, but standards of proof perceived to be irrational will erode the necessary belief that the guilty are sought out.

26 Bentham's idea that innocence unavoidably will become known applies to innocents deliberately convicted, when their innocence was known to the authorities. It does not apply to innocents convicted in good faith by clear and convincing evidence. Indeed, their conviction may well add to deterrence.

27 Similarly, “capacity to deter” does not depend on jury unanimity, or on the number of jurors, or finally on constitutional or common law exclusionary rules, some of which are quite irrational.

28 The malum poenae, St. Thomas thought, would, through suffering, induce repentance and thus ultimately annul the malum culpae (sin). He believed that guilt adds to the suffering imposed by punishment. Surely the point of the penitentiary, as it used to be called not so long ago, was to redeem the guilty by their psychological suffering – from which the innocent being imprisoned would be exempt, even though they suffered the same material deprivations.

29 Even though they are found guilty beyond a reasonable doubt, most imprisoned convicts persuade themselves that they are blameless. They are outraged by their punishment and full of rage against “the system” that inflicted it. It seems that blameworthy people are rarely able to accept blame, whereas there are many people outside prisons willing to accept blame for almost anything, including the punishment of convicts – which may upset some non-convicts more than the crimes the convicts were found guilty of.

30 The Bayesian probability of guilt is no less than it is with a more restrictive standard for defendants without relevant prior convictions.

31 The primary purpose of trials should be, once more, to determine guilt or innocence. Secondary purposes, such as disciplining the police, however important, should not be allowed to interfere. Hence, all relevant evidence should always be admitted. (Further discussion of the curious trial rules which have accumulated is beyond my present scope.)

32 The converse conclusion might appear as logical: these are the crimes we want to deter most and should most readily punish. However, they also are exceptional enough to enable us to take special precautions to avoid convicting innocents without much fear of reducing deterrence.


Altmetric attention score

Full text views

Total number of HTML views: 0
Total number of PDF views: 0 *
Loading metrics...

Abstract views

Total abstract views: 0 *
Loading metrics...

* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.

Usage data cannot currently be displayed