Paradox of Self-Amendment by Peter Suber Of these I prefer the first, although there is no need to choose. However, I will argue that strict self-amendment is permissible in any case, even if Ross's contradiction is ineliminable. The two solutions are compatible, and may be invoked alternately as circumstances require. If an AC is amended by appeal to its own procedures, then even if this creates a contradiction, it may nevertheless create a valid new AC, if the necessary acceptance is obtained. On the other hand, if the AC is amended by procedures that it does not itself specify, the same conclusion follows: if accepted, the product can be valid law, although it would not in that case be self-amendment. If an AC is followed, self-amendment has occurred and the contradiction is overruled by acceptance; if the AC is not followed, self-amendment has not occurred and the new AC would be unauthorized unless directly authorized by acceptance. As a matter of fact, the people and officials are much less likely to accept an amendment of an AC that did not follow the AC's procedures. This is a fact about law only because it is a fact about acceptance, not because it is a fact about logic. Even Ross admits that an amendment to an AC that did not follow the AC's own procedures would not usually be accepted, although he denies that acceptance can confer validity. Ross admits this because he attributes to ordinary citizens the belief that an AC may be amended by its own procedures and, if the AC is to be amended at all, that it must follow its own procedures. Then he chastises the ordinary citizen for the contradiction latent in this belief.[Note 14] According to Ross, if we want to amend an AC, then applying the AC to itself in self-amendment is a necessary condition of doing so plausibly or acceptably, but it is a sufficient condition of doing so invalidly or illegally. ACs and other rules of change may as well be thought self-applicable for the acceptance model. The legal validity of self-application is to be judged by the acceptance it inspires, not the contradictions it entails. An indissoluble contradiction is powerless to nullify a legal act in the face of sufficient acceptance. The paradox of self-amendment exists only for the inference model, or some other concept of law that allows logical tests to overcome legal and social tests of legal validity. If the paradox of self-amendment exists only for the inference model, then changing models will dissolve the paradox, at least for law. The paradox of self-amendment has forced us to choose between fundamental philosophies of law. The acceptance model is just one alternative. A second alternative, developed to a degree in Section 10, may be called the procedural model. The procedural model allows the outcomes of any valid procedure to be valid law even if there is a formal contradiction between the outcome and the antecedents of the procedure. The direct acceptance theory, however, is sufficient to escape Ross's paradox, and in Hart's version of it sufficiently complex to take up our time. But by acknowledging that other models may exist and suffice, we must be careful to avoid affirming the truth of the direct acceptance theory just because it circumvents the conditions of the self-amendment paradox, although we are clearly entitled to deny Ross's inference model because it fails to do so. For the merits of the direct acceptance theory, beyond the demerits of the inference model and the paradox of self-amendment, see Section 21.D. What follows is a slightly more complete account of Hart's acceptance theory, long enough to show its weak points, but short enough to prevent our being diverted from the main track of our discussion. Notes 1. Oliver Wendell Holmes, Jr., The Common Law, Little, Brown, and Co., 1881, p. 1. 2. There is some support for this view of Hart's theory in his The Concept of Law, Oxford University Press, 1961, pp. 92, 93, 97, l06, 205, and 229. One of the most formalistic statements of Hart's theory is to be found in John M. Farago, "Judicial Cybernetics: The Effect of Self-Reference on Dworkin's Rights Thesis," Valparaiso University Law Review, 14 (1980) 371-425, at p. 377-78.n.3: We might view Hart's model as an instrumental application of deductive logic to jurisprudence... Thus, the rule of recognition provides a single undefined assumption, against which we can evaluate the various axioms (rules) of our system. Propositions of law are demonstrably true (or false) by application of the rule of recognition and the primary and secondary rules. The rule of recognition itself provides a recursive rule for determining whether any given proposition is a theorem of the system. 3. Hart, op. cit., at pp. 149-50 (emphases in original). 4. Joseph Raz prevents an infinite regress of authorities with a similar theory. Arguing more against Kelsen than Ross, Raz says, The authors of the first constitution [or the constitution without a prior authority] have legislative power, which is not given to them by law, to make the first constitution. The first constitution is law because it belongs to an efficacious legal system (a fact which cannot be determined until some time after the constitution is first issued)....[L]aws can indirectly authorize their own creation. Joseph Raz, The Concept of a Legal System: An Introduction to the Theory of Legal System, Oxford University Press, 1980, at p. 138; emphases added. 5. Hart, ibid., at p. 142. See also Stoliker v. Waite, 101 N.W.2d 299, 305 (1960) in which the Michigan Supreme Court said that the obstacle in the state AC to rapid and easy amendment is "secure from judicial profanation, though its security is measured only by the tradition of the law, our sense of self-restraint." 6. Hart, ibid., at pp. 121-50. 7. Hart, ibid., at p. 124. 33