Journal of Contemporary Legal Issues

Spring 1994

Harm v. Culpability:

Which Should Be the Organizing Principle of the Criminal Law?

*299 THE ROLE OF HARM AND EVIL IN CRIMINAL LAW: A STUDY IN LEGISLATIVE

DECEPTION?

Paul H. Robinson [FNa1]

What is the role of the occurrence of harm or evil in criminal law? What should it be? Answers to these questions commonly use the distinction between what is called an "objective" and a "subjective" view of criminality. [FN1] To oversimplify, the objective view maintains that the occurrence of the harm or evil defined by the offense is highly relevant. The subjectivist view maintains that such harm or evil is irrelevant; only the actor's culpable state of mind regarding the occurrence of the harm or evil is important.

The labels tend to overstate a rather subtle distinction. The "objectivist" or "harmful consequences" view is not so objective as to require that the harm or evil of the offense actually occur in order to impose liability. The objectivist imposes liability for inchoate conduct, as for example, when the actor comes close to bringing about a real offense harm or evil. [FN2] More on this in a moment. The "subjectivist" view, in turn, is not so subjective as to only require a culpable state of mind. An intention alone is insufficient for liability; some act is required to prove the actor's willingness to act upon, to externalize, his or her subjective culpability. [FN3] And, while the occurrence of the harm or evil may not be important to the subjectivist, the nature of the harm or evil intended or risked is important to determine the degree of the actor's culpability. Intending to cause death is more serious than intending to trip. The primary difference between the two views, then, is the role of resulting harm or evil. The objectivist thinks it central, the subjectivist thinks it marginal. Later in this paper, I discuss the nature of this difference in greater detail.

*300 The different perspectives also have been labelled as the difference between "traditionalists" and "modernists." [FN4] Common law is said to embody the "traditional" objective view of criminality, where the gravamen of an offense is its resulting harm or evil. The "modernist" subjective view of criminality, in contrast, focuses upon the actor's culpable state of mind toward bringing about the offense harm or evil, without regard for whether the harm or evil actually occurs. George Fletcher explains:

The traditionalists root their case in the way we feel about crime and suffering. Modernists hold to arguments of rational and meaningful punishment. Despite what we might feel, the modernist insists, reason demands that we limit the criminal law to those factors that are within the control of the actor. The occurrence of harm is beyond his control and therefore ought not to have weight in the definition of crime and fitting punishment. The tension between these conflicting schools infects virtually all of our decisions in designing a system of crime and punishment. [FN5]

As Fletcher conceives of it, then, a person is either a "traditionalist" or a "modernist." One either thinks that the occurrence of harm or evil is important to criminal liability or thinks that it is not. The distinction has been useful, I acknowledge. It has helped articulate the nature of the shift from common law to modern legislation. But, as currently expressed, it is a distinction that in many important respects obscures rather than sharpens the points of dispute in criminal law debate.

Let me first illustrate how useful the distinction has been in capturing the common thread of the differences between common law and modern liability rules. Then, I shall explain why the traditionalist-modernist distinction is misleading: it incorrectly suggests that one must be either an objectivist or a subjectivist on all issues. In fact, most American jurisdictions take a subjectivist view of the minimum requirements of liability but many simultaneously take an objectivist view of grading.

The remainder of the article looks more closely at the current objectivist- subjectivist dispute in grading, presenting the competing arguments and analyzing the most common statutory formulations. Ironically, while most jurisdictions are subjectivists as to the minimum requirements of liability, none are true subjectivists as to grading. Many might like to be subjectivist but instead have adopted a criminal code that, while subjectivist in rationale, is designed to look as if it is objectivist, and in many respects actually is. The article explains the logical reasons why subjectivist code drafters might take this somewhat peculiar approach of being partially objectivist and trying to appear even more objectivist than they actually are.

*301 I. OBJECTIVE "TRADITIONALISTS" VERSUS SUBJECTIVE "MODERNISTS"

Consider four examples of differences between common law and modern liability rules. First, recall that the common law defined the agreement requirement in conspiracy as two or more persons agreeing. Not only did the defendant have to agree with another, the other person genuinely had to agree with the defendant. [FN6] Agreeing with an undercover agent, for example, did not constitute an adequate agreement for conspiracy liability; there was no real danger of the commission of a crime. Modern codes typically drop this "bilateral" agreement requirement, and simply require that the actor agree with another. [FN7] Culpable state of mind, not actual danger of commission, is the focus.

Similarly, the common law recognized an unconvictable perpetrator defense to complicity, under which the accomplice escaped liability if, for example, the perpetrator was an undercover agent. [FN8] Modern codes typically reject this defense, arguing that an accomplice's liability ought to depend upon what the actor did and the actor's state of mind, and that a perpetrator's defense ought not affect the accomplice's liability. [FN9] It *302 is the actor's culpable state of mind, not his contribution to an actual offense, that matters.

In the context of attempts, the common law permitted a defense if an actor attempted an offense that, unknown to him, was impossible to complete under the circumstances as they actually existed. Thus, the actor who purchases goods believing them to be stolen, cannot be held liable for an attempt to receive stolen goods where the goods purchased were not, in fact, stolen. [FN10] Modern codes, in contrast, reject this impossibility defense to any form of inchoate liability. An actor's liability for an attempt is determined from the point of view of the actor, under the circumstances as he or she believes them to be. [FN11] If the actor believes that the goods are stolen, he can be held liable for an attempt to purchase stolen goods.

Also in the context of attempts, the common law adopted a variety of "proximity" tests to determine whether an actor's conduct had moved from mere preparation to a criminal attempt. The tests had in common their focus on how close the actor had come (his "proximity") to completing the offense conduct. [FN12] Modern codes, following the Model Penal Code, look instead at how far the actor has gone in externalizing his intention. Once the actor has taken a "substantial step," he has shown his willingness to act upon his intention and, therefore, his conduct is deemed sufficient to impose a criminal attempt liability. [FN13] Asking a secretary to type a letter that is the first of several steps in an elaborate fraud scheme, for example, would not constitute an attempt at common law but could constitute a "substantial step" sufficient for liability under modern codes.

Each of these four instances reflects a "traditionalist" common law consistent in imposing liability only where an offense harm or evil or a credible danger of one actually exists. Completion of an attempt must be possible and must almost materialize; a conspiracy must be a true agreement between two actual conspirators; an accomplice must actually assist a real perpetrator. Where the potential for the harm or evil exists only in the actor's mind or is remote or speculative, traditionalists liability deemed inappropriate.

The subjective "modernist" view focuses upon the actor's intention to bring about the offense harm or evil. An actor need not come close to a *303 substantive harm or evil; he need only engage in some conduct, a "substantial step," toward that end, to demonstrate his willingness to act upon his intention. Completion of an attempt need not be possible, provided that the actor thinks it is possible. The coconspirator need not actually agree with the actor, provided that the actor thinks he does. The perpetrator need not actually commit an offense, provided that the accomplice thinks that he will.

II. MINIMUM REQUIREMENTS FOR LIABILITY VERSUS THE CRITERION FOR GRADING

While the distinction between objective "traditionalists" and subjective "modernists" has proved useful in many respects, it obscures the analysis in other respects because it oversimplifies the possible roles of harm and evil. It suggests that one must be either a "traditionalist" or a "modernist" as to the significance of the occurrence of harm or evil. In reality, one might well conclude that harm and evil are highly relevant to some criminal law decisions but not to others. The same person might rationally be a traditionalist- objectivist on one issue but a modernist-subjectivist on another.

I have argued elsewhere that criminal law performs at least three distinct functions: announcing the rules of conduct, establishing the minimum requirements for liability, and setting the conditions for grading. [FN14] There often is value in distinguishing these functions when evaluating criminal law doctrine, and the doctrines expressing the significance of harm and evil are one of those instances. The four examples of objective-to-subjective reforms, cited above, each concern the law's function of establishing the minimum requirements for liability. The subjectivist view appears to be particularly well suited to this function. There is a near consensus among jurisdictions that have undertaken reform that neither the absence of the offense harm or evil nor the absence of its close proximity ought to shield an actor from criminal liability. [FN15] The subjectivist view seems to be the overwhelming modern view of the minimum requirements for criminal liability.

On the other hand, there is considerable disagreement among modern jurisdictions as to whether the occurrence of harm or evil ought to affect the degree or grade of liability assigned to an actor. A majority subscribes to the view that occurrence of the offense harm or evil is highly relevant to grading; it ought to increase the degree of liability and punishment. This is an objectivist view of grading. A significant minority *304 view take a subjectivist view of grading, and generally hold the occurrence of offense harm or evil irrelevant.

