Vol ## THE PROBLEM OF UNIFORMITY COST In fact, the problem of uniformity cost is potentially far more ignificant than the example above suggests. The distribution of rewards from both cultural and technological innovation is highly skew. For example, uncertainty about demand or about feasibility leads recording companies, motion picture studios, pharmaceutical companies, and biotechnology research firms to invest millions of dollars that will never be recouped in innovation. In these industries, profits from chart-busting songs, blockbuster movies, and blockbuster drugs must be sufficient to cover the losses incurred on other investments. Consequently, industries such as these demand robust intellectual property rights to maximize the profitability of successful innovations When these rights apply uniformly, the social costs are magnified 2. Uniformity Cost Typology Uniformity costs can be categorized as Type I or Type II. Type I uniformity costs arise when the creators of the same class of subject matter face different magnitudes or types of appropriability problem. For Although any counterfactual query introduces certain biases and uncertainties, and posing factual to interested part significant percentage of inventions would have been developed and brought to market without the prospect of patent protection; (2)this effect varies significantly by industry and ()that nonetheless 80% of patentable inventions were patented in industries with high patent-dependencies (pharmaceuticals, chemicals, petroleum, machinery, and fabricated metal products) and 60% of inventions were patented in less patent-dependent industries(primary metals, electrical equipment, instruments, office equipment, motor vehicles, rubber, and textiles). See id at 175-76 See, e.g., F.M. Scherer, The Innovation Lottery, in EXPANDING THE BOUNDARIES OF INTELLECTUAL PROPERTY: INNOVATION POLICY FOR THE KNOWLEDGE SOCIETY I Rochelle Cooper Dreyfuss et al., eds. 2001)(collecting data showing skew distributions variety of industries); F.M. Scherer, Dietmar Harrhoff Jorg Kukies, Uncertainty and the Size Distribution of Rewards from Innovation, 10 J OF EVOLUTIONARY ECON. 175 (2000)(showing through empirical study that distributions of rewards for innovation is highly skew) 26 See, e.g., Edwin Mansfield, et al, Social and Private Rates of Return from Industriai Innovations, 91 QUARTERLY. J ECON. 221, 233-34(1977) 27 See, e.g, Arthur S. Devany W. David Walls, Motion Picture Profit, the Stable Paretian Hypothesis, and the Curse of the Superstar, 28 J. ECON. DYNAMICS CONTROL 035, 1042(2004)(estimating from gross profit data over 13-year span that only 22% of movies made were profitable and of those, 35% made 80% of the total profits earned) Henry G Grabowski, Patents and New Product Development in the Pharmaceutical and Biotechnology Industries(working paper 2002)(finding that the search for blockbuster drugs is what drives the r&d process in pharmaceuticals" and that"[t]he median new drug does not cover the R&d costs of the average compound)at http://www.econ.duke.edu/papers/oTher/grabowski/pateNts.pdf(visitedAug.1,2005)
Vol. ##] THE PROBLEM OF UNIFORMITY COST 10 In fact, the problem of uniformity cost is potentially far more significant than the example above suggests. The distribution of rewards from both cultural and technological innovation is highly skew.25 For example, uncertainty about demand or about feasibility leads recording companies, motion picture studios, pharmaceutical companies, and biotechnology research firms to invest millions of dollars that will never be recouped in innovation.26 In these industries, profits from chart-busting songs, blockbuster movies, and blockbuster drugs must be sufficient to cover the losses incurred on other investments.27 Consequently, industries such as these demand robust intellectual property rights to maximize the profitability of successful innovations. When these rights apply uniformly, the social costs are magnified. 