Published in Dimitrios I. Roussopoulos, ed., The Anarchist Papers, 1986.


Graham Baugh

Since its original publication in 1970, Robert Paul Wolff's In Defense of Anarchism has generated considerable controversy. Both advocates of State authority and anarchists have found his arguments unsatisfactory. Wolff himself has admitted that in formulating his argument he has "simply taken for granted an entire ethical theory."1 It is only in his later work, The Autonomy of Reason, that Wolff sketches out the ethical theory upon which the arguments of In Defense of Anarchism ultimately depend.2 A proper understanding of Wolff's anarchism and its inadequacies therefore requires an analysis of his arguments in relation to this ethical theory.

Through a formal analysis of rational agency tied to an instrumental view of reason Wolff seeks to demonstrate that insofar as people are rational and they are to act, they must be autonomous. By autonomy Wolff means acting only on one's self-chosen policies rationally connected to one's self-chosen ends. Autonomy therefore requires rational choice and deliberation and rational action. By rational action Wolff means action based on rational policies which consist of the most efficient means for achieving an individual's goals. Rational consistency requires that these policies be consistently applied or acted upon, and that they be consistent, taken together, with each other. [108]

Two important consequences follow from Wolff's analysis of rational agency. The first is that ultimately the only obligation rational agents have is to be autonomous, for all obligations are a strict function of what is instrumentally rational for each person in relation to his or her goals. The second is that, because of Wolff's commitment to an instrumental view of reason, all choice of ends or goals is determined by neither reason nor desire, but is both nonrational and uncaused. "There are," Wolff argues, "in principle, no ends that reason requires and no ends that it rules out."3

It is on this basis that Wolff denies the validity of all claims to legitimate political authority, which Wolff defines as "the right to command, and correlatively, the right to be obeyed."4 From a Wolffian perspective, the exercise of legitimate political authority is either unnecessary or positively immoral. Someone claiming a right to command and to be obeyed will either order other people to engage in activity which constitutes a rational means to their respective ends, which they are already obligated as rational agents to engage in, or that person will order people to engage in activity which is contrary to their own ends, which would be irrational and therefore immoral. To recognize in another person any right to command and to be obeyed would simply be an irrational policy, for there is no necessary connection between the commands such a person may make and the individual goals and policies of each agent. In fact, Wolff's analysis is subversive of the very concept of a right, for rights either entitle people to engage in activities they are obligated as rational agents to engage in, in which case they are superfluous, or they entitle people to engage in activities contrary to their own ends, which would be irrational.5 All government claims to legitimate authority must therefore be rejected.

Wolff's argument also provides the basis for a rejection of all government claims to legitimate coercive power, conceived as the right to enforce one's commands. Not only is the concept of a right problematic, [109] but any form of coercion conflicts with rational agency or autonomy. For Wolff, an agent's actions are rational and autonomous only insofar as they are caused by the agent's own reason, "by one's conception of a rational connection between an end that one has posited for oneself and a bit of behaviour which either is an instance of or else will accomplish that end."6 When someone does something out of fear of the application of a coercive sanction, rather than because he or she has decided that it is the rational thing for him or her to do, the rational connection between the agent's goals and policies is broken. To avoid the sanction the agent must act according to policies which are not rationally connected directly to his or her own goals, but which are instead intended to facilitate the achievement of State goals, forcing the agent to adopt the subsidiary goal of escaping punishment as well. In this case neither the agent's goals nor policies are freely chosen. To this extent the agent is no longer autonomous.