Some jurisdictions take an objectivist view of both the minimum requirements of liability and of grading. Such jurisdictions, typically those with older codes, retain the common law position, which was similarly objectivist in both respects. Other jurisdictions take a subjectivist view of both, including the Model Penal Code and the several jurisdictions following its lead. However, more jurisdictions adopt a subjectivist view with regard to minimum requirements and an objectivist view with regard to grading. That is, they consider purely subjective criminality sufficient to impose liability, yet they increase the degree of liability when the offense harm or evil actually materializes. There are not, then, "traditionalists" and "modernists," as Fletcher suggests, but rather objective and subjective perspectives on the minimum requirements of liability and objective and subjective perspectives on grading.

III. OBJECTIVIST VERSUS SUBJECTIVIST VIEW OF GRADING: THE ARGUMENTS

Why do modern codes take what Fletcher calls the "traditionalist" view of harm and evil in grading? Why do they reject the subjectivist view of grading? To set the stage for this discussion, let me review briefly the primary provisions that implement a subjectivist or objectivist view of grading. I will use the Model Penal Code as representative of the subjectivist view.

The most important provision is § 5.05(1), which grades all inchoate offenses the same as the grade of substantive offense, with the exception that the inchoate form of a first degree felony ( e.g., murder) is graded as a second degree felony. [FN16] Thus, an unsuccessful conspiracy to commit arson is the same grade offense as if the arson occurs. An uncompleted plan to rape is graded the same as if the rape occurs. A solicitation to illegally dump toxic chemicals is graded the same as if the chemicals were dumped.

Subjectivist grading is employed in the Code's complicity provision, which stipulates that an actor is as an accomplice if he "aids or agrees or attempts to aid" in the commission of an offense. [FN17] Thus, an unfulfilled agreement or unsuccessful attempt to assist or encourage is graded the same as the substantive offense that does not materialize. The actor *305 who agrees to stand watch for a perpetrator bent on arson is liable for arson even if he gets the date confused and does not show. In other words, inchoate complicity is punished not as inchoate liability but as full substantive liability.

Adhering to an objective view of grading, a majority of jurisdictions reduce the grade of inchoate conduct below that of the corresponding substantive offense. [FN18] Similarly, many jurisdictions require actual assistance or encouragement for full complicity; an unsuccessful attempt at complicity can only be punished as an attempt. [FN19] Where the actor takes steps to burn a building but another arsonist gets to it first, the actor is liable only for attempted arson, graded less than the substantive offense. Where the actor tries but fails to aid an arsonist, unbeknownst to the arsonist, and therefore has no causal connection with the offense harm or evil, his liability similarly is attempt liability not substantive offense liability, and accordingly graded less. These objectivist views are adopted even by jurisdictions that otherwise are heavily influenced by the Model Penal Code and generally accept its subjectivist view of the minimum requirements of liability. [FN20]

Why these differences in perspective on the significance of harm and evil in grading? The objectivist's preference for increasing liability where the actor causes or contributes to the actual occurrence of the offense harm or evil may be explained in part by a strong intuitive sense:

*306 The notion that there should be a difference in punishment is deeply rooted in the popular conscience, and to ignore it is to risk jury nullification . . . . The successful criminal and the person who engaged in an unsuccessful attempt are in some sense not of equal culpability. They are different in that the defendant who succeeded in his purpose has done an injustice to another; he has brought about an objective evil in the form of an injury to another; he has actively participated in the cruelty and tragedy of life. He has done a worse act than the person who failed and is in some sense a worse man. It is this largely intuitive judgment that finds expression in existing law. [FN21]

The community's shared intuitive sense that resulting harm and evil increases blameworthiness is confirmed by recent empirical studies. In one study, subjects were asked to assign punishment to actors in each of a series of scenarios, including the following:

Because Smith made disparaging remarks about them, Luman and Alma decide to kill Smith. They go to Smith's house and while Alma serves as a watchman, Luman enters the house and stabs Smith in the chest, killing him instantly . Luman and Alma flee, but both are subsequently apprehended.

Because Smith made disparaging remarks about them, Luman and Alma decide to kill Smith. They go to Smith's house and while Alma serves as a watchman, Luman shoots at Smith through a window but misses him . Luman and Alma flee but are subsequently apprehended. [FN22]

In addition to the appropriate degree of punishment, subjects were asked questions that tested their perception of each scenario. Their responses confirm that they generally perceived the actors in the different scenarios as having the same culpable state of mind and as having engaged in similar conduct-conduct designed to cause Smith's death. Yet, 97.3 percent of the subjects believed that both the perpetrator and the accomplice deserve considerably more punishment where they cause a death, in the first scenario, than in the second scenario, where no death occurs. Other studies confirm the same relief in a variety of other situations. [FN23]

Indeed, the studies confirm that subjects even share a common intuition as to the importance of a strong causal connection between the actor's conduct and the prohibited result. In the above study, one of the scenarios given subjects described actors with the same culpable state of mind engaging in the same conduct with a resulting death, as described in the first scenario, but where the death was rather remote and accidental in its occurrence:

*307 Because Smith made disparaging remarks about them, Luman and Alma decide to kill Smith. They go to Smith's house and while Alma serves as a watchman, Luman enters the house and stabs Smith in the chest. Luman and Alma leave thinking that Smith is dead. A neighbor calls the police and an ambulance. Smith is taken to the hospital, undergoes surgery, and recovers . Luman and Alma flee, but both are subsequently apprehended. Two months later, Smith is killed in a traffic accident on his way to the hospital for post- operative treatment for his injury . [FN24]

89.2 percent believed that both perpetrator and accomplice deserve considerably less punishment in this scenario, where the causal connection to the death is weak, than in the first scenario, where the causal connection to the death is strong and immediate. [FN25] Apparently, it is not the occurrence of a resulting harm itself that increases an actor's blameworthiness but rather the occurrence of harm causally attributable to the actor.

Assume, then, that a strong community intuition exists for increasing punishment where harm or evil actually occurs and is attributable to the actor. Why precisely should the community intuition be of interest to drafters of the community's criminal code? Code drafters typically are guided by either retributivist or utilitarian considerations (or a combination of the two [FN26] ) in determining the rules for the distribution of liability and punishment. Let me consider each in turn.

As to desert, moral philosophers will have only fleeting interest in the existence of a strong community intuition on an issue. Deserved punishment, they will observe, is not a function of the community's intuitions on desert but rather is derived from principles of right and good. Conflict with a strong community view might prompt a philosopher to look more carefully at his or her analysis but, in itself, proves nothing; the community may be wrong.

After an independent analysis, might the philosopher come to the same conclusion as the community on this issue? There is disagreement among jurisprudentians as to the significance of resulting harm and evil. What have been called the "intent-based" theorists [FN27] point out that the actual occurrence of the harm or evil is a matter of luck, beyond the control of the actor. An actor can control whether he attempts to cause a harm or evil or risks causing it, but he can do no more than this. If an unfore seen intervening event interrupts the causal chain, it cannot reduce the actor's blameworthiness for his earlier conduct and his accompanying *308 state of mind. These are, of course, the standard subjectivist arguments noted above. The "harm-based" theorists, in contrast, will point to the harm and suffering that occurs when the harm or evil actually comes about and insist that the actor's degree of blameworthiness is increased with the increase of the harm or evil caused. [FN28] This is, of course, simply a restatement of the objectivist view.

Turning to a utilitarian crime prevention analysis, several points argue against taking resulting harm or evil into account. It is the actor's conduct and accompanying state of mind that establish his or her dangerousness, not the fortuity of whether the intended or risked harm or evil actually occurs. [FN29] This is the explanation the Model Penal Code drafters give for grading attempts the same as the substantive offense:

The theory of this grading system may be stated simply. To the extent that sentencing depends upon the antisocial disposition of the actor and the demonstrated need for a corrective sanction, there is likely to be little difference in the gravity of the required measures depending on the consummation or the failure of the plan. [FN30]

The person who attempts the offense but fails-the intended victim happens to move just as the bullet is fired-is no less dangerous by virtue of the failure of the plan. This same explanation applies as well to the actor who attempts but fails to aid another. The accomplice is no less dangerous by virtue of the lack of success in trying to assist or encourage the principal actor. [FN31]

*309 A utilitarian also may focus upon deterrence as a crime prevention mechanism. [FN32] In the context of attempts, one might argue that a reduced grade for attempts provides an incentive for offenders to stop before completing the offense. But the Code drafters implement this crime prevention mechanism by providing a renunciation defense, under which an actor has an incentive to completely and voluntarily renounce his attempt (or conspiracy or solicitation) up until the moment of its fruition. [FN33]

