Prometheus and the Future of Software Patentability

By: Gareth Tan

“It is noted that the creation of programs has undergone substantial and satisfactory growth in the absence of patent protection.” —“To Promote the Progress of…Useful Arts,” Report of the President’s Commission on the Patent System (1966).


In the field of computer law, a patent on software might be defined as a government-granted monopoly on ideas embodied in methods or processes relating to the development and use of computer software.1 A patent holder is granted such a monopoly in exchange for disclosing how the method is performed. Those instructions are then placed in the public domain. Patents are currently valid for twenty years and grant their holders, among other rights, “the right to exclude others from making, using…or selling the invention throughout the United States.”2

One focus in the debate over software patentability concerns the fact that software usually contains much mathematical reasoning, a fair amount of which is found in algorithms. These are “finite sets of rules that give a sequence of operations for solving a specific type of problem.”3 Since the advent of the modern computer industry about four decades ago, the courts have struggled with the question of whether software, and the algorithms they contain, should be granted patent protection.

With its decision in Mayo Collaborative v. Prometheus Laboratories4, the Supreme Court paves the way towards a wise resolution of this debate by strongly suggesting that software patents ought to be less leniently granted than before. This is a return to its position of the late 1970s. In the case, the Court moves towards recognizing that granting patents on software and the algorithms they contain often also involves granting monopolies to individuals or companies on natural and abstract concepts. The Court correctly perceives that to do so would hinder technological development and innovation. Prometheus is thus a decision that must be commended.

The Patent-Eligibility Trilogy

The law governing what subject matter is eligible for patent protection is set forth in 35 U.S.C. § 101:

“Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.”5

The statute further defines “process” to mean a “process, art, or method” as well as a “new use of a known process, machine, manufacture, composition of matter, or material.”6

The question of whether computer programs containing algorithms would fall under the definition of a ‘process’ was the subject of three cases decided by the Supreme Court in the late 1970s and the early 1980s. These cases – Gottschalk v. Benson7, Parker v. Flook8, and Diamond v. Diehr9 – are known as the patent eligibility trilogy. In them, the Court held that pure algorithms, while apparently included in this definition of ‘process’, could not be patented because of an implicit exception in §101 preventing the patenting of natural laws. However, programs that contained algorithms could be patented as long as they transformed an article into a different state or thing.

In the first algorithm patentability case, Gottschalk v. Benson, the respondents attempted to patent a “method of converting signals from binary coded decimal (BCD) form into binary” and an associated data processing method.10 Their patent simply recited the steps of the algorithm. Writing for a unanimous Court, Justice Stevens found that “one may not patent an idea. But in practical effect that would be the result if the formula for converting BCD numerals to pure binary numerals were patented in this case.”11 For a “principle, in the abstract, is a fundamental truth; an original cause; a motive; these cannot be patented, as no one can claim in either of them an exclusive right.”12

Benson, too, articulated the machine-or-transformation test for the first time with regards to software patents. “Transformation and reduction of an article ‘to a different state or thing’ is the clue to the patentability of a process claim that does not include particular machines,”13 the Court wrote, though it was not a necessary condition for patentability. With the exception of the short time in which Parker v. Flook proposed a new patentability test, the machine-or-transformation test would be established as precedent for more than three decades.

Problems with software patentability are likely to arise when patents are claimed on software not tied to any particular computer system. In Parker v. Flook, second in the trilogy, the respondent attempted to patent software that could calculate when an alarm should sound within a catalytic conversion process to indicate that it was proceeding abnormally. The software executed an algorithm that used Arrhenius’ equation, a well-known chemical equation, to calculate these alarm limit values.14

The Court decided that Flook’s system “contained no claim of patentable invention.”15 When dealing with a claimed process containing a “law of nature or mathematical algorithm” the Court directed that a procedure known in the scholarly literature as analytic dissection should be used. The algorithm in such a procedure must be “treated as though it were a familiar part of the prior art”16 – that is, part of the publicly available knowledge in a field. Only if the remaining parts of the “process itself” were “new and useful”, and contained some “inventive concept”, could the patent be granted.17 A fortiori, one cannot simply patent abstract principles by adding “conventional or obvious” “post-solution activity” (such as a direction to apply a formula to a practical problem) to the claim.18

The Court further clarified its previous holding in Benson by deciding that even claims that did not completely pre-empt the use of an algorithm could still be barred by the natural law exception. In Benson, the patent described had threatened to pre-empt every single usage of the BDC-binary algorithm, while the patent in Flook was limited to the petrochemical industry.