2. Uniformity Cost Typology Uniformity costs can be categorized as Type I or Type II. Type I uniformity costs arise when the creators of the same class of subject matter face different magnitudes or types of appropriability problem. For Although any counterfactual query introduces certain biases and uncertainties, and posing a counterfactual to interested parties poses others, Mansfield’s data suggest that (1) a significant percentage of inventions would have been developed and brought to market without the prospect of patent protection; (2) this effect varies significantly by industry; and (3) that nonetheless 80% of patentable inventions were patented in industries with high patent-dependencies (pharmaceuticals, chemicals, petroleum, machinery, and fabricated metal products) and 60% of inventions were patented in less patent-dependent industries (primary metals, electrical equipment, instruments, office equipment, motor vehicles, rubber, and textiles). See id. at 175-76. 25 See, e.g., F.M. Scherer, The Innovation Lottery, in EXPANDING THE BOUNDARIES OF INTELLECTUAL PROPERTY: INNOVATION POLICY FOR THE KNOWLEDGE SOCIETY 1 (Rochelle Cooper Dreyfuss et al., eds. 2001) (collecting data showing skew distributions in variety of industries); F.M. Scherer, Dietmar Harrhoff & Jörg Kukies, Uncertainty and the Size Distribution of Rewards from Innovation, 10 J. OF EVOLUTIONARY ECON. 175 (2000) (showing through empirical study that distributions of rewards for innovation is highly skew). 26 See, e.g., Edwin Mansfield, et al, Social and Private Rates of Return from Industrial Innovations, 91 QUARTERLY. J. ECON. 221, 233-34 (1977). 27 See, e.g., Arthur S. DeVany & W. David Walls, Motion Picture Profit, the Stable Paretian Hypothesis, and the Curse of the Superstar, 28 J. ECON. DYNAMICS & CONTROL 1035, 1042 (2004) (estimating from gross profit data over 13-year span that only 22% of movies made were profitable and of those, 35% made 80% of the total profits earned); Henry G. Grabowski, Patents and New Product Development in the Pharmaceutical and Biotechnology Industries (working paper 2002) (finding that “the search for blockbuster drugs is what drives the R&D process in pharmaceuticals” and that “[t]he median new drug does not cover the R&D costs of the average compound”) at http://www.econ.duke.edu/Papers/Other/Grabowski/Patents.pdf (visited Aug. 1, 2005)
Vol ## THE PROBLEM OF UNIFORMITY COST example, in the absence of copyright some composers would still create new music; whereas, others may pursue a different line of work. The public would benefit if copyright applied to only music created by the latter group. Instead, under U.S. copyright law all music is protected by the same entitlement. Even when some copyright incentives are needed the magnitude of that need will vary based on the time, effort, and capital at risk or the incentives may be needed to solve different kinds of problem With respect to software, for example, open source programmers rely on rights under copyright to prohibit private appropriation of common-pool software; whereas, many commercial software producers rely on copyright to prohibit unauthorized public appropriation of a privately-held software Copyright law treats all software as literary works and supplies the same rights to both groups. 30 Type II uniformity costs also arise out of variable appropriability problems. Even when all creators within an industry or technological field face roughly the same type and magnitude of appropriability problem, among industries and technological fields the magnitude and type of problem will certainly vary Nonetheless, patent grants the same entitlement to inventors of pharmaceutical drugs and novelty toys, and copyright grants roughly the same entitlement to, for example, authors of novels and computer programs 3. Qualifying the problem Granting uniform entitlements in patent and copyright law necessarily will impose some Type I and Type II costs, and the question for policymakers is how best to reduce these. The magnitude of social costs incurred when the government rewards all innovators with the same entitlement depends on the currency If the government were to More thorough-going welfare analysis would include, among other things, the rection costs imposed on unprotected composers' from being treated differently than cted composers 17U.sC.§l02(a)(2),(7);id.§ 30see17USC.§101 See, e.g., Rebecca S. Eisenberg, Patents, Product Exclusivity, and Information Dissemination: How Law Directs Biopharmaceutical Research and Development, 72 FORDHAM L. REV. 477, 486(2003)("Our patent laws are one-size-fits-all, applying essentially the same rules to biopharmaceutical research that apply to automotive engineering, information technology, semiconductors, and rocket science. But the needs of these fields for patent protection differ. ) Burk Lemley, Policy Levers, supra note XX, at 1584(Appropriability is itself an amalgam of a complex set of variables, many of which are themselves industry-specific