A different kind of deterrence argument does argue against equal grading for inchoate and completed conduct. It focuses upon the inherent ineffectiveness of a threat to punish completed offense the same as inchoate offenses. The deterrence mechanism presupposes that the individual contemplating a criminal act is deterred from it by looking ahead to the threatened punishment and choosing to avoid that punishment. But this "looking ahead" cannot be to the punishment that the justice system will actually assign. It must be to the individual's conceptualizations of what punishment will follow which specific criminal conduct. And the individual's conceptualization is more likely to be in line with the community's conceptualization than the sometimes esoteric assignments of the legal system. If the system will inevitably be perceived as imposing less liability on inchoate conduct, the imposition of full liability by the code is ineffective in providing greater deterrence and, therefore, is wasteful. [FN34]

The most compelling utilitarian argument against ignoring resulting harm and evil in grading is of a different sort: the actual occurrence of resulting harm or evil should be punished less because that is what the public sees as a just distribution of punishment . Giving significance to harm and evil is one of several things, albeit one of the more important, that the criminal law should do to conform its rules to the moral intuitions of the community. [FN35]

This line of argument may be summarized as follows. By having the criminal law mirror the moral intuitions of the community, as in taking account of the occurrence of harm or evil, the criminal law can enhance its reputation with the community as a moral authority, the violation of *310 which deserves moral condemnation. Moral condemnation is an inexpensive yet powerful form of deterrent threat. It demands none of the costs that attend imprisonment or even supervised probation yet, for many citizens, common sense and empirical evidence suggest that it is a sanction to be very much avoided. [FN36] The more important social acceptance is to the citizen, the more terrible this threatened sanction of the shame of criminal conviction becomes. Such a marvelously cost-efficient compliance mechanism is possible, however, only if the system has moral credibility. Each time the system is seen to deviate from the community's notion of justice, as in ignoring the significance of resulting offense harm or evil in grading, the ability of subsequent convictions to draw community condemnation is weakened.

The potential for disutility from loss of moral credibility is implicitly recognized even when utilitarians employ the crime prevention mechanisms of incapacitation or rehabilitation of dangerous offenders. The current treatment of mental illness illustrates the point. Mental illness can cause an actor to violate the criminal law and can signal future violations, yet every jurisdiction provides some kind of defense for mental illness. The criminal law forgoes the conviction of dangerous mentally-ill offenders not because it ignores the need for incapacitation and treatment but because to criminally convict a clearly insane offender would be inconsistent with the community's expectation that criminal conviction is consonant with moral condemnation.

Aside from preserving the crime-deterrent effect in the moral condemnation of criminal conviction, the perceived "justice" of the system is crucial to obtaining the cooperation and acquiescence of those persons involved in the process (offenders, potential offenders, witnesses, jurors, etc.). Greatest cooperation will be elicited where the criminal law's rules and the community's notions of justice generate identical results. [FN37] When the system is seen to convict an actor when the community applauds his action, martyrs are created and revolutionary forces born. When the system fails to convict an actor that the community labels as morally offensive, *311 confidence in the formal system of justice is undermined and informal, "vigilante," systems of justice are encouraged. Similar disutility, albeit of a lesser effect, occurs when the system grades or punishes conduct noticeably more or less than the community's intuitions of justice would permit.

In some instances, the system's ineptness may mitigate some of the potential for injury. Poor comprehension of jury instructions make it easier for juries to exercise their own conceptions of justice. And, even where the legal instructions are understood, juries may simply disregard them if they see the instructions as failing to allow a result congruous with their shared conception of justice. Of course, the jury can effectively assert the community's view only if it has within its authority verdict options that correlate with the actor's blameworthiness. Thus, if the jury wishes to give a reduced grade for an incomplete attempt but is not given this as a verdict option, the verdict it does return-full liability or no liability-will be seen as failing to do justice.

Where jury nullification is effective, it may minimize injury to the system's moral credibility, but it nonetheless drives home to the jurors the untrustworthiness of the legal rules as rules of justice. With juries drawn from the community, and returning to the community to describe their experiences, the long-term effect of such jury nullification is not likely to enhance the law's moral credibility.

Note that it is not the moral accuracy of the criminal law as moral philosophers would define it that is important for the utilitarian value of doing justice. Rather, it is the community's perception of the law's moral correctness that assists the law's effectiveness in preventing crime. Perceived injustice will hurt the law's efficient functioning, even in the face of moral philosophical support for the rule. Results perceived as just will not cause disutility, no matter that moral philosophers judge the rule unjust. It is the community's view of justice that provides the standard by which the system's perceived moral credibility is judged. It follows that the moral force and credibility of the criminal law can be maintained or restored only by having the doctrine mirror the community's notions of justice.

To summarize, one can find both consequentialist and non-consequentialist arguments in support of giving significance to the occurrence of harm or evil in grading, but one also can find counterarguments of both sorts. If desert is the guiding principle, moral philosophers disagree over the significance of resulting harm or evil. If efficient crime prevention is the goal, the traditional arguments support the subjectivist view but more recent empirical data suggests there may be greater utility in following the community's sense of justice, which would take account of resulting harm and evil, the objectivist view.

*312 IV. INCONSISTENCY IN APPLICATION OF THE SUBJECTIVE VIEW OF GRADING

Given the arguments available to the subjectivist, one would expect to find a fairly consistent and complete execution of that view in the jurisdictions that adopt it. But no jurisdiction, even those that claim adherence to the principles of the subjectivist view of grading, is consistent or complete in its execution. The reasons for this failure are worth examining, but let me first demonstrate the inconsistencies and incompleteness, using the Model Penal Code again as an instructive subjectivist vehicle.

If the occurrence of the offense harm or evil should play no role in grading, one may wonder, for example, why the Code creates an exception for first degree felonies in grading inchoate offenses. [FN38] If the arguments for grading inchoate conduct the same as the completed offense are sound, why should they not apply to first degree felonies as well? The Code's commentary offers a deterrent efficacy explanation:

It is doubtful . . . that the threat of punishment for the inchoate crime can add significantly to the net deterrent efficacy of the sanction threatened for the substantive offense that is the actor's object, which he, by hypothesis, ignores. Hence, there is a basis for economizing in use of the heaviest and most afflictive sanctions by removing them from the inchoate crimes. The sentencing provisions for second degree felonies, including the provision for extended terms, should certainly suffice to meet whatever danger is presented by the actor. [FN39]

Thus, the drafters seem to concede that deterrence arguments in support of their policy are unpersuasive; dangerousness is the key. Whether the harm or evil actually occurs does not affect the actor's dangerousness.

But then one may wonder why the Code, like all other modern codes, distinguishes between offenses that differ only in that one punishes an actor when harm or evil occurs and the other punishes an actor, at a lower grade, when the harm or evil does not occur. Note, for example, the dramatic difference in grading between manslaughter and endangerment. The Model Penal Code grading is typical: the former is a second degree felony, the latter, a misdemeanor. [FN40] Yet, the actor's conduct and culpability may be the same under the two offenses; the sole distinguishing variable is existence of a resulting harm or evil. Similarly, recklessly causing a catastrophe is a third degree felony, while the same recklessness where the catastrophe does not occur is punished as a misdemeanor. [FN41] The deterrent-efficiency arguments that the drafters give to *313 explain the exception for grading inchoate first degree felonies does not apply to any of these offenses; in each instance, the no-harm offense is punished only as a misdemeanor, a grade that may not "suffice to meet whatever danger is presented by the actor." [FN42] Perhaps because the drafters do not see the apparent contradiction in their position, the commentary gives no explanation for why the occurrence of harm or evil should not be relevant in the general grading of inchoate conduct but should be relevant when two substantive offenses are defined and graded disparately to take account of the occurrence of harm or evil.

Other incongruities in the standard implementation of the subjective view are equally mystifying. Recall that, under the subjectivist view, an attempt or agreement to aid in an offense results in full substantive liability for the attempted complicity, not merely inchoate liability. [FN43] This is consistent with the subjectivist view that an actor's liability ought to be based on the actor's own conduct and attendant state of mind, rather than on subsequent events over which the actor has no control, such as whether the attempt to aid is successful. Yet, the standard subjectivist complicity formulation also provides that an accomplice may not be liable for full substantive liability unless the perpetrator actually commits the offense. For example, Model Penal Code § 2.06 provides that, while a perpetrator's defense does not redound to the benefit of the accomplice, as it would have at common law, an accomplice cannot be liable for the substantive offense except upon "proof of the commission of the offense." [FN44] It is unclear what exactly this requires; presumably, at the least, the objective harm or evil of the offense must have occurred. [FN45] Consistent with this, the Code explicitly provides that complicity in a perpetrator's failed attempt can only be punished as an attempt. [FN46]

But one might ask, 'If causing the occurrence of the offense harm or evil is immaterial to the grading inquiry, why should it matter to an accomplice's liability whether the perpetrator does or does not actually commit the offense?' To echo the subjectivist argument in support of full substantive liability for inchoate assistance, the accomplice is no less dangerous (or blameworthy) simply because the perpetrator subsequently fails to commit the offense. The accomplice has shown a willingness to aid such an offense. Similarly, if the unsuccessful accomplice is to be held for full substantive liability, based solely upon his or her subjective culpability, why should not the successful accomplice (to the unsuccessful *314 perpetrator) be held to the same result? [FN47] Indeed, one could argue that the successful accomplice (to the unsuccessful perpetrator) has more clearly demonstrated his dangerousness, by carrying through with all of his complicit conduct, than the unsuccessful accomplice. If subjective culpability is to be the sole criterion, is it not wrong to distinguish the two cases? And, if a distinction is to be made, does not the standard formulation have it backwards based on a subjectivist perspective?