Because autonomy is the necessary precondition for all rational action, anything which inhibits or conflicts with it is illegitimate, or put another way, contrary to reason. The State, conceived as a group of persons claiming a right to make and enforce commands, and to be obeyed, is incompatible with autonomy and therefore illegitimate. Only a society without a State, 'anarchy,' provides the possibility of complete autonomy. Because autonomy is the necessary precondition for all rational action, anarchy is a rational, that is to say a moral necessity. Wolff concludes that "philosophical anarchism would seem to be the only reasonable belief for an enlightened man."7

Critics of Wolff have denied not only the validity of this argument, but the very anarchist status of the argument itself. For example, Jeffrey Reiman claims that Wolff's anarchism "is a personal, inner moral atittude -- and not the political doctrine of anarchism at all!"8 The anarchist theorist John P. Clark agrees with Reiman that "Wolff's argument that autonomy and moral authority are incompatible constitutes neither a defense of anarchism as a political theory nor a proof of the [110] unjustifiable nature of the state and government."9 Reiman argues that a political anarchist, as opposed to a Wolffian moral anarchist, "decries the state in the name of... freedom from coercion, not in the name of any inner freedom or moral autonomy."10 Alan Ritter provides support for this claim when he notes that none of the four major classical anarchist theorists -- Godwin, Proudhon, Bakunin nor Kropotkin -- "use Kantian autonomy as the normative basis for their opposition to state coercion."11

These attempts to deny the anarchist status of Wolff's position fail, however, for primarily two reasons. First, as I argued earlier, by autonomy Wolff means both deciding for oneself what one ought to do and acting on those decisions. Autonomy so conceived is incompatible with any form of coercion, so Wolff's argument for autonomy is equally an argument against the coercion of the State. Second, the classical anarchist thinkers value autonomy very highly, in the sense of both moral autonomy and rational agency, even if they do not assign either kind of autonomy supreme normative value in their theories and arguments. As John P. Clark himself has argued, the concept of personal autonomy, the "ability on the part of the individual to make meaningful, critical, unprejudiced decisions," is central to the classical anarchist conception of freedom itself.12 Virtually all of the major anarchist thinkers emphasize the need for individuals to act according to their own reason rather than the dictates of authority. The supposed distinction between Wolff's 'moral' anarchism and 'political' anarchism is then not as clear-cut as Wolff's critics claim.

Nevertheless, I would like to argue that Wolff's argument fails as a defence of anarchism, and further that his critics were right to question its anarchist status, albeit for the wrong reasons. What Wolff has really defended is a form of moral and political scepticism quite distinct from the political theory of anarchism, and far less coherent or convincing. [111]

In a section of In Defense of Anarchism which has perplexed some of Wolff's critics, "Beyond the Legitimate State," he seems to have implicitly recognized some of the inadequacies of his approach. Completely contradicting his preceding anarchist argument, Wolff claims that the State "cannot be ineradicably other."13 Wolff argues that "it must be possible" to create a State composed of rational agents who "determine to set private interest aside and pursue the general good... which accomplishes that end without depriving some of them of their autonomy."14 I would now like to argue that Wolff is correct, that according to his own premises it is theoretically possible to reconcile individual autonomy with the State.

It is conceivable that a State authority could promulgate and enforce laws in such accordance with an individual agent's goals and policies that it would be rational for that agent to obey the commands of such an authority and to support the coercive apparatus necessary for their effective implementation. Moreover, if support of the State and obedience to its commands constitute the most effective means to the agent's ends, then he or she has a positive obligation as a rational agent to support and obey that State. To fail to do so would be irrational.

To expand on Wolff's own example, a group of agents committed to the pursuit of a common good, but who due to unavoidable limitations in knowledge lack the same competence as a State (perhaps with an extensive bureaucracy, expert social scientists, greater access to information, and so on) for determining which policies would most effectively realize the common good, and who lack the power to effectively implement these policies while the State does not, would be obligated as rational agents to obey the commands of that State and to support the coercive State apparatus needed to implement the required policies in an efficient and effective manner.

It should be noted that this is not a case of a mere coincidence between the individual agent's policies and the laws of the State; rather, the rationality of the agent's policies, and hence their obligatory character, follow from the antecedent rational efficacy of the law in providing for the effective achievement of the agent's goal. Neither does the agent independently evaluate each command, as the agent's [112] obedience to the command is based in part on his or her rationally based acknowledgement that he or she lacks the same competence as the State authority which issued the command to determine what would be the best policy for the achievement of the agent's goal. Here we have a case where the obligation to obey does not conflict with individual autonomy but in fact follows from the agent's own autonomous choice of ends. The State is not ineradicably other. Political philosophy, despite Wolff's premature obituary, would seem to be very alive and well.15

One might respond that this is hardly a crushing objection to Wolff's argument, for it is very unlikely that any State will ever be able to satisfy Wolff's criterion of legitimacy, and if the State and individual autonomy can be made at least theoretically compatible, so much the better. The problem is that not only can autonomy and the State be reconciled according to Wolff's own premises, but his moral theory can provide some persons with a rational justification for wielding coercive power, in the form of the State, over others.