Vol. ##] THE PROBLEM OF UNIFORMITY COST 11 example, in the absence of copyright some composers would still create new music; whereas, others may pursue a different line of work. The public would benefit if copyright applied to only music created by the latter group.28 Instead, under U.S. copyright law all music is protected by the same entitlement.29 Even when some copyright incentives are needed, the magnitude of that need will vary based on the time, effort, and capital at risk or the incentives may be needed to solve different kinds of problem. With respect to software, for example, open source programmers rely on rights under copyright to prohibit private appropriation of common-pool software; whereas, many commercial software producers rely on copyright to prohibit unauthorized public appropriation of a privately-held software. Copyright law treats all software as literary works and supplies the same rights to both groups.30 Type II uniformity costs also arise out of variable appropriability problems. Even when all creators within an industry or technological field face roughly the same type and magnitude of appropriability problem, among industries and technological fields the magnitude and type of problem will certainly vary.31 Nonetheless, patent grants the same entitlement to inventors of pharmaceutical drugs and novelty toys, and copyright grants roughly the same entitlement to, for example, authors of novels and computer programs. 3. Qualifying the Problem Granting uniform entitlements in patent and copyright law necessarily will impose some Type I and Type II costs, and the question for policymakers is how best to reduce these. The magnitude of social costs incurred when the government rewards all innovators with the same entitlement depends on the currency used. If the government were to 28 More thorough-going welfare analysis would include, among other things, the disaffection costs imposed on unprotected composers’ from being treated differently than protected composers. 29 See 17 U.S.C. § 102(a)(2), (7); id. § 106. 30 See 17 U.S.C. § 101. 31 See, e.g., Rebecca S. Eisenberg, Patents, Product Exclusivity, and Information Dissemination: How Law Directs Biopharmaceutical Research and Development, 72 FORDHAM L. REV. 477, 486 (2003) (“Our patent laws are one-size-fits-all, applying essentially the same rules to biopharmaceutical research that apply to automotive engineering, information technology, semiconductors, and rocket science. But the needs of these fields for patent protection differ.”); Burk & Lemley, Policy Levers, supra note XX, at 1584 (“Appropriability is itself an amalgam of a complex set of variables, many of which are themselves industry-specific.”)
Vol ## THE PROBLEM OF UNIFORMITY COST grant a uniform monetary entitlement to all inventors - say a bounty of $ million-whether their invention were a life-saving biomedical device or a novelty toy, the social costs of uniformity would be apparent and such a system would be grossly inefficient Policymakers have chosen, in the main, to grant legal rather than monetary entitlements to innovators. Uniform exclusive rights are not mmediately problematic because three market-based features of the intellectual property system reduce uniformity cost: demand elasticity, price discrimination, and Coasean bargaining. The social costs of intellectual property rights arise only when there is demand for protected information. If demand for a novelty toy that would have been invented in the absence of protection is 0, then even though granting uniform patent rights was unnecessary, uniformity cost is 0 because no potential buyers have been excluded. Uniformity costs rise with demand Even when these uniformity costs arise, under traditional economic analysis, perfect price discrimination theoretically would eliminate the underdistribution of protected information. That is, if intellectual property owners are able to engage fully in first-degree price discrimination selling or licensing to each user willing to pay more than marginal cost static deadweight loss would be 0. As others have shown, however, even as a matter of theory perfect price discrimination would not eliminate all social costs of intellectual property rights. Moreover, even if perfect 32 For a site dedicated to identifying such low-demand inventions, see Patently Silly