The obvious difference between unsuccessful complicity in a complete offense and successful complicity in an unsuccessful offense is that the harm or evil of the offense has occurred only in the former, which is the only one for which the supposedly subjectivist Code imposes full substantive liability. But the subjectivist can hardly rely on this difference, at least not without renouncing the subjectivist view in grading that the occurrence of harm and evil ought to be irrelevant.

The care taken to distinguish unsuccessful complicity in a complete offense from successful complicity in an unsuccessful offense is all the more peculiar when one remembers that attempt liability, in the latter case, will be punished at the same grade as the substantive offense, the liability in the former case. If the grading ultimately is the same, what is the point of having such a carefully structured distinction within criminal law doctrine?

To make the same point more broadly, one may ask, 'Why would the subjectivist in grading have result elements in any offense definition?' Result elements are found in a variety of offenses, including such offenses as felonious restraint, [FN48] sexual assault, [FN49] and arson. [FN50] In each instance, where all elements of an offense are proven except the result element, an actor is liable for an attempt to commit the offense. [FN51] Yet, after the doctrine carefully distinguishes the presence and absence of the prohibited result, by including the result as a requirement of the substantive offense's definition, it then imposes the same grade of liability for both the inchoate and the completed conduct! What is the point of the exercise? If the result element is to be ignored in answering the grading inquiry, *315 why not define the offense without it? Why define offenses to include elements that are supposedly irrelevant to the liability inquiry? [FN52]

Similar observations can be made with regard to the standard subjectivist treatment of offenses other than those with result elements. If the actual occurrence of the evil conduct is irrelevant, why define offenses to distinguish the substantive offense and the attempt? Why not define each offense as "an actor is liable for [the offense] if he does or attempts to do . . . "? (What constitutes an "attempt" could be defined just as it is now.) The Code's careful segregation of inchoate offenses from complete offenses is, again, peculiar in light of its general policy to punish the attempt at the same grade as the substantive offense.

V. ILLOGICAL INCONSISTENCIES OR USEFUL DECEPTION?

If the occurrence of harm and evil is to be ignored, why does complicity liability require that the perpetrator actually perform the conduct constituting the offense? Why are result elements ever required in the definition of offenses? Why does a code ever distinguish attempts from completed offenses? Several possible explanations may be offered. Is it because, liability and grading aside, the criminal law must be careful to label conduct to reflect its relative seriousness? This cannot be the subjectivists' *316 answer because the subjectivist view is that the occurrence of the offence harm or evil is irrelevant to the seriousness of the conduct. Why should labelling take irrelevant factor into account?

One might speculate that the inconsistencies are simply the result of poor drafting or poor thinking. But anyone who has studied subjectivist codes, such as the Model Penal Code, would quickly dismiss this explanation. The Model Penal Code is a thoughtfully and carefully drafted document. The quality of the intellects participating and the decades devoted to the project assured circumspection and caution.

Perhaps the inconsistencies are the product of political compromise, reflecting the fact that there are competing arguments and strongly held views on each side of the debate on the significance of resulting harm and evil in grading. Attempts at compromise on fundamental issues are unheard of in modern codifications, especially where the proponents of one view are not likely to be persuaded by the arguments of their adversaries. In the case of the Model Penal Code, after all, the aim of the American Law Institute was to bring criminal code reform to the United States, and this required a model code that was politically appealing to state legislatures. With regard to the issue of the Code's basic purpose, for example, the Code expressly includes in its statement of purpose both the utilitarian purposes of deterrence, rehabilitation, and incapacitation, as well as the retributivist purpose of just deserts, purposes that may frequently be inconsistent, and even irreconcilable. [FN53]

On the other hand, the political-compromise explanation is less than compelling when one notes that the resulting position is not much of a compromise. No matter that the codes are drafted to generally distinguish substantive and inchoate liability, in the end, all inchoate offenses are converted to full substantive liability under the inchoate grading provision. Only first degree felonies are exempt, a small percentage of the Code's hundreds of offenses. [FN54] And, while some substantive offense definitions appear to take account of the occurrence of harm or evil-manslaughter vs. endangerment, causing vs. risking catastrophe-the great majority do not. [FN55] Further, the compromise theory does *317 not explain the structural changes that have no effect on results, such as defining substantive offenses and inchoate offenses as distinct but then grading them the same.

The most plausible explanation is that subjectivist drafters sought to create the appearance of doctrine that takes account of the occurrence of harm and evil because only that would give the doctrine the moral credibility with the community that it needs, while in reality making the occurrence of harm or evil insignificant because that is what the subjectivist drafters believed better serves the goal of crime prevention.

This grand illusion theory suggests another explanation for the inchoate grading exception for first degree felonies: these offenses, such as murder and kidnapping with serious bodily injury, are the most serious offenses and a failure to grade inchoate conduct lower in these cases would create the greatest and most obvious disparity between the community's intuitive judgment and the legal rules. A similar explanation exists for the subjectivists' giving only inchoate liability for complicity in an unsuccessful offense: full liability for such complicity would be too obvious a deviation from the community's expectations. [FN56]

The concern for preserving the appearance of a code that mirrors community intuitions is illuminated in several other provisions of modern codes. For example, where an actor thinks he is committing one offense but because of a mistake he is actually committing another offense, the Model Penal Code drafters proposed as their first alternative that the actor be liable for the offense that he thought he was committing:

When ignorance or mistake affords a defense to the offense charged but the defendant would be guilty of another [and included] offense had the situation been as he supposed it was, he may be convicted of that other offense . [FN57]

This is consistent with the subjectivist view, of course, where liability is premised upon the actor's view of the facts. In the final draft, however, *318 the drafters selected a second alternative, which would hold the actor liable for the offense he actually committed , thus apparently giving deference to the actual harm or evil rather than the intended harm or evil. [FN58] But while the offense label is changed, the liability is not; the offense of conviction is graded the same as the offense intended rather than the offense committed. [FN59] Thus, the man who thinks he is committing statutory rape of a female unrelated to him but who in reality is his overage daughter, is liable for incest graded as if it were statutory rape. (The drafters' first alternative would have had the man liable for the statutory rape that he thought he was committing.) Resulting harm or evil is the guiding principle.

The drafters' desire for a code that seems to take account of the occurrence of harm and evil, while generally seeking to ignore the same, may well have been a clever strategy, given the arguments presented above concerning the importance of criminal law mirroring community notions of justice. To deviate too conspicuously or too greatly is to risk the law's moral credibility and the cooperation, acquiescence, and coercion to compliance that moral credibility perpetuates. The drafters have every reason, then, to want the code to seem to mirror the community's moral intuitions, especially on matters such as the occurrence of harm and evil for which the intuitions are nearly universal and strongly held.

Such a calculated deception strategy recently has been given additional scholarly approval by Mier Dan-Cohen, who would term it a useful instance of "acoustic separation." [FN60] It may be useful, he argues, for the decisionmakers to be able to distribute liability under rules to which the community is not privy. [FN61] The danger with the strategy, as with many attempts at "acoustic separation," is the serious effects of acoustic leakage, which in an open society seems likely. If the community comes to understand the deception, the system may well lose more credibility than if the code simply overtly deviated from community views. Further, the deception may make it difficult for subsequent reform measures to regenerate credibility for the system. Once deceived, the community understandably may be suspicious and cynical about even genuine reforms meant to make grading more credible. They may understandably ask, another calculated deception?

Consider the past difficulties of sentencing reform. The system sought to maximize deterrent effect by publicly imposing long prison terms but, to conserve resources, allowed early release through the less public determinations of parole commissions. In many systems, including the federal, many offenders could be released almost immediately and none could be *319 made to serve more than a third of the publicly-announced sentence without a special showing by the government. [FN62] The initial justification for the practice cited the need to observe the offender in prison before it could be determined when he or she was rehabilitated or no longer dangerous. Yet this justification became obviously unpersuasive when parole commissions began setting release dates immediately upon each prisoner's arrival in prison, using the identical data available to the sentencing judge. But a primary advantage of parole commission sentencing over judicial sentencing remained: its less public, almost covert, nature.