While this area of law seemed to be settled, the Court would take advantage of a statement in Flook – that “a process is not unpatentable simply because it contains a law of nature or a mathematical algorithm”19 – to revisit the matter just three years later in Diamond v. Diehr. The respondent in that case attempted to claim a patent on a method of calculating how long rubber should be left heated in a molding machine using the same chemical equation used by the respondent in Flook. Surprisingly, the Court held this time that Diehr’s patent was to be granted.

With all of the justices dissenting in Flook now in the majority, the Court in Diehr subtly disregarded the holding of that earlier case while pretending that its new holding was completely consistent with Flook.20 Recalling Benson, the Court reiterated its holding that ideas, as well as “manifestations of nature”, are not patentable.21 However, it then decided that the “recent holdings in Gottschalk v. Benson…and Parker v. Flook…stood for no more than these long-established principles”22, seemingly oblivious to Flook’s crucial requirement of inventiveness. Further eroding the decision in Flook, the Court then dismissed the analytic dissection test. The Court in Diehr held to the contrary that it was “inappropriate to dissect the claims into old and new elements and then to ignore the presence of the old elements in the analysis.”23 While the novelty of the invention less its algorithmic parts was crucial to the analytic dissection test, the Court here held that novelty was irrelevant to patentability under § 10124, and thus declined to rule on the matter. “It may later be determined that the respondents’ process is not deserving of patent protection because it fails to satisfy the statutory conditions of novelty…or nonobviousness,” it wrote.25

Instead, the Court once more adopted the machine-or-transformation test first announced in Benson. Software containing a normally unpatentable mathematical formula could be patented when it “implements or applies that formula in a structure or process which, when considered as a whole, is performing a function which the patent laws were designed to protect (e.g. transforming or reducing an article to a different state or thing).”26 As Justice Stevens pointed out in his dissent, “Diehr…does not claim to have discovered anything new about the process for curing synthetic rubber.”27 Yet, under the majority’s analysis, the embedding of the algorithm in a transformative process that vulcanizes rubber had somehow effected a transformation, allowing it to be shielded from the natural law exception, clearing the way for patents to be granted on software containing algorithms that performed precisely those transformative actions.

The Court did not revisit these issues for almost three decades, and its pronouncements on software patentability from the patent-eligibility trilogy have stood unchanged until the 2010s. While Benson denied that bare algorithms could be patented and Flook required a consideration of the novelty of software that contained algorithms, Diehr liberalized those requirements. By adopting the lenient machine-or-transformation test, and continuing to affirm it as a “useful and important clue” for determining patentability under §101 as recently as 2010 in Bilski v. Kappos,28 the Court opened the way for the patent office29 and the lower courts30 to usher in an era of widespread software patentability.31

Prometheus and Precedent

Mayo v. Prometheus, the Court’s latest decision regarding the patentability of abstract ideas contained within larger claimed inventions, is superficially a biomedical case. However, as the foregoing consideration of Flook and Diehr has shown, the Court’s pronouncements in this area are likely to have a significant impact on whether software can be patented.

In this most recent case, Prometheus patented a way to determine whether a particular dosage of a drug was too low or too high by measuring the levels of chemical substances formed as breakdown products in the blood.32 The correlation between these chemical substances and drug dosage is a naturally occurring one, and Prometheus claimed a patent on a “method of optimizing therapeutic efficiency.”33 The patent concerned the process of injecting a patient with the drug then comparing the measured amounts of substances against its table indicating whether the measured dosage was too low or too high.

Prometheus’s patent was declared to be invalid in a unanimous decision by the Supreme Court. The reasoning behind the Court’s decision is unfortunately flawed, and stands in the same relation to Diehr as Diehr stood to Flook: a reinterpretation of an earlier decision that purports to be consistent with precedent but in fact subtly changes it. Nevertheless, Prometheus is distinguished from the other cases by being the first case in which the Supreme Court inserts itself into a debate about policy. Where in the earlier cases it was eager to defer to Congress for further guidance,34 in Prometheus it actively inserts itself into the patent debate and goes to great lengths to argue that its decision is correct because it fulfills the policy objectives of patent law.