availableathttp://www.patentlysilly.com/.Ofcoursedemandfortheinventiondoes not refer to only demand in product markets. Any potential user of information for which a patent owner might make a credible threat must be plotted on the invention's demand curve Increases in demand for a work also attract free riding competitors so that increases demand increase both the magnitude of the appropriability problem and the magnitude of social cost. C. Glynn S. Lunney, Jr. Reexamining Copyright's Incentives-Access Paradigm, 49 VAND. L REV. 483, 557(1996)(incentives and access both functions of degree of market power conferred by exclusive rights). Although we should expect rising demand to generate correlated offsetting effects in many cases, when creators of popular works do not require the power over price that patent or copyright promise, uniformity costs rise $4 See HAL VaRIAn CARL SHAPIRO, INFORMATION RULES: Harold Demsetz, The Private Production of Public Goods, 13 J. L. Econ. 293(1970) Economists have become less certain about the theoretical efficiency of perfect price discrimination by natural monopolists or firms engaged in monopolistic competition. See, e. g.,V. Bhaskar and Ted To, Is Perfect Price Discrimination Really Efficient? An Analysis Of Free Entry, 35 RAND J OF ECON. 762(2004); Aaron S. Edlin, Mario Epelbaum Walter P. Heller, Is Perfect Price Discrimination Really Efficient? Welfare and Existence in General Equilibrium, 66 ECONOMETRICA 897(1998). Moreover, many attempts to modify intellectual property law to enhance opportunities for price
Vol. ##] THE PROBLEM OF UNIFORMITY COST 12 grant a uniform monetary entitlement to all inventors – say a bounty of $1 million – whether their invention were a life-saving biomedical device or a novelty toy, the social costs of uniformity would be apparent and such a system would be grossly inefficient. Policymakers have chosen, in the main, to grant legal rather than monetary entitlements to innovators. Uniform exclusive rights are not immediately problematic because three market-based features of the intellectual property system reduce uniformity cost: demand elasticity, price discrimination, and Coasean bargaining. The social costs of intellectual property rights arise only when there is demand for protected information. If demand for a novelty toy that would have been invented in the absence of protection is 0, then even though granting uniform patent rights was unnecessary, uniformity cost is 0 because no potential buyers have been excluded.32 Uniformity costs rise with demand.33 Even when these uniformity costs arise, under traditional economic analysis, perfect price discrimination theoretically would eliminate the underdistribution of protected information. That is, if intellectual property owners are able to engage fully in first-degree price discrimination – selling or licensing to each user willing to pay more than marginal cost – static deadweight loss would be 0.34 As others have shown, however, even as a matter of theory perfect price discrimination would not eliminate all social costs of intellectual property rights.35 Moreover, even if perfect 32 For a site dedicated to identifying such low-demand inventions, see Patently Silly, available at http://www.patentlysilly.com/. Of course, demand for the invention does not refer to only demand in product markets. Any potential user of information for which a patent owner might make a credible threat must be plotted on the invention’s demand curve. 33 Increases in demand for a work also attract free riding competitors so that increases in demand increase both the magnitude of the appropriability problem and the magnitude of social cost. Cf. Glynn S. Lunney, Jr. Reexamining Copyright’s Incentives-Access Paradigm, 49 VAND. L. REV. 483, 557 (1996) (incentives and access both functions of degree of market power conferred by exclusive rights). Although we should expect rising demand to generate correlated offsetting effects in many cases, when creators of popular works do not require the power over price that patent or copyright promise, uniformity costs rise. 34 See HAL VARIAN & CARL SHAPIRO, INFORMATION RULES; Harold Demsetz, The Private Production of Public Goods, 13 J. L. & Econ. 293 (1970). 35 Economists have become less certain about the theoretical efficiency of perfect price discrimination by natural monopolists or firms engaged in monopolistic competition. See, e.g., V. Bhaskar and Ted To, Is Perfect Price Discrimination Really Efficient? An Analysis Of Free Entry, 35 RAND J. OF ECON. 762 (2004); Aaron S. Edlin, Mario Epelbaum & Walter P. Heller, Is Perfect Price Discrimination Really Efficient? Welfare and Existence in General Equilibrium, 66 ECONOMETRICA 897 (1998). Moreover, many attempts to modify intellectual property law to enhance opportunities for price