It is important to remember that Wolff himself defines the State as "a group of persons who have supreme authority."16 Clearly, a group of persons could adopt a goal, the exploitation of the working class, for example, which requires a coercive State apparatus for its achievement. To remain autonomous, this group of persons must adopt the most instrumentally efficacious means for the achievement of this end, which in this case includes the creation of a coercive State apparatus under its control to oppress the workers. Although the State, conceived as a "group of persons" may not have a right to command and to be obeyed, it is also the case according to Wolff's own analysis of rationality, morality and obligation, that from the perspective of this "group of persons" it may very well be rational and hence obligatory to command and to compel obedience to its commands. As Carole Pateman remarks, "no political ruler need quake at the implications of Wolff's 'anarchism.'"17 [113]

Wolff's adoption of an instrumental view of reason is fatal to his entire enterprise. If all choice of ends is nonrational, then Wolff can offer no rational arguments for adopting either anarchy or autonomy as collective or individual goals. Neither can he provide rational arguments against adopting goals positively inimical to the ideals of anarchy and autonomy. Worse, given Wolff's moral theory, the sort of rational anarchist community he himself advocates is a conceptual impossibility.

In place of the nation-State Wolff proposes the creation of a rational community of autonomous persons who regulate their mutual affairs through a process of unanimous collective agreement.18 Yet due to Wolff's instrumental view of reason, the members of such a community may not be able even to engage in meaningful moral discourse, let alone to rationally agree to the collective pursuit of various goals.

From the perspective of Wolff's moral theory, the meaning of moral concepts is strictly relative to each agent's subjective choices, for what is morally right and obligatory for an individual agent is whatever is the most rational and efficient means to his or her ends, and these ends are themselves subjective and nonrational. Wolff believes that if there is no rational ground for opposing something, there can be "no moral ground" for opposing it either.19 There can therefore be no rational or moral ground for opposing any particular choice of ends, for all ends are equally nonrational. Because what is rational for one agent (in relation to his or her goals) is not rational for another agent with different goals, what is morally 'good' and 'obligatory' for the first agent will not be morally 'good' and 'obligatory' for the other. Indeed, if the other agent has adopted goals and policies positively incompatible with the first agent's, the two agents will have both conflicting conceptions of the good and conflicting obligations. Any [114] moral discourse between such agents will consist of competing but incommensurable and mutually unintelligible moral claims.

In Wolff's anarchist community there could be no meaningful debate or discussion of collective choice of ends. There could be limited rational discussion of policies but only in an instrumental sense: "given that x is your goal, adopt policy y, the most efficient means to that goal." Because policies are rational (and hence binding for autonomous agents) only in relation to goals, if one agent opposed the policy of another, or his or her autonomy was threatened by that policy, the only way he or she could convince that person to refrain from acting on that policy would be to convince him or her to abandon the goal which made adoption of the policy rationally binding (or 'obligatory' in Wolff's sense). But because choice of ends is nonrational, this cannot be done in a rational manner. The only way to get someone to change his or her goals and concomitant policies would be through threats, bribes, manipulation or some form of nonrational persuasion, but never through rational argument. So even if agreement among agents could somehow be obtained, one of the primary goals of Wolff's rational community, "that reciprocity of consciousness which is achieved and sustained by equals who discourse together publicly for the specific purpose of social decision and action," would be in essence defeated.20