As people came to understand that the publicly-imposed sentences were not in fact served, judges had to impose longer and longer terms for the sentence to make a meaningful statement. The spiral of sentencing inflation had effects similar to monetary inflation; ultimately it takes a wheelbarrow of currency to buy shoes. Sentences far exceeding an offender's remaining life expectancy, sometimes sentences of hundreds of years, became a necessity for serious offenses. Without them, the credibility of the criminal justice system would erode further.

In the American federal system, the Sentencing Reform Act of 1984 sought to stop the spiral of increasingly long but nonetheless fake sentences by introducing "real-time" sentencing. Under current American law, federal offenders must actually serve 85 percent of the term imposed (15 percent is reserved as an incentive for good behavior in prison). [FN63] But members of the public, now accustomed to long terms, frequently are upset by comparatively short terms that seem to trivialize the offense, even if the time served is longer than the actual time the offender would have served under the old fake- sentence system. The result is that the average sentence actually served by federal prisoners has increased significantly. [FN64] It seems likely that we will pay the price for our earlier deception for some years to come.

To put a more admirable gloss on the subjectivist structuring of the Model Penal Code, one might speculate that such was an attempt to make the Code a useful model even if its position on the insignificance of resulting harm and evil were rejected. Perhaps the drafters knew that their view on harm and evil was not shared by most members of the community and that in the political process surrounding adoption of a criminal code it was likely that many jurisdictions would seek to deviate from the Model Code to make harm and evil matter. To maximize the chance that other valuable contributions of the Code would be adopted, *320 the draftersmay have thought it best to make it easy to alter the Code into a document that takes account of harm and evil. Thus, by defining distinct substantive and inchoate offenses, and equating their grade in a single provision, a jurisdiction could simply alter that inchoate grading provision if it rejected the Code's view of the insignificance of resulting harm and evil.

If this was the drafters' strategy, they should be congratulated for their political acumen. In the United States, three-quarters of the jurisdictions reject the notion of grading inchoate offenses the same as the completed offense. [FN65] Nearly two-thirds of American jurisdictions have adopted codes that have been heavily influenced by the Model Penal Code, [FN66] but less than 30% of these have adopted the Code's inchoate grading provision or something akin to. [FN67] To the many jurisdictions that disagree with the Code on the significance of harm and evil, the drafters' use of the inchoate grading provision, rather than defining all offenses in their inchoate form, no doubt seems a blessing. They can reverse the Code's position simply by altering the relevant grading provisions. The remainder of the Code, with result elements intact, provides many useful advances over prior law in many important respects.

But even this strategy of the Model Penal Code drafters, if that is what it was, can be deceptive, if perhaps inadvertently so. While it may seem that dropping the inchoate grading provision will purge the Code of its disregard for harm and evil, the truth is that the Code's indifference to harm and evil is more pervasive. Recall, the example, that the Code's complicity provision requires only that the actor "aids or agrees or attempts *321 to aid." [FN68] If a jurisdiction rejects the subjectivist view of grading, it would want to delete the italicized language. Yet, of the States heavily influenced by the Model Penal Code that have dropped the Code's inchoate grading provision, more than a third have failed to drop the "agrees or attempts to aid" language from the complicity provision. [FN69]

VI. SUMMARY AND CONCLUSION

Many disputes about criminal law doctrine can be explained as conflicts between an objectivist view that harm and evil ought to be relevant to liability and a subjectivist view that it ought not. But, this Article suggests, such an opposing-camps characterization of the objectivist- subjectivist distinction is too simplistic to capture the variety of roles that may be given to the occurrence of harm and evil. One might conclude harm and evil irrelevant in determining whether criminal liability ought to be imposed but highly relevant in the subsequent determination of the degree of liability. In other words, a person might take a subjectivist view as to minimum requirements of liability but an objectivist view as to grading.

In fact, more modern jurisdictions take this view than any other view. Most agree that the minimum requirements for liability, as in the definition of inchoate offenses, may be primarily subjective. No harm or evil need occur or have come close to occurring; the actor need only externalize his or her culpable state of mind. On the other hand, once the minimum requirements threshold is passed, most jurisdictions choose to impose greater liability where a harm or evil actually occurs than where it only appears to the actor that it would occur.

Among these jurisdictions that do claim a subjectivist view of grading, such as the Model Penal Code, their codes are inconsistent and incomplete in implementing the subjectivist view. But, while their treatment of harm and evil in grading may seem seriously conflicted, the inconsistencies *322 may well have been carefully calculated. [FN70] The drafters may have appreciated that their subjectivist view of grading offenses does not match that of the community and that such a deviation from the community's view of just punishment could undercut the effectiveness of the criminal law in gaining compliance. Although they are good utilitarians, or perhaps because they are, subjectivist drafters generally produce codes that appear to track community notions of deserved punishment, at least in their most visible features.

The strategy of appearances may have a degree of success but it risks discovery. If the community comes to appreciate the codes' deception, it may lose faith in the moral integrity of the punishments imposed and that may more than offset any increase in utility gained from the deception. A better course may be to try to change community views on the significance of harm and evil, if that is possible. Until that time, and unless that occurs, utilitarians might best prefer a code that reflects the community's views of just deserts, which presently includes giving significance to the occurrence of harm and evil.

[FNa1]. I wish to acknowledge the help of participants in seminars at the University of Edinburgh's Centre for Criminology and the Social and Philosophical Study of Law and Department of Private Law and at IIT Chicago- Kent College of Law, and of participants at the Conference on Harm and Culpability at the University of San Diego School of Law.

[FN1]. See, e.g., Lawrence Crocker, "Justice in Criminal Liability: Decriminalizing Harmless Attempts," Ohio State Law Review 53 (1992): 1057-1110 (comparing "subjective" and "objective" "theories of criminal liability"); George P. Fletcher, Rethinking Criminal Law (Boston: Little, Brown, 1978), pp. 135-44, 389, 472-83 (referring to three "patterns of liability": "subjective," "harmful consequences" and "manifest").

[FN2]. Generally, a "harm" or a "harmful" result is used here to refer to a tangible or intangible consequence of conduct, such as those described in the result elements of offense definitions. "Evil" conduct refers to conduct that is objectionable and prohibited for its own sake rather than because of a harmful consequence that it brings about. Taken together, the "harm and evil" of crimes is meant to include all prohibitions of the criminal law. The phrase "harm and evil" is sometimes used in this article as shorthand for "the occurrence of harm and evil."

[FN3]. For further discussion of this function of the act requirement, see Paul H. Robinson, "Should the Criminal Law Abandon the Actus Reus - Mens Rea Distinction?" in Action and Value in Criminal Law , eds. Stephen Shute, John Gardner & Jeremy Horder (Oxford: Clarendon, 1993).

[FN4]. George P. Fletcher, A Crime of Self Defense: Bernhard Goetz and the Law on Trial (New York: Free Press, 1988), p. 64.

[FN5]. Ibid.

[FN6]. See, e.g., Archbold v. State , 397 N.E.2d 1071 (Ind. Ct. App. 1979), in which the court held that " [t]he offense of conspiracy does not occur, a crime is not committed, until two or more persons form the intent to commit a felony. The joint intent is the proscribed conduct ." Id at 1073 (emphasis in original).

[FN7]. See, e.g., Model Penal Code § 5.03(1)(a)&(b):

The unilateral approach makes it immaterial to the guilt of a conspirator whose culpability has been established that the person or all of the persons with whom he conspired have not been or cannot be convicted. Traditional law has frequently held otherwise, reasoning from the definition of conspiracy as an agreement between two or more persons that there must be at least two guilty conspirators or none.

Model Penal Code § 5.03 comment 398-402 (1985). See also People v. Schwimmer , 66 A.2d 91, 411 N.Y.S.2d 922 (1978), aff'd, 47 N.Y.2d 1004, 420 N.Y.S.2d 218, 394 N.E.2d 288 (1979) (court held that since the legislature had adopted a unilateral agreement requirement, defendant's agreement alone was sufficient to support conspiracy to steal diamonds, although the coconspirators were undercover police officers).