The conclusion that the Court reaches in Prometheus is an attempted reconciliation of Diehr and Flook, though the Court refuses to admit that it is a reconciliation, instead stating that both of “the controlling precedents…Diehr and Flook…reinforce our conclusion.”35 For a process containing an abstract idea to be patentable, the Court holds in Prometheus that it must have “additional features that provide practical assurance that the process is more than a drafting effort designed to monopolize the law of nature itself.”36 Thus no process that consists of only additional “well-understood, routine, conventional” steps which “when viewed as a whole, add nothing significant beyond the sum of their parts” can be patented.37

This holding is largely compatible with the key holding of Flook. Like Flook, which held that inventiveness of the process surrounding an abstract concept is a necessary condition for patentability, Prometheus holds equivalently that no conventional process surrounding the abstract concept can lead to patentability. Secondly, the Court was careful to emphasize that even an attempt to patent pure algorithms without pre-empting all usage of them was not acceptable. “Our cases have not distinguished among different laws of nature according to whether or not the principles they embody are sufficiently narrow,” the Court writes, finding Flook held, inter alia, a “narrow mathematical formula unpatentable.”38

In attempting to fit Diehr into this new framework, however, the Court exercises dubious logic to reinterpret Diehr as falling comfortably within the framework’s boundaries. Diehr, the Court notes, “nowhere suggested that all these steps” (of curing the rubber) “or at least the combination of those steps, were in context obvious, already in use, or purely conventional.” From this it then concludes that “these other steps apparently…transformed the process into an inventive application of the formula.”39 This is manifestly untrue. In deciding the case the Court in Diehr explicitly reserved its judgment on whether the rubber curing process was in any way novel not because it meant to imply that they were original, as the Prometheus court suggests. It did so because it found novelty concerns to be entirely irrelevant to the question of whether an invention was patentable under §101.

In sum, Diehr held that the question of “whether a particular invention is novel is ‘wholly apart from whether the invention falls into a category of statutory subject matter.’”40 In Prometheus the Court recognizes instead that “in evaluating the significance of additional steps, the §101 patent-eligibility inquiry and, say, the §102 novelty inquiry might sometimes overlap.”41 It finally announces that it has “neither said nor implied that the machine-or-transformation test trumps the law of nature exclusion,”42 despite Diehr having unleashed an era of jurisprudence meaning something close to that.43 It cannot be said that the Court has adhered faithfully to precedent.

Prometheus and Policy

What makes Prometheus an inspired decision, however, is the Court’s engagement with the philosophy of patent law in justifying its decision. The Court appears less interested in following precedent – as explained above – than in ensuring that the objectives of patent protection are served.

The Court conceives of patent law as the traditional tension between providing “monetary incentives” to the innovator “that lead to creation, invention, and discovery” while not impeding “the flow of information that might…spur…invention.”44 In the absence of specific guidance from Congress, however, the Court hints that the latter is the more important objective with regards to abstract principles. “Even though rewarding with patents those who discover new laws of nature…might well encourage their discovery…there is a danger that the grants of patents that tie up their use will inhibit future innovation premised upon them, a danger that becomes acute when a patented process amounts to no more than an instruction to ‘apply the natural law.’”45

The foregoing analysis suggests that Prometheus, as a precedent, is likely to have significant effects on the future of the age of software patentability that began with Diehr. Unlike the divided decisions in Flook (6–3) and Diehr (5–4), the reconciliation of the patentability of abstract concepts in Prometheus was achieved by a unanimous vote. Secondly, the Court in Prometheus effectively adopted the less lenient patent regime described in Flook and dismissed the principles laid down in Diehr. Finally, its considerations on patent law policy suggest that its decision-making will likely place the need for a free flow of innovation and ideas above the competing interest of inventors to be rewarded. When, at the end of its decision, the Court cautioned that “we must hesitate before departing from established general legal rules lest a new protective rule that seems to suit the needs of one field produce unforeseen results in another”, the Court is almost certainly referring to the software industry.46 While the biotechnology industry might wish to have a lenient standard for the natural law exception so companies working on genes might justly patent them, a similar standard applied to algorithms in the software industry would deprive developers of the basic tools of their work.

The decision in Prometheus is a step in the right direction and is to be commended. Patents in the software industry indeed do not function to protect inventors as well as might be hoped. The major players in the software industry engage in what is known as ‘defensive patenting’, in which “competitors within an industry each build significant patent portfolios and then threaten to use those portfolios in response to offensive patent threats.”47 In such a situation, new software developers who manage to somehow afford the cost of applying for and receiving a patent are still likely to be threatened by large software companies with patent portfolios. This is the case because it is probable that some part of their software infringes upon one or more of the tens of thousands of vague patents such large software companies have. These new developers will likely have to sign a cross-licensing agreement to avoid a lawsuit, allowing competition to proceed “just as if there were no patents.”48 In other words, the new developers’ original patent might “protect” them “from competition from you or me, but not …megacorporations.”49

Worse still, the speed of technological development in the computer industry prevents an inventor from realistically using patent law to protect his innovations. Unlike biotechnological developments – the approvals of which proceed glacially due to the FDA – “by the time an inventor has filed for, obtained, and enforced a patent, the software industry has moved on.”50 One study found that an average of 12.3 years elapsed between the time at which an inventor filed a patent to the time a court determination was made on the final validity of a patent.51 While it is true that these figures were for all patents, rather than just software patents, there is no evidence to suggest that software patents are accorded any expedited treatment in the courts. Moreover, the youth of a great number of software companies52 would ensure the validity of this argument if the system of civil trials does not undergo dramatic change.