Vol ## THE PROBLEM OF UNIFORMITY COST price discrimination would theoretically avoid reduction in social value, perfect first-degr discrimination in the intellectual property context is a practical impossibility. The real question is whether policymakers should design intellectual property entitlements to facilitate price discrimination so as to reduce uniformity cost. As Michael Meurer has shown, some forms of price discrimination are socially beneficial and others are socially harmful. Consequently, even when th le law can encourage price discrimination, the problem of uniformity cost reemerges with respect to the need to tailor entitlements to promote only beneficial price discrimination Finally, when demand is positive and price discrimination is imperfect, the Coase Theorem asserts that uniformity cost will affect allocative efficiency only if reallocation or reapportionment of uniform entitlements by contract is too costly. 38 Commentators disagree about the general magnitude of transaction costs in intellectual property sales an licensing, but all will agree that the costs are greater than 0. Indeed, most agree that difficulties in valuing patents and copyrights raise transaction costs to the point that allocative efficiency will depend upon discrimination likely are undesirable. See generally Julie E. Cohen, The Perfect Curve, 53 VAND. L REV. 1799(2000); Michael Meurer, Copyright Law and Price Discrimination 23 CARDOZO LAW REVIEW 55(2001)[hereinafter Meurer, Price Discrimination]; Brett M. Frischmann, An Economic Theory Of Infrastructure And Commons Management, 89 MINN. L REv. 917, 978-80(2005)(discussing distortionary effects of promoting price discrimination) Copyright and Product Differentiation, 79 N.Y.U.L. REV. 212, 255(2004)(noting that perfect price discrimination is impossible), Daniel Farber Brett McDonnell, Why(and How) Fairness Matters at the P/Antitrust Interface, 87 MINN. L. REV. 1817, 1867(2003) ge generally Meurer, Price Discrimination, supra note XX; Michael Meurer, Price Discrimination, Personal Use and Piracy: Copyright Protection of Digital Works, 45 BUFF.L.REV.845(1997) See ronald Coase, The Problem of Social Cost, 3 J LAw EcoN. 1,8(1960)(In these conditions [of high transaction costs] the initial delimitation of legal rights does have an effect on the efficiency with which the economic system operates ). while arguing that policymakers should recognize the effects they have on allocative efficiency when fashioning legal rights for high-transaction-cost environments, Coase also recognized that distributional justice matters and that "the choice between different social arrangements for the solution of economic problems should be carried out in broader terms than this [maximizing total output] and that the total effect of these arrangements in all spheres of life should be taken into account. " ld. at 21 See robert P. Merges, Of Property Rules, Coase, and Intellectual Property, 94 COLUM L REV. 2655, 2661(1994)[hereinafter Merges, Of Property Rules]("Despite a few brave attempts to assume away the obvious, those who have considered the application of the Coase theorem to IPRs have noted the pervasive presence of transaction costs
Vol. ##] THE PROBLEM OF UNIFORMITY COST 13 price discrimination would theoretically avoid reduction in social value, perfect first-degree price discrimination in the intellectual property context is a practical impossibility.36 The real question is whether policymakers should design intellectual property entitlements to facilitate price discrimination so as to reduce uniformity cost. As Michael Meurer has shown, some forms of price discrimination are socially beneficial and others are socially harmful.37 Consequently, even when the law can encourage price discrimination, the problem of uniformity cost reemerges with respect to the need to tailor entitlements to promote only beneficial price discrimination. Finally, when demand is positive and price discrimination is imperfect, the Coase Theorem asserts that uniformity cost will affect allocative efficiency only if