[FN8]. In Regina v. Richards , 1 Q.B. 776 (Crim. App. 1974), defendant wife had hired defendant perpetrators to beat her husband "bad enough to put him in the hospital for a month." Ibid. at 778. After beating the husband, but failing to inflict the type of serious bodily harm the wife had requested, the perpetrators were convicted of unlawful wounding; the wife as accomplice was, in turn, convicted of wounding with intent to do grievous bodily harm. On appeal, her conviction was overturned because the only offense committed by the perpetrators was unlawful wounding, and an accomplice, held the court, "cannot be guilty of a graver offence than that in fact which was committed." Ibid. at 780.

[FN9]. See, e.g., Model Penal Code § 2.06(7):

An accomplice may be convicted on proof of the commission of the offense and of his complicity therein, though the person claimed to have committed the offense has not been prosecuted or convicted or has been convicted of a different offense or degree of offense or has an immunity to prosecution or conviction or has been acquitted.

See also Regina v. Cogan & Leak , 1 Q.B. 217 (Crim. App. 1976), in which Leak induced Cogan to rape his (Leak's) wife. Perpetrator Cogan received a defense for his mistake as to the wife's lack of consent. Leak argued that he could not be convicted of complicity in the rape as Cogan had not been found guilty due to his mistake. In rejecting this rationale, the court noted that "the wife had been raped. Cogan had sexual intercourse with her without her consent. The fact that Cogan was innocent of rape because he believed that she was consenting does not affect the position that she was raped." Ibid. at 223.

[FN10]. See, e.g., People v. Jaffe , 185 N.Y. 497, 78 N.E. 169 (1906), in which impossibility was held to be a defense to the receipt of stolen property where the goods had been "wholly within [the owners'] control [at the time of sale] and [were] offered to the defendant by their authority." Ibid. at 499, 78 N.E. at 169.

[FN11]. See, e.g., Model Penal Code § 5.01(1)(a).

[FN12]. The common law proximity tests include the most-often encountered "dangerous proximity test," which requires a consideration of (1) the gravity of the offense intended, (2) the nearness of the act to completion of the crime, and (3) the probability that the conduct will result in the offense intended. See, e.g., Commonwealth v. Peaslee , 177 Mass. 267, 59 N.E. 55 (1901) ("a mere collection and preparation of materials in a room for the purpose of setting fire to them, unaccompanied by any present intent to set the fire, would be too remote," ibid. at 273, 59 N.E. at 57).

[FN13]. See, e.g., Model Penal Code § 5.01(1)(c).

[FN14]. Paul H. Robinson, "A Functional Analysis of Criminal Law," Northwestern University Law Review 88 (1994): 857-913.

[FN15]. For a recent dissenter on this point, see Lawrence Crocker, "Justice in Criminal Liability: Decriminalizing Harmless Attempts," Ohio State Law Review 53 (1992): 1057-1110.

[FN16]. Model Penal Code § 5.05(1). For American jurisdictions following the Model Penal Code's lead, see below note 67. Other countries have similar provisions. See, e.g., West German Criminal Code art. 23(2); (English) Criminal Attempts Act 1981 § 4. These latter provisions may reflect a desire to maximize judicial sentencing discretion rather than a principled view that inchoate conduct generally ought to be punished the same as a completed offense, as is the Model Penal Code view. See Andrew Ashworth, "Criminal Attempts and the Role of Resulting Harm Under the Code, and in the Common Law," Rutgers Law Journal 19 (1988): 725-772.

[FN17]. Model Penal Code § 2.06(3)(a)(ii) (emphasis added).

[FN18]. The American jurisdictions that authorize a sentence at one grade lower than the completed substantive offense include: Alaska Admin. Code tit.13A § 4-2(d); Alaska Stat. § 11.31.100(d)-(e) (1991); Ariz. Rev. Stat. Ann. § 13-1001 (West 1990); Ark. Code Ann . § 5-3-203 (Michie 1991); Colo. Rev. Stat. § 18-2-101(4)-(9) (1991); Fla. Stat. Ann. § 777.04(4); Ill. Ann. Stat. ch 38, para. 8-4(c) (Smith-Hurd 1991); Kan. Stat. Ann. § 21-3301 (1991); Ky. Rev. Stat. Ann. § 506.010 (Michie/Bobbs-Merrill 1991); Me. Rev. Stat. Ann. tit 17-A, § 152-4 (West 1990); Mo. Ann. Stat. § 564.011 (Vernon 1991); Neb. Rev. Stat. § 28-201 (1990); N.M. Stat. Ann. § 30-28-1 (Michie 1991); N.Y. Crim. Proc. Law § 110.05 (Consol 1991) (lowers grade for attempts except for certain first degree felonies, including first degree murder); Ohio Rev. Code Ann. § 2923.02(E) (Anderson 1991); Or. Rev. Stat. § 161.405 (1989); Tenn. Code Ann. § 39-12-107(a) (1991); Tex. Penal Code Ann . § 15.01 (West 1991); Utah Code Ann. § 76-4-102 (1991); Va. Code Ann. § § 18.2-25 - 18.2-28 (Michie 1991); Wash. Rev. Code Ann. § 9A.28.020(3) (1991). Jurisdictions that authorize sentences for inchoate offenses as some fraction of the completed substantive offense, or a specified lesser term, or a combination of lesser grading approaches include: Cal. Penal Code § 664 (West 1991); D.C. Code Ann. § 22-103 (1991); Ga. Code Ann. § 16-4-6 (1991); Idaho Code § 18-306 (1991); La. Rev. Stat. Ann. § 14:27 (West 1991); Mass. Gen. Laws Ann. ch. 274, § 6 (West 1991); Mich. Comp. Laws Ann. § 750.92 (West 1991); Minn. Stat. Ann. § 609.17 (West 1991); Nev. Rev. Stat. Ann. § 193.330 (Michie 1989); Okla. Stat. Ann . tit. 21, § 42 (West 1991); P.R. Laws Ann. tit. 33, § 3122 (1989); S.D. Codified Laws Ann. § 22-4-1 (1991); Vt. Stat. Ann. tit. 13, § 9 (1990); V. I. Code Ann. tit. 14, § 331 (1991); W.Va. Code § 61-11-8 (1991); Wis. Stat. Ann. § 939.32 (West 1991).

[FN19]. See, e.g., authorities cited below, note 20.

[FN20]. For jurisdictions that model their code after the Model Penal Code but drop the Code's grading of inchoate conduct in § 5.05(1), see below note 69. Jurisdictions influenced by the Model Penal Code but that drop the "agrees or attempts to aid" formulation contained in § 2.06(3)(a)(ii) of the Code include: Ala. Code § 13A-2-23; Alaska Stat. § 11.16.110(2)(B) (1991); Colo. Rev. Stat. Ann. § 18-1-603 (1991); Minn. Stat. Ann. § 609.05 (West 1991); N.Y. Penal Code § 20.00 (Consol 1991); N.D. Cent. Code § 12.1-03- 01(1)(b) (1991); Ohio Rev. Code Ann. § 2923.03(A)(2) (Anderson 1991); Utah Code Ann. § 76-2-202 (1991).

[FN21]. John H. Mansfield, "Hart and Honore, Causation in the Law -A Comment," Vanderbilt Law Review 17 (1964): 487-524, 494-95 (summarizing explanation given by Hart and Honore).

[FN22]. See Paul H. Robinson & John M. Darley, Justice, Liability, and Blame: Community Views and the Criminal Law (Boulder: Westview, 1994) (in press) (Appendix B).

[FN23]. See generally ibid.

[FN24]. Ibid. Appendix B.

[FN25]. Ibid.

[FN26]. As to this unhappy situation, see Paul H. Robinson, "Hybrid Principles for the Distribution of Criminal Sanctions," Northwestern University Law Review 82 (1987): 19-42 (without an articulation of the interrelation between such conflicting purposes, no real distributive principle for liability exists; instead, the collection of conflicting purposes can be used, as needed, to justify nearly any result drafters or judges might wish; ibid. at 41-42).

[FN27]. Andrew Ashworth, supra note 16 at 742.

[FN28]. There may be agreement between the two schools that some mitigation is appropriate where the attempt conduct was incomplete. The "harm-based" theorist will cite the absence of harm or evil. The "intent-based" theorist will point out that because the conduct was incomplete the actor might not have completed the required conduct, suggesting that his intent was not sufficiently resolute to commit the offense. It is this argument that desert theorists use to support recognition of a renunciation defense. See, generally, ibid. at 739.

[FN29]. See Schulhofer, "Harm and Punishment: A Critique of Emphasis on the Results of Conduct in the Criminal Law," University of Pennsylvania Law Review 122 (1974): 1497-1607, 1514-16.

[FN30]. Model Penal Code § 5.05(2) comment 490 (1985).