Further empirical evidence supports the assertion that patents do not serve to protect innovators in the software industry. In a survey conducted in 2008, researchers found that a mere 24% of software or internet startups had filed for any patents at all.53 The software companies that did hold patents considered those patents just as important as a means of enhancing the company’s reputation or product image as they were a way to protect the company’s innovative ideas.54 Another study found, too, that “the number of lawsuits involving software patents had more than tripled since 1999”, representing a “substantial increase in litigation risk and hence a disincentive to invest in innovation.”55 Yet it also found that “the probability that a software patent is in a lawsuit has stopped increasing” as of 2011 due to judicial decisions reining in the “worst excesses” of software patents.56


The Supreme Court’s decision in Prometheus is a logically dishonest but also brilliant accomplishment for patent law. Despite the protests of the biotechnology industry, the Court is right to have considered that the patent system, whatever its flaws, currently applies equally to all industries. By taking this into consideration and reasserting the natural law exception neglected since Diehr, the Court rightly acknowledges the need to protect the basic tools of computer programmers from the closed doors of patents.

Prometheus will scarcely be the last pronouncement from the Supreme Court on the issue of software patentability – it is, after all, a biotechnological case. Yet the Court has, after three decades of confusion and chaos, taken a step in the right direction for the software industry, and has opened the path to future decisions that continue to encourage invention and innovation for computer technology.

Gareth Tan is a fourth-year in the College majoring in Political Science.

  1. See Stallman, Richard. Free Society, Free Software: Selected Essays of Richard M. Stallman, Second Edition. Free Software Foundation (2010), p. 141 for an excellent overview of what software patents can cover. Automatic expansion of text snippets in word processors, natural order recalculation in updating spreadsheets, file formats such as MPEG2, parts of operating system kernels and (of course) algorithms like LZW (see infra note 31) were patented. ;↩

  2. 35 U.S.C. §154. ;↩

  3. Knuth, Donald. The Art of Computer Programming, Volume I: Fundamental Algorithms, p. 4. (1997) Third edition, Addison-Wesley Professional. ;↩

  4. Mayo Collaborative Services, et al. v. Prometheus Laboratories, Inc. No. 10–1150, 566 U.S. __ (Sup. Ct., Mar. 20, 2012) ;↩

  5. 35 U.S.C. §101. ;↩

  6. 35 U.S.C. § 100. ;↩

  7. Gottschalk, Acting Commissioner of Patents v. Benson, et al., 409 U.S. 63 (1972). ;↩

  8. Parker, Acting Commissioner of Patents and Trademarks v. Flook, 437 U.S. 584 (1978). ;↩

  9. Diamond, Commissioner of Patents and Trademarks v. Diehr, et al., 450 U.S. 175 (1981). ;↩

  10. Benson, 409 U.S. at 73. For example, the binary coded decimal 0101 0011 (5 3) could be converted to the binary number 110101 (53). ;↩

  11. Id. at 71–2. ;↩

  12. Id. at 67, quoting Le Roy v. Tatham, 55 U.S. 156, 175 (1852). ;↩

  13. Id. at 70. ;↩

  14. Flook. 437 U.S. at 597. ;↩

  15. Id. at 594. ;↩

  16. Id. at 592. ;↩

  17. Id. at 591, 594. ;↩

  18. Id. at 590. See Bilski v. Kappos, infra at p. 15–16, holding that merely adding instructions to use well-known random analysis techniques was “token postsolution” activity. But see Diehr, supra at 187, finding all the processes surrounding a algorithm used in a very similar rubber curing process to not be insignificant postolution activity. See also Novick, Mitchell P. and Wallenstein, Helene. “The Algorithm and Computer Software Patentability: A Scientific View of a Legal Problem” 7 Rutgers Journal of Computers, Technology, and Law 313 (1979), p. 326, which discusses the difficulty of distinguishing the facts of Flook and Diehr. The first was held to have insignificant postsolution activity and the latter was not, demonstrating the problems with determining what precisely sufficient postsolution activity is. ;↩