Any agent whose choice of ends is determined by manipulation or some form of nonrational persuasion is not autonomous, according to Wolff's analysis, because the agent's action (of choosing) is not caused by his or her own reason. But to have any reasonable prospect for obtaining other agents' agreement to the pursuit of some goal, rather than merely hoping that everyone will miraculously share the same preferences, an agent will have to devise ways of manipulating and otherwise influencing other agents' nonrational preferences. Other agents will then agree to the pursuit of some goal, not because they were rationally convinced, but because they were successfully manipulated. Thus, the device of unanimous collective agreement will fail to preserve each agent's autonomy, because to get it to function effectively (rather than miraculously) each agent will have to treat all other members of the community as heteronomous agents incapable of rational choice or rational deliberation regarding their nonrational goals. Instead of [115] a rational community of autonomous agents there will be a group of heteronomous agents mutually manipulating one another.

Although Wolffian autonomous agents cannot rationally agree among themselves to the pursuit of various ends, perhaps they can rationally agree to adopt a rational method for the resolution of conflict between their nonrational ends and corresponding policies which will either preserve or maximize their individual autonomy. However, it is impossible to derive from Wolff's purely formal analysis of rational agency any notion of justice or fairness which would limit the autonomy of each agent to a definite sphere of legitimate action, so that each agent would enjoy the maximum amount of autonomy consistent with the autonomy of others.

Any agent who acts according to self-chosen policies, which are internally consistent and rationally connected to his or her goals, is autonomous in a Wolffian sense. It is the formal structure of action, and not the actual content of the agent's goals and policies, which makes each agent's actions rational and autonomous, and hence for Wolff moral. That the agent may have adopted goals and policies which threaten the autonomy of others is not strictly relevant. Autonomous agents are not bound to respect the autonomy of others.

Even if agents did agree to a principle of equal respect, perhaps for prudential reasons, it is difficult to see how such a principle could be applied. Virtually any goal or policy may conflict, either directly or indirectly, with other goals and policies. Upon what basis can such conflict be resolved? Although it may be possible for individual agents to rank their goals on some sort of rational basis, Wolff cannot derive an interpersonal, objective standard or ranking of goals from his instrumental analysis of moral reason. For Wolff, all ends are equally nonrational, so to adopt one common standard or ranking of goals for the resolution of interpersonal conflict either would arbitrarily favour some agents over others, or it would implicitly assume that which Wolff explicitly denies -- that some ends are rationally or morally superior to others.

Perhaps Wolff would endorse some sort of preference utilitarianism, but the maximization of the satisfaction of everyone's preferences, or autonomy, is a substantive principle which does not strictly follow from Wolff's formal analysis of rational agency, a goal no more rational than any other. Moreover, such a principle can justify the infringement [116] of some people's autonomy, or the frustration of some people's preferences, in order to maximize the autonomy of others. As Keith Graham argues, it can also provide a possible justification for either unanimous or majoritarian direct democracy rather than anarchy, because in both unanimous and majoritarian direct democracy any conflict between goals and preferences can be resolved in favour of the greatest number.21

Ultimately, Wolff's concepts of autonomy and obligation are incoherent. One must question in what possible sense agents whose actions are ultimately determined by nonrational choice can be said to be free or responsible. To be autonomous, for Wolff, is to be "moved by reason", rather than by desire, impulse, and so on.22 But it is to be moved by reason only in the sense of acting according to policies rationally connected to one's nonrational goals. These nonrational goals, which make the adoption of policies, and action in general, rational, can be said to be the motor of agency, for they provide the necessary impetus and rationale for action. As Wolff paraphrases Kant, "who wills the end wills the means."23 For the autonomous agent, then, choice of ends and action are inextricably intertwined.