[FN31]. The Model Penal Code's focus upon utilitarian concerns of incapacitation or reform of dangerous offenders is confirmed by the Code's defense for an inherently unlikely attempt. The Code gives a mitigation or defense to the non-dangerous actor who tries to commit an offense through inherently ineffective means. See Model Penal Code § 5.05(2). Consider the actor who attempts to kill another by use of voodoo, sincerely believing that it will be effective. In some cases the actor may well be of no danger to society. (This is true only if the person does not choose more effective means when voodoo fails in its expected effect.) Yet, if she believes in the effectiveness of the method, it is hard to argue against her blameworthiness. The inherent ineffectiveness of the means, like other instances of impossible attempts, it might be argued, does not take away from the fact that the actor believed that she was committing the offense, and therefore her blameworthiness for the attempt. If a jurisdiction adopts a subjective view of criminality because of a blameworthiness rationale rather than a dangerousness rationale, it would wish to modify the Code by rejecting such a defense. In fact, many jurisdictions modeling their codes after the Model Penal Code have dropped the Code's defense for an inherently unlikely attempt. Model Penal Code jurisdictions that drop § 5.05(2) include: Conn. Gen. Stat. Ann . § 53a-51 (West 1985); Haw. Rev. Stat. § 705-502 (1985); Ind. Code Ann. § 35-41-5- 1(a) (West 1985); N.H. Rev. Stat. Ann. § 629:1(IV.) (1986); Wyo. Stat. § 6-1-304 (1988).

[FN32]. Which of competing crime control mechanisms the utilitarian ought to rely upon depends upon, among other things, the relative effectiveness and cost of the distributive rules generated by each, together and in combination. See Paul H. Robinson, "Hybrid Principles for the Distribution of Criminal Sanctions," supra note 26, at pp. 31-33.

[FN33]. Model Penal Code § 5.01(4). One might wonder whether adequate incentive could be provided by giving a one grade reduction in liability, rather than a complete defense. That the Code gives a complete defense may suggest that other factors, perhaps blameworthiness judgments, are also at work. Note that the renunciation defense is not available to the actor who, from his perspective, has completed an offense that in fact is impossible. See ibid. (limiting the defense to prosecutions under § 5.01(1)(b)&(c)).

[FN34]. One can construct other deterrence-based arguments in support of reduced liability for unsuccessful attempts but none are compelling. For example, one could argue that greater public attention is given completed offenses, hence the wide disseminationof the deterrent sanction. Thus, one can maximize the deterrent effect per sanction unit by concentrating the sanctions of completed offenses at the expense of the less noteworthy failed attempts. We probably have insufficient data on the mechanism of general deterrence to know whether such an argument is correct.

[FN35]. See, generally, Paul H. Robinson & John M. Darley, supra, note 22.

[FN36].

Influence by the social group can be instrumental. [They] reward and punish their members, either by withholding or conferring signs of group status and respect, or more directly by channelling material resources toward or away from particular members. In focusing on peer group pressures [it has been shown] that law breaking is strongly related to people's judgments about the sanctions or rewards their behavior elicits from members of their social groups. People are reluctant to commit criminal acts for which their family and friends would sanction them.

Tom Tyler, Why People Obey the Law (New Haven: Yale, 1990), pp. 23-24.

[FN37]. It has been pointed out, for example, that in a system that attempts to maximize the utility of the sanction,

[h]arsh punishments for minor offenses do not work precisely because they depart from popular notions about how people should be treated. . . . Where the model requires very harsh penalties that seem unjust or undeserved, the result would almost certainly be nullification and, therefore, actual sanctions would become less severe than they might have been under a system that permitted mitigation.

Seidman, "Soldiers, Martyrs and Criminals: Utilitarian Theory and the Problem of Crime Control," Yale Law Journal 94 (1984): 315-349, 331-32.

[FN38]. Model Penal Code § 5.05(1).

[FN39]. Model Penal Code § 5.05(2) comment 490 (1985).

[FN40]. Compare Model Penal Code § 210.3(1)(a)&(2) to § 211.2.

[FN41]. Compare Model Penal Code § 220.3(1) to § 220.3(2). In the same vein, some offenses are graded according to the extent of the harm actually occurring. Criminal mischief, for example, is a third degree felony if over $5,000 damage is done, a misdemeanor if over $100 is done, a petty misdemeanor if over $25 is done, otherwise a violation. Model Penal Code § 220.3(2).

[FN42]. Text at note 40, supra.

[FN43]. See, supra note 17.

[FN44]. Model Penal Code § 2.06(7).

[FN45]. It seems likely that "proof of the commission of the offense" is intended to mean proof that the perpetrator satisfied all of the objective elements of the offense: the required conduct, circumstance, or result elements. Hence, an unconvictable perpetrator defense of sorts remains, where the perpetrator does not satisfy the objective elements of the offense.

[FN46]. Model Penal Code § 5.01(3).

[FN47]. The Code drafters might argue that the actor's assistance is less dangerous where the offense does not occur. This may be true in many cases but surely is untrue in others. Neither the commentary to the tentative draft nor that to the final draft provides much of a clue to the drafters' rationale for retaining the common law "proof of commission" requirement, particularly in light of the Code's overall subjectivist view. See Model Penal Code § 2.04 comment 38-39 Tent Draft No. 1, 1953) (subsequently renumbered 2.06); Model Penal Code § 2.06 comment 327-328 (1985). Some writers supporting the subjectivist view in grading do in fact suggest that liability of the accomplice should not depend upon the perpetrator actually committing the offense. See Andrew Ashworth, "Criminal Attempts and the Role of Resulting Harm Under the Code, and in the Common Law," supra note 16, at p. 766; Buxton, "Complicity in the Criminal Code," Law Quarterly Review 85 (1969): 252, 278.

[FN48]. See, e.g., Model Penal Code § 212.2.

[FN49]. See, e.g., Model Penal Code § 213.4.

[FN50]. See, e.g., Model Penal Code § 220.1.

[FN51]. See, e.g., Model Penal Code § 5.01(1)(b).

[FN52]. In a related vein, one may ask: if results generally are irrelevant to criminal liability, why should the doctrine insist on a demanding test for causation? The nearly universal American view, even among jurisdictions that claim a subjectivist view of grading, is that, to be held accountable for a result, an actor's conduct must be "an antecedent but for which the result in question would not have occurred." Model Penal Code § 2.03(1)(a). England specifically rejects such a requirement that the actor's conduct be a necessary cause; the English require only that the conduct be " a cause" of the result.

[A]s a matter of law, [the cause] was sufficient if the prosecution could establish that it was a cause, provided it was a cause outside the de minimis range, and effectively bearing upon the acceleration of the moment of the victim's death.

Cato v. Regina , 1 All E.R. 260, 265 (1976), 62 Crim. App. 41 (1976). It must be outside the de minimis range, but describing it as a "substantial cause" is said to be setting the requirement too high. Ibid. at 45, 46.

During the ALI floor debate on the Model Penal Code's causation section, it was proposed that the Code require only that the actor's conduct be "a substantial factor in producing the result." ALI Floor Debate on Model Penal Code § 2.03(1)(a), ALI Proceedings (1962), pp. 77-79, 135-39. The drafters opposed the proposal and it was defeated in favor of the present necessary cause test. To be consistent, should not the subjectivist drafters reason that some degree of randomness exists in every potential causal chain and that an actor's liability ought not depend upon such "moral luck"? Liability ought to depend upon what an actor does and the actor's state of mind at the time of his conduct, for these are the things than an actor can control. Thus, where an actor inflicts a lethal stab wound, intending to kill the victim, but the victim is soon after shot and killed instantly by another independent actor, it is not to the actor's credit that he did not cause her death, in the "but for" cause sense. He did everything he could toward that end. (The fact pattern is from State v. Wood , 53 Vt. 558 (1881), where the actor escaped liability for murder because he failed to satisfy the necessary cause test.) If the drafters believe that the occurrence of the prohibited result generally is irrelevant, on what ground do they insist on the strongest "necessary cause" connection between the actor's conduct and the result? Why not a "sufficient cause" requirement, for example, or an even weaker causation requirement? (For those who question the significance of resulting harm, a sufficient cause test might be seen as an acceptable compromise to the necessary cause test. Although it recognizes the significance of the result, the actor's accountability for that result will depend upon the nature of his own conduct, and not upon the fortuity of the non-occurrence of a sufficient cause under the necessary cause test.

[FN53]. Model Penal Code § 1.02(1).

[FN54]. The first degree felonies in the Model Penal Code are murder, § 210.2, a kidnapping where the victim is not released alive and in a safe place, § 212.1, a rape where the actor inflicts serious bodily injury upon anyone or where the victim is not a voluntary social companion of the actor upon the occasion of the crime and had not previously permitted him sexual liberties, § 213.1(1), and a robbery where the actor attempts to kill anyone or purposely inflicts or attempts to inflict serious bodily injury, § 222.1(2).