  19. Flook, 473 U.S. at 590. ;↩

  20. The appeals courts were baffled for the next two decades, and continue to be confused by the Court’s rulings. See cases such as In re Alappat, 33 F.3d 1526, 1545 (Fed. Cir. 1994), State Street Bank & Trust Co. v. Signature Financial Group, 149 F.3d 1368 (Fed. Cir. 1998)., and In re Bilski, 545 F.3d 943 (Fed. Cir. 2008), in which the Federal Circuit adopted and discarded numerous tests for software patentability. ;↩

  21. Diehr, supra at 185 (quoting Diamond v. Chakrabarty, 447 U.S. 303, 309 (1980) (quoting Funk Brothers v. Kalo Inoculant, 333 U.S. 127 130 (1948))). ;↩

  22. Id. at 185. ;↩

  23. Id. at 188. ;↩

  24. Id. at 189. ;↩

  25. Id. at 191. ;↩

  26. Id. at 192. It may be observed that this test seems to make the Benson algorithm patentable. After all, early computers had to transform their magnetic storage tapes from one state to another when performing calculations on stored data. This similarity may explain in part why patents were actually granted on algorithms after Diehr (see infra note 31), though much of the impetus must have come from Diehr being the first Supreme Court case in which a software patent was approved as patentable. ;↩

  27. Id. at 206. ;↩

  28. Bilski et al. v. Kappos, Under Secretary of Commerce for Intellectual Property and Director, Patent and Trademark Office, 561 U.S. _, slip opinion p. 8. See also Lemley, Mark A. and Risch, Michael. “Life After Bilski”, 63 Stanford Law Review 1315 (2011), p. 1316, noting that because the Supreme Court failed to suggest alternatives the machine-or-transformation test after Bilski threatened to become “effectively mandatory.” ;↩

  29. Samuelson, Pamela. “Benson Revisited: The Case Against Patent Protection for Algorithms and other Computer Program-Related Inventions”, 39 Emory Law Journal 1025 (1990), p. 1093. ;↩

  30. Bessen, James and Hunt, Robert M. “An Empirical Look at Sofware Patents.” 16 Journal of Economics & Management Strategy, No. 1, 157 (2007) p. 160. ;↩

  31. Unfortunately, patents were also granted on algorithms disguised as software. For the most infamous example, see U.S. Patent 4,558,302 (1985) on the LZW compression algorithm, best known for being used in the animated GIFs flooding Tumblr today. ;↩

  32. I.E., the suitability of a thiopurine dosage for autoimmune disorders can be measured through correlations with the metabolites 6-thioguanine and 6-methyl-mercaptopurine. ;↩

  33. Prometheus, supra note 4, p. 5 of slip opinion. ;↩

  34. See Benson, 409 U.S. at 72 (“It may be that the patent laws should be extended to cover these programs, a policy matter to which we are not competent to speak.”), Flook, 437 U.S. at 595 (“Difficult questions of policy concerning the kinds of programs that may be appropriate for patent protection…can be answered by Congress on the basis of current empirical data…”), Diehr, 450 U.S. at 182 (looking to Committee Reports to justify its understanding of Congress’ attitude towards software patentability). ;↩

  35. Id., p. 11. ;↩

  36. Id. ;↩

  37. Id. ;↩

  38. Id. at p. 20. ;↩

  39. Id. at p. 12. ;↩

  40. Diehr, 450 U.S. at 190. ;↩

  41. Prometheus, 566 U.S. ____, p. 21. ;↩

  42. Id., p. 19. ;↩

  43. See the cases in, supra, note 20. ;↩

  44. Id., p. 23. ;↩

  45. Id., p. 17. ;↩

  46. Id., p. 7. ;↩

  47. Id. ;↩

  48. Stallman, supra, note 1, p. 152. ;↩

  49. Id. ;↩

  50. Burk, Dan L. and Mark A. Lemley, The Patent Crisis and How the Courts Can Solve It, p. 57 University of Chicago Press, 2009. ;↩

  51. Allison, John R. and Mark A. Lemley, “Empirical Evidence on the Validity of Litigated Patents.” 26 AIPLA Quarterly Journal No. 3 185 (1998), p. 235. ;↩

  52. For example, according to their Wikipedia pages Evernote was founded in 2008, Foursquare in 2009, Uber in 2009, and Instagram in 2010 (;;; (last accessed December 28, 2012) ;↩

  53. Graham, J.H. et al. “High Technology Entrepreneurs and the Patent System: Results of the 2008 Berkeley Patent Survey.” 24 Berkeley Technology Law Journal, Vol 4, p. 1277. ;↩

  54. Id., at 1301. ;↩

  55. Bessen, James. “A Generation of Software Patents”, Boston University School of Law Working Paper No. 11–31, p. 20. ;↩

  56. Id. ;↩

Read the article:
Prometheus and the Future of Software Patentability