However, the act of choosing ends, which is uncaused by reason, is not an autonomous act but mere nonrational behaviour. But, Wolff argues, agents cannot be held responsible for their nonrational behaviour, as opposed to their rational actions.24 In that case, agents cannot be held responsible for 'choosing' the ends they do choose, for such choice is not rational, or in the true sense of the word, autonomous. And if they cannot be held responsible for the ends they choose, how can they possibly be held responsible for adopting the policies which rationally follow from them? To be 'autonomous' they must follow these policies, but by doing so they make themselves slaves to their own nonrational choices. But to refuse to act according to these policies would also be irrational. Thus, either the agent's actions are ultimately determined by arbitrary choice or they are determined by some other nonrational cause (perhaps the arbitrary choice of another person), but [117] in neither case are they determined by reason. Yet even Wolff admits that to be free and responsible, which is what it means to be autonomous, one must be rational.25

Wolff cannot provide a coherent account of obligation either, for obligation arises from various social practices, such as promising, and Wolff's purely individualistic and abstract analysis of rational agency simply cannot provide a coherent account of any social practice, for social practices are constituted by intersubjective meanings and impersonal rules which transcend pure, subjective rational agency and arbitrary choice.

Wolffian agents lack a common moral vocabulary by which to constitute and sustain social practices. All moral concepts and values will be relative to each agent's subjective, nonrational goals and corresponding policies. Similarly, the 'rules' constituting a social practice will be different for each agent, being the most efficient and rational policies consistent with his or her goals and other policies, rather than impersonal standards. Even if agents can agree to abide by the same rules, despite their lack of a common moral vocabulary, these rules will always be subordinate to each agent's ends. Whenever it is rational to break the rules, the agent will be 'obligated' to do so. The rules constituting a social practice will then have no binding force -- they won't really be rules at all. If the rules are binding and intersubjective, they will threaten individual autonomy just as much as any positive law, for the agent will be required to act in ways not always rationally consistent with a specific personal goal. Wolffian agents must always be superior to any obligations or commitments owed to others, whether an individual to whom one has made a promise or a political authority to whom one has promised obedience.

Wolff's moral contractarianism, his idea that the moral principles governing agents' interrelationships and interaction must (and can) be based on unanimous agreement, and that "all obligations are grounded in the collective commitments of a society of rational agents," is incompatible with his analysis of rational agency and his corresponding notion of individual 'obligation.'26 According to Wolff's own analysis, [118] the act of publicly committing oneself to a policy or a practice, of voluntarily assuming an obligation, adds nothing to the binding character of the policy or practice for the individual agent because any obligation to abide by a policy or practice ultimately depends on the individual agent's subjective commitment to the ends which make adoption of the policy or practice rational, and hence 'obligatory' insofar as the agent is to be autonomous. Any 'obligation' to adhere to a policy or practice is based on the efficacy of the policy or practice in realizing the agent's ends, not on any public act of agreement. Such agreements, if binding, may also threaten individual autonomy by requiring agents to abide by agreements even though it may no longer be rationally advantageous for each agent to do so. But if one's obligations are always subordinate to what is most rational for oneself, then obligations simply will have no real, interpersonal, binding force, in which case they really won't be obligations at all in the normal sense of the word.

Without any meaningful intersubjective moral notions and values, without even a concept of rationality which transcends subjective individual choice, in effect without the very notions which form the basis of social practices, Wolffian agents simply will have no meaningful notions of what a promise, a contract, an agreement or the corresponding obligations mean. Not only will they be unable to collectively and rationally agree to the pursuit of certain goals, they will lack the very conceptual apparatus by which to make any binding agreements whatsoever.

Bakunin captured the essential absurdity of such an abstract form of individualism long ago:

How ridiculous then are the ideas of the individualists of the Jean-Jacques Rousseau school and the Proudhonian mutualists who conceive society as the result of the free contract of individuals absolutely independent of one another and entering mutual relations only because of the convention drawn up among them. As if these men had dropped from the skies, bringing with them speech, will, original thought, and as if they were alien to anything having social origin. Had society consisted of such absolutely independent individuals, there would have been no need, nor even the slightest possibility of them entering into an association; society itself would be non-existent, and those free [119] individuals, not being able to live and function upon the earth, would have to wing their way back to the heavenly abode.27
Ultimately, Wolff's 'autonomous' agents are left stranded in a nihilistic desert, unable to engage in any meaningful moral discourse or social practices, reduced to treating each other as morally 'mute' means to one another's arbitrary ends. It is a far cry from the 'rational community' Wolff originally envisaged. Not even the Categorical Imperative can help us here.