[FN55]. It is a second degree felony to aid a "suicide or an attempted suicide ." Model Penal Code § 210.5(2). One is liable for the offense of obscenity if one sells, delivers, or provides, or " agrees to sell, deliver, or provide" obscene material. Model Penal Code § 251.4(2)(a). Simple assault is defined as causing or attempting to cause bodily harm. Model Penal Code § 211.1(1)(a). The Code's definition of robbery provides another example. An actor need not actually commit theft or cause injury in order to be liable for robbery; an attempted theft and a threat to cause injury are sufficient. Model Penal Code § 222.1(1)(a)-(c) (emphasis added). For other similar examples, see Model Penal Code § § 224.7, 224.9, 242.5.

Similarly, to be liable for arson, an actor need only "start" a fire; the fire need not actually have the intended or any destructive effect; arson requires only that the actor act with a purpose to destroy. Model Penal Code § 220.1(1). Similarly, arson under the Code does not require that a danger to persons actually be created. "Occupied structure" is defined broadly to include places where people normally are present, "whether or not a person is actually present." Model Penal Code § 220.1(4).

The Code's subjectivist view of grading also manifests itself in several grading provisions, as when offenses are graded according to the actor's subjective belief in the extent of the harm he caused or attempted to cause. Theft, for example, is graded according to the value of the property that the actor "attempted to steal," rather than what he actually stole. Model Penal Code § 223.1(2)(c). A similar effect occurs in the grading of burglary and robbery, where the grade of the offense is aggravated if the actor "inflicted or attempted to inflict " injury. See Model Penal Code § § 221.1(2) (aggravated burglary, "inflicts or attempts to inflict bodily injury"), 221.1(2) (aggravated robbery, "inflicts or attempts to inflict serious bodily injury").

[FN56]. Perhaps it is for these reasons that the necessary cause test for causation was preferred by the Model Penal Code drafters over a sufficient cause test. Causation is most commonly an issue in homicide cases, which have the greatest visibility to the public, and the necessary cause test more accurately reflects the community's views of when an actor ought to be causally accountable for a result.

[FN57]. Model Penal Code § 2.04(2) (Tent Draft No. 4, 1955) (italics added).

[FN58]. Model Penal Code § 2.04(2). The commentary explains the preference for Alternative No. 2 as based upon procedural advantages. See Model Penal Code § 2.04 comment 273-274 & n. 11 (1985).

[FN59]. Ibid.

[FN60]. Mier Dan-Cohen, "Decision Rules and Conduct Rules: On Acoustic Separation in Criminal Law," Harvard Law Review 97 (1984): 625-677.

[FN61]. Ibid. at 630-632.

[FN62]. See 18 USC § 4205. The court could fix the date of parole eligibility as any time after incarceration to as late as one-third of the sentence, § 4205(b)(1), or could leave it to the discretion of the parole commission, § 4205(b)(2); but, under ordinary circumstances, parole eligibility could be no later than one-third of the imposed sentence, § 4205(a).

[FN63]. At the end of each year served, 54 days (approximately 15% of a year) is credited toward service of the prisoner's sentence unless "he has not satisfactorily complied with . . . institutional disciplinary regulations." 18 USC § 3624(b).

[FN64]. See General Accounting Office of the United States Congress, Report on the United States Sentencing Commission Guidelines (August 1992).

[FN65]. See note 18.

[FN66]. Jurisdictions whose criminal codes have been heavily influenced by the Model Penal Code include: Ala. Code tit.13; Alaska Stat. § § 11.16.100 to 11.81.900; Ariz. Rev. Stat. Ann . § § 13-101 to 13-4202; Ark. Code Ann. § § 41-123 et seq; Cal. Penal Code ; Conn. Gen. Stat. § § 53a-1 to 53a-215; Del. Code Ann. tit. 11; Fla. Stat. Ann. § § 775.01 to 893.15; Ga. Code Ann. § § 26-101 et seq; Haw. Rev. Stat. § § 701-101 et seq; Idaho Code tit. 18; Ill. Ann. Stat. ch. 38; Ind. Code Ann. tit. 35; Iowa Code Ann. § § 687.0 et seq; Kan. Stat. Ann . tit. 21; Ky. Rev. Stat. Ann. § § 500.00 et seq; La. Rev. Stat. Ann. § § 14:1 et seq; Me. Rev. Stat. Ann. tit. 17-A; Mich. Comp. Laws Ann. chs. 750 to 759; Minn. Stat. Ann. § § 609.01 et seq; Mo. Ann. Stat. § § 556.011 et seq; Mont. Code Ann. § § 45-1-100 et seq; Neb. Rev. Stat. § § 28-101 et seq; Nev. Rev. Stat. § § 193 to 207; N.H. Rev. Stat. Ann. § § 625:1 et seq; N.J. Stat. Ann. tit. 2C; N.M. Stat. Ann. § § 30-1-1 et seq; N.Y. Penal Law ; N.D. Cent. Code § § 12.1-01-01 et seq; Ohio Rev. Code Ann. § § 2901.01 et seq; Okla. Stat. Ann . tit. 21; Or. Rev. Stat. § § 161.005 et seq; Pa. Cons. Stat. Ann. tit. 18; P.R. Laws Ann. tit. 33; S.D. Codified Laws Ann. § § 22-1-1 et seq; Tex. Penal Code Ann. tits. 1 to 10; Utah Code Ann . § § 76- 1-101 et seq; Wash. Rev. Code Ann. tit. 9A; Wis. Stat. Ann. § § 939.01 et seq.

[FN67]. Jurisdictions that have adopted the inchoate grading approach of Model Penal Code § 5.05(1) include: Conn. Gen. Stat. Ann. § 53A-51 (West 1991); Del. Code Ann. tit. 11, § 531 (1990); Ind. Code Ann. § 35-41-5- 1(a) (West 1991); Miss. Code Ann. § 91-1-7 (1991) (capital crimes are limited to a maximum term of 10 years); N.J. Stat. Ann. § 2C:5-4(a) (West 1991) (but grading attempted murder the same as the completed offense); N.H. Rev. Stat. Ann. § 629:1(IV) (1990); N.D. Cent. Code § 12.1-06-01-3 (1991); Pa. Cons. Stat. Ann. tit. 18, § 905(a) (Purdon 1991). An additional five states have adopted provisions that go beyond the Model Penal Code's treatment by not exempting first degree felonies; these include: Del. Code Ann. tit. 11, § 531 (1990); Haw. Rev. Stat. § 705-502; Md. Crim. Law Code Ann. art. 27, § 644A (1988); Mont. Code Ann. § 45-4-103(3) (1991); Wyo. Stat. § 6-1-304 (1991) (but exempting death penalty).

[FN68]. Model Penal Code § 2.06(3)(a)(ii).

[FN69]. Of the States heavily influenced by the Model Penal Code that have dropped the Code's inchoate grading provision (Alabama, Alaska, Arizona, Arkansas, California, Colorado, Florida, Georgia, Idaho, Illinois, Kansas, Kentucky, Louisiana, Maine, Michigan, Minnesota, Missouri, Nebraska, Nevada, New Mexico, New York, Ohio, Oklahoma, Oregon, Puerto Rico, South Dakota, Tennes see, Texas, Utah, Washington, and Wisconsin), the following have failed to drop the "agrees or attempts to aid" language from the complicity provision: Az. Rev. Stat. Ann . § 13-301 (West 1990); Ark. Stat. Ann. § 5-2-403 (1991); Ill. Ann. Stat . ch . 38, § 5-2 (Smith-Hurd 1991); Kan. Stat. Ann. § 21- 3303(a) (1991); Ky. Rev. Stat. Ann. § 502.020 (Michie 1991); Me. Rev. Stat. Ann. tit. 17-A, § 57-3 (1990); Mo. Ann. Stat. § 562.041 (Vernon 1991); Or. Rev. Stat. § 161.155(A)(2)(b) (1989); Tenn. Code Ann. § 39-11- 402(2) (1991); Tex. Penal Code Ann. § 7.02(a)(2) (Vernon 1991); Wash. Rev. Code Ann. § 9A.08.020(3)(a)(ii) (1991). The failure is particularly disconcerting given that the complicity revision may be one of the more obvious of the provisions in which the drafters have "integrated" their view of the insignificance of harm and evil. For a more subtle example, consider the Code's definition of what constitutes a prohibited risk, discussed at Paul H Robinson, A Functional Analysis of Criminal Law, Northwestern University Law Review 88 (1994): 857-913.

[FN70]. Alternatively, the apparently inconsistent structure might have been the expression of an unarticulated intuition of the dangers of deviating from the community view on an issue that had such strong support.