reallocation or reapportionment of uniform entitlements by contract is too costly.38 Commentators disagree about the general magnitude of transaction costs in intellectual property sales and licensing, but all will agree that the costs are greater than 0.39 Indeed, most agree that difficulties in valuing patents and copyrights raise transaction costs to the point that allocative efficiency will depend upon discrimination likely are undesirable. See generally Julie E. Cohen, The Perfect Curve, 53 VAND. L. REV. 1799 (2000); Michael Meurer, Copyright Law and Price Discrimination, 23 CARDOZO LAW REVIEW 55 (2001) [hereinafter Meurer, Price Discrimination]; Brett M. Frischmann, An Economic Theory Of Infrastructure And Commons Management, 89 MINN. L. REV. 917, 978-80 (2005) (discussing distortionary effects of promoting price discrimination). 36 See, e.g., Lemley, Free Riding, supra note XX, at 1059 n.115; Christopher S. Yoo, Copyright and Product Differentiation, 79 N.Y.U. L. REV. 212, 255 (2004) (noting that perfect price discrimination is impossible); Daniel Farber & Brett McDonnell, Why (and How) Fairness Matters at the IP/Antitrust Interface, 87 MINN. L. REV. 1817, 1867 (2003) (same). 37 See generally Meurer, Price Discrimination, supra note XX; Michael Meurer, Price Discrimination, Personal Use and Piracy: Copyright Protection of Digital Works, 45 BUFF. L. REV. 845 (1997). 38 See Ronald Coase, The Problem of Social Cost, 3 J. LAW & ECON. 1, 8 (1960) (“In these conditions [of high transaction costs] the initial delimitation of legal rights does have an effect on the efficiency with which the economic system operates.”). While arguing that policymakers should recognize the effects they have on allocative efficiency when fashioning legal rights for high-transaction-cost environments, Coase also recognized that distributional justice matters and that “the choice between different social arrangements for the solution of economic problems should be carried out in broader terms than this [maximizing total output] and that the total effect of these arrangements in all spheres of life should be taken into account.” Id. at 21. 39 See Robert P. Merges, Of Property Rules, Coase, and Intellectual Property, 94 COLUM. L. REV. 2655, 2661 (1994) [hereinafter Merges, Of Property Rules] (“Despite a few brave attempts to assume away the obvious, those who have considered the application of the Coase theorem to IPRs have noted the pervasive presence of transaction costs.”)
Vol # THE PROBLEM OF UNIFORMITY COST the content of intellectual property entitlements. This is particularly true because the externalities that justify patent and copyright law differ fundamentally from those that inspired Coase, and the laws choice is not between granting an entitlement to party a or to party b but between granting an entitlement to party A or to the public at large, comprised of an users.2 Consequently, allocative inefficiency in intellectual property law potentially imposes a far more significant social cost than it does with respect to real property Thus, even after demand elasticity, price discrimination, and Coasean bargaining have been accounted for, we find that if the law is strictly uniform when granting intellectual property rights, society pays too much for numerous innovations that would be created with less robust protection, and the optimal level-of protection must be set lower than necessary to induce the creation of certain costly but socially desirable Innovations The uniformity-cost perspective calls for a reorientation in the economic analysis of intellectual property law. Those who argue that perfect price discrimination alone would be a complete solution to the 40 See, e. g, Clarisa Long, Proprietary Rights and Why Initial Allocations Matter EMORY L.J. 823(2000)(arguing that uncertainty in valuation of patents on basic research tools is likely to block efficient licensing of such tools); Gallini Scotchmer, supra note KX, at 67(The optimal design of IP depends importantly on the ease with which rights holders can contract around conflicts in rights. ) Lemley, Economics of Improvement supra note XX, at 1053(analyzing components of transaction costs and concluding that the result of all these factors is that the transaction costs of intellectual property licenses are significant. ) James Bessen, Holdup and Licensing of cumulative Innovations With Private Information, 82 ECON. LTRS 321(2004)(showing that"the