The sort of moral outlook which emphasizes efficiency in its account of rational (and moral) agency, and which treats ends as nonrational, precluding any intelligent debate over anything of moral importance, is typical of modern bureaucratic, technocratic society, socialist or capitalist, and exceedingly narrow and impoverished. It is not so much the mode of production in modern society, but its instrumental mode of reason which distinguishes it from other forms of society, and it is this particularly modern mode of reason which Wolff articulates.

This helps explain both why Wolff's concept of future society has a decidedly Marxist bent, and why he advocates a classically Marxist strategy for attaining it, for Marxism itself is irremediably tied to an instrumental view of reason which ultimately analyzes all social relationships in terms of their functional rationality.28 Wolff's Utopia is a society where politics is reduced to a problem of rational coordination, the administration of things, in which the State becomes the mere instrument of "rational men of good will" who eliminate the irrational domination of society by the market, subduing the market and society "to their wills" through the medium of the State, thereby moving "from the realm of necessity into the realm of freedom."29 In such a society the individual and the general will are united and "a political order which harmonizes authority and autonomy" is finally attained.30 The way to achieve such a society, of course, is to use the most instrumentally effective means possible, and, as Wolff has indicated [120] elsewhere, this means consists of nothing other than the State itself.31 Only a very powerful and centralized State, Wolff claims, is capable of the massive restructuring of the economy and the redistribution of wealth which provide the necessary preconditions for the creation of a decentralized anarchist society in which each person enjoys full autonomy. In order to bring about anarchy one must first strengthen, not lessen, the power of the State.

Genuine anarchists have always strongly opposed this essentially Marxist idea that, in Bakunin's words, "anarchy or freedom is the aim, but the State or dictatorship is the means. Therefore in order to emancipate the masses they must first be enslaved."32 Anarchists reject the instrumental view that the end justifies the means, insisting that ends and means are inseparable. Furthermore, they have increasingly come to reject the whole base/superstructure model and economic reductionism upon which the Marxist position is based.33 Political power, anarchists argue, is not merely the 'excrescence' of the 'economic base,' nor will it necessarily disappear with the abolition of capitalism. The State is not, any more than technology, a morally neutral instrument which can be put to good uses, but by its very nature an instrument of class domination, whether of the capitalist class or a new class of Communist bureaucrats. Wolff's so-called 'philosophical anarchism' therefore is not only incoherent but contrary to anarchist ideals and in direct opposition to the mainstream of anarchist thought.

We see, then, that Wolff's critic, Jeffrey Reiman, was right to say that Wolff's "moral anarchism bears only the most superficial and largely misleading resemblance to political anarchism," if not for entirely correct reasons.34 As Carole Pateman argues, Wolff's moral and political scepticism regarding the legitimacy of authority and the objectivity of substantive moral ends is a form of "scepticism which is inherent in abstractly individualist liberalism" but which "forms no part of a [121] political theory of anarchism."35 But from the failure of Wolff s particular argument neither the legitimacy of the State nor the falsity of anarchism in general follow. The State requires an independent justification. Whether such a justification has been successfully provided by any political philosopher is a question that cannot be answered here. Similarly, whether a successful defence of an alternative (and coherent) theory of anarchism is possible must also be left undecided. But at least this much is certain: Wolff has clearly demonstrated how not to defend anarchism.


1 Robert Paul Wolff, In Defense of Anarchism: With a Reply to Jeffrey Reiman's "In Defense of Political Philosophy" (New York: Harper & Row, 1976), p. viii.

2 Wolff, The Autonomy of Reason: A Commentary on Kant's "Groundwork of the Metaphysics of Morals" (New York: Harper & Row, 1973), pp. 129, 223.

3 Ibid., p. 223. For a more detailed analysis of Wolff's ethical theory, see my thesis: Strange Anarchy: The Philosophical Anarchism of Robert Paul Wolff (University of British Columbia, 1984).

