Chiang noted that "'canonical' articles very much depend on one's methodological predispositions," and he suggested one "doctrinal" piece, two "economics" pieces, and one "lawyer economics" piece. Chiang's work is available here on SSRN, and his papers on the levels-of-abstraction problem and inequitable conduct have been reviewed here at Written Description; I highly recommend his latest working paper, Forcing Patentee Claims, which "explains why patentee-written claims are in fact efficient and desirable, even though they systematically cover more than the actual invention."
1. Doctrinal: Giles S. Rich, Principles of Patentability, 28 Geo. Wash. L. Rev. 393 (1960) (reprinted here and here, all links to Hein).
Judge Giles Rich is one of the most influential figures in U.S. patent law—he drafted (along with P.J. Federico) the Patent Act of 1952, he served on the CCPA and then the Federal Circuit from 1956 to 1999, and he authored many influential articles (including 4 on Madison's list as well as this one). This article walks through the constitutional basis of patent law and the patentability requirements and rejects "the unsound notion that to be patentable an invention must be better than the prior art." (He certainly would have disapproved of the original version of the 1962 Kefauver-Harris amendments, which had a comparative effectiveness requirement for pharmaceutical patents!) In addition to this main argument, Rich discusses nonobviousness and explains why the new standard of section 103, while still presenting "knotty problems," is a clearer and less subjective standard than the old "invention" requirement. He notes that "improvements" made "by the expected skill of ordinary workers in the arts" are unpatentable "[b]ecause they will be made anyway, without the 'fuel of interest' which the patent system supplies"—which sounds like a cost-benefit or "inducement standard" approach to nonobviousness. Also interesting (albeit tangential): Rich lists methods of teaching a foreign language as inventions that are clearly unpatentable under 101, but many such patents exist today.
2. Economics: William D. Nordhaus, Invention, Growth, and Welfare: A Theoretical Treatment of Technological Change (1969) (link to WorldCat entry).
One paragraph can't give justice to a 168-page book, but this work is probably best known for Nordhaus's calculation of the optimal life of a patent: he concluded that the optimal life is 1.1-34 years (p. 81), but that welfare does not improve much once the life reaches 6-10 years (p. 84) (though he also noted that these results are "highly tentative" because uncertainty in the model and parameters "make[s] final judgments extremely hazardous," p. 86). Nordhaus's calculation is based on a static approach to innovation, in which longer life always increases the amount of innovation, and the only tradeoff is the extension of the inefficiencies associated with monopoly (p. 76)—he does not consider cumulative and interdependent innovations. In an extension of Nordhaus's work, F.M. Scherer suggested that because "patent life is a policy variable," the best policy might be "early compulsory licensing or short patent lives" with extensions granted "upon a showing that the patent recipient exhibited exceptional creativity or undertook unusual technical and/or commercial risks." In response to Scherer, Nordhaus concluded that (1) "a fixed patent life is not optimal in theory, although it may be unavoidable in practice," and "too long a patent life is better than too short a patent life"; (2) "the complications arising from risk, drastic inventions, imperfect product markets, and 'inventing around' patents generally point to a longer rather than shorter patent life"; and (3) compulsory licensing "will be desirable if and only if the optimal life is less than the actual life."
3. Economics: Kenneth J. Arrow, Economic Welfare and the Allocation of Resources for Invention (1962) (chapter in The Rate and Direction of Inventive Activity: Economic and Social Factors).
This chapter is often cited for the Arrow Information Paradox—"there is a fundamental paradox in the determination of demand for information; its value for the purchaser is not known until he has the information, but then he has in effect acquired it without cost"—but this is only part of a broader discussion of information as a commodity. Arrow notes that "an ideal socialist economy" separates the reward for information from its price, whereas in our "free enterprise economy" invention is supported by property rights, which lead to underinvestment in invention. Patent laws reduce innovation incentives because they cannot allow perfect appropriation without being "unimaginably complex," and "precisely to the extent that [patent laws are] successful, there is an underutilization of the information" because "the price is positive and not at its optimal value of zero." Basic research "is especially unlikely to be rewarded." Arrow also argues that "the incentive to invent is less under monopolistic than under competitive conditions but even in the latter case it will be less than is socially desirable." Finally, he concludes that the government and nonprofits should "compensate for the underallocation of resources to invention by private enterprises," but notes that this raises "two problems" for which he doesn't really have answers: "how shall the amount of resources devoted to invention be determined, and how shall efficiency in their use be encouraged?"
4. Lawyer economics: Robert P. Merges & Richard R. Nelson, On the Complex Economics of Patent Scope, 90 Colum. L. Rev. 839 (1990).
In this article, Merges and Nelson explore the policy issue of patent scope. To use one of Chiang's favorite examples, should the Wright brothers have received a patent on all flying machines, or only on wooden flying machines with wings and rudders? In one of the few earlier treatments of this issue, Kitch argued that broad patent "prospects" (e.g., on all flying machines) allow an inventor to efficiently coordinate the development of an idea and to avoid wasteful racing. But Merges and Nelson argue, both theoretically and with historical examples, that racing is rarely so wasteful and that "when it comes to invention and innovation, faster is better." They summarize their basic conclusion:
Without extensively reducing the pioneer's incentives, the law should attempt at the margin to favor a competitive environment for improvements, rather than an environment dominated by the pioneer firm. In many industries the efficiency gains from the pioneer's ability to coordinate are likely to be outweighed by the loss of competition for improvements to the basic invention.Merges and Nelson look in particular at the different innovative processes in (1) cumulative technologies, (2) chemical technologies, and (3) science-based technologies, and they discuss ways in which patent doctrine can be used to tailor patent scope in different areas.
Chiang built on this economic approach to patent scope in his levels-of-abstraction article (and there is also a vast economic literature on the subject), but he more recently wrote that "[t]he problem with this economic balancing approach is that courts lack the capability to directly measure such economic variables as social benefits and monopoly costs with any degree of precision"—instead, he argued for "keying patent scope to the novelty of the idea."