possibility of ex ante licensing does not eliminate the problem of holdup in cumulative nnovation”) See, e.g, Merges, Of Property Rules, supra note XX, at 2657-64, Lemley, Free Riding, supra note XX; Mark A. Lemley, What's Different About Intellectual Property, 82 TEX L. REv. 1097(2005)(reply essay ); Mark A. Lemley, Ex Ante Versus Ex Post Justifications for Intellectual Property, 71 U CHI. L REV. 129(2004) See generally Frischmann, supra note xx (discussing variety of demand-side considerations for information resources In my view, interpersonal and intrapersonal incommesurability problems make the notion of an optimal level of protection incoherent. Nonetheless, the case for some level of protection is persuasive for at least some forms of information. Those who share my doubts about the utility of optimality analysis should understand"optimal level" to mean the level of protection that democratically-representative policymakers would choose to bring about a desired amount of investment in innovation, recognizing the incommensurable interests that are sacrificed with each change in the level of protection See lunney, Quiet Revolution, supra note XX, at 50-51
Vol. ##] THE PROBLEM OF UNIFORMITY COST 14 the content of intellectual property entitlements.40 This is particularly true because the externalities that justify patent and copyright law differ fundamentally from those that inspired Coase,41 and the law’s choice is not between granting an entitlement to party A or to party B but between granting an entitlement to party A or to the public at large, comprised of an unknown and often unknowable proportion of higher- and lower-valued users.42 Consequently, allocative inefficiency in intellectual property law potentially imposes a far more significant social cost than it does with respect to real property. Thus, even after demand elasticity, price discrimination, and Coasean bargaining have been accounted for, we find that if the law is strictly uniform when granting intellectual property rights, society pays too much for numerous innovations that would be created with less robust protection, and the optimal level43 of protection must be set lower than is necessary to induce the creation of certain costly but socially desirable innovations.44 The uniformity-cost perspective calls for a reorientation in the economic analysis of intellectual property law. Those who argue that perfect price discrimination alone would be a complete solution to the 40 See, e.g., Clarisa Long, Proprietary Rights and Why Initial Allocations Matter, 49 EMORY L.J. 823 (2000) (arguing that uncertainty in valuation of patents on basic research tools is likely to block efficient licensing of such tools); Gallini & Scotchmer, supra note XX, at 67 (“The optimal design of IP depends importantly on the ease with which rights holders can contract around conflicts in rights.”); Lemley, Economics of Improvement, supra note XX, at 1053 (analyzing components of transaction costs and concluding that “[t]he result of all these factors is that the transaction costs of intellectual property licenses are significant.”); James Bessen, Holdup and Licensing of Cumulative Innovations With Private Information, 82 ECON. LTRS 321 (2004) (showing that “[t]he possibility of ex ante licensing does not eliminate the problem of holdup in cumulative innovation”). 41 See, e.g., Merges, Of Property Rules, supra note XX, at 2657-64; Lemley, Free Riding, supra note XX; Mark A. Lemley, What’s Different About Intellectual Property, 82 TEX. L. REV. 1097 (2005) (reply essay); Mark A. Lemley, Ex Ante Versus Ex Post Justifications for Intellectual Property, 71 U. CHI. L. REV. 129 (2004). 42 See generally Frischmann, supra note XX (discussing variety of demand-side considerations for information resources). 43 In my view, interpersonal and intrapersonal incommesurability problems make the notion of an optimal level of protection incoherent. Nonetheless, the case for some level of protection is persuasive for at least some forms of information. Those who share my doubts about the utility of optimality analysis should understand “optimal level” to mean the level of protection that democratically-representative policymakers would choose to bring about a desired amount of investment in innovation, recognizing the incommensurable interests that are sacrificed with each change in the level of protection. 44 See Lunney, Quiet Revolution,, supra note XX, at 50-51