4 Wolff, In Defense of Anarchism, op. cit., p. 4.

5 In his "Reply to Reiman" Wolff states that he does not now believe that persons "have obligations and rights" in a state of nature, which is consistent with my analysis of his argument. See In Defense of Anarchism, Ibid., p. 90.

6 Wolff, The Autonomy of Reason, op. cit., p. 216.

7 Wolff, In Defense of Anarchism, op. cit., p. 19-

8 Jeffrey Reiman, In Defense of Political Philosophy: A Reply to Robert Paul Wolff's "in Defense of Anarchism" (New York: Harper & Row, 1972), p. xxii.

9 John P. Clark, "What is Anarchism?" The Anarchist Moment (Montreal: Black Rose Books, 1984), p. 121.

10 Reiman, In Defense of Political Philosophy, op. cit., p. 48.

11 Alan Ritter, Anarchism: A Theoretical Analysis (Cambridge: Cambridge University Press, 1980), p. 115.

12 John P. Clark, Max Stirner's Egoism (London: Freedom Press, 1976), p. 61.

13 Wolff, In Defense of Anarchism, op. cit., p. 78.

14 Ibid., p. 78.

15 In his "Reply to Reiman," Wolff proclaims that "[p]olitical philosophy... is dead." See In Defense of Anarchism, Ibid., p. 110.

16 Ibid., p. 3.

17 Carole Pateman, The Problem of Political Obligation: A Critical Analysis of Liberal Theory (London: John Wiley & Sons, 1979), p. 137.

18 This concept is discussed in the concluding chapter of The Autonomy of Reason, pp. 223-225. Wolff admits that it differs from the conception of unanimous direct democracy that he put forward in In Defense of Anarchism (pp. 22-27) as a solution to the conflict between authority and autonomy. In response to his critics, Wolff has concluded that this earlier formulation failed to resolve the conflict. "It may be that men are bound by the collective commitments they make," Wolff writes, "but such commitments do not create the sort of political authority I was attempting to analyse." See "Reply to Reiman," In Defense of Anarchism, op. cit., p. 88.

19 Wolff, The Autonomy of Reason, op. cit., p. 225.

20 Wolff, The Poverty of Liberalism (Boston: Beacon Press, 1968), p. 192.

21 Keith Graham, "Democracy and the Autonomous Moral Agent," in Contemporary Political Philosophy: Radical Studies, ed. Keith Graham (Cambrigde: Cambridge University Press, 1982), pp. 113-137.

22 Wolff, The Autonomy of Reason, op. cit., p. 216.

23 Ibid., p. 143.

24 Ibid., p. 136.

25 Ibid., p. 222. Indeed, Wolff defines autonomy itself as "the condition of taking full responsibility for one's actions." See In Defense of Anarchism, op. tit., p. 14.

26 Wolff, The Autonomy of Reason, op. tit., p. 224.

27 Michael Bakunin, The Political Philosophy of Bakunin, ed. G. P. Maximoff (New York: Free Press, 1953), p. 167.

28 See Marshall Sahlins, Culture and Practical Reason (Chicago: University of Chicago Press, 1976). For a defence of Marxism as a form of technological determinism, see G. A. Cohen, Karl Marx's Theory of History: A Defence (Princeton: Princeton University Press, 1978).

29 Wolff, In Defense of Anarchism, op. tit., p. 76.

30 Ibid., p. 78.

31 Wolff, interviewed in The Black Flag of Anarchism, Research Group One (Baltimore: Great Atlantic Radio Conspiracy, 1973), tape #43.

32 Bakunin, Selected Writings, ed. Arthur Lehning (New York: Grove, 1974), p. 270.

33 For recent anarchist critiques of Marxian analysis and politics, see Murray Bookchin, The Ecology of Freedom (Palo Alto: Cheshire Books, 1982) and Toward an Ecological Society (Montreal: Black Rose Books, 1980); and John P. Clark, The Anarchist Moment: Reflections on Culture, Nature and Power (Montreal: Black Rose Books, 1984).

34 Reiman, In Defense of Political Philosophy, op. cit., p. 52.

35 Pateman, The Problem of Political Obligation, op. cit., p. 137.