Search the archive.

China and Russia employ cyberattacks and other illegal means to steal U.S. intellectual property.

China’s aspiration to achieve technological dominance is understandable. Many of the market-distorting measures Beijing undertakes to build up its domestic technology sector, such as subsidies and tariffs, are employed by other industrial nations even though they are in conflict with trade commitments. What puts China in a class by itself, however, is the extensive, state-supported use of illegal means to secure access to the intellectual property and trade secrets of U.S. technology companies. As of mid-2020, the U.S. Federal Bureau of Investigation (FBI) had over a thousand cases of technology theft by China under investigation.

While we claim to be competing, there are two giant holes through which America helps China develop technologies that are potentially vital to winning the competition. The first is our failure to respond to theft of intellectual property (IP). Even if claims of trillions in IP losses are exaggerated, the amount is certainly in the hundreds of billions. American retaliation against PRC beneficiaries of coerced or stolen IP is nearly non-existent, guaranteeing more theft.13

Outbound investment is the second. The US attempts to protect technology embedded in goods sales through the export control regime, though the regime has its own problems. For inbound investment, the Committee on Foreign Investment in the United Sates (CFIUS) has a solid track record in preventing loss of sensitive technology through acquisition here. Yet the same technologies that are controlled in exports or shielded off from acquisition can be developed in China, with American funding.

The Patent Eligibility Restoration Act also fails to achieve its purpose of bringing clarity to the subject matter eligibility standard. The proposed legislation cites the supposed “extreme confusion and uncertainty” surrounding the current subject matter eligibility standard as its reasoning for getting rid of the judicially created exceptions but then uses subjective terms in the proposed legislation without providing guidelines for their meanings, while also ignoring the confusion and uncertainty that are certain to result from wiping out 200 years of jurisprudence. Moreover, the Act’s use of language that is taken directly from some cases that the Act intends to overrule would lead to uncertainty, since the meanings of those words and phrases would be unsettled without caselaw to support them. This would lead to years of widespread confusion regarding the standard until the Court could interpret their meanings, and possibly beyond that point.

As previously stated, the Act also asserts that the judicial exceptions should be eliminated because their application by judges has caused “extreme confusion and a lack of consistency” within the judicial branch of the federal government and federal agencies, and among patent practitioners.249 However, as discussed previously, independent analysis shows district court and PTO judges are applying the Alice/Mayo framework accurately and consistently arriving at the correct outcome for correct reasons.250 Further, Federal Circuit judges seldom disagree with one another on the outcomes of subject matter eligibility cases.251 Accordingly, it seems that Federal Circuit judges are not actually confused by how to apply the standard, nor are they applying it inconsistently. So why trade this system that seems to be working well at the moment for a completely new one that would introduce a new set of problems?

The U.S. Patent and Trademark Office (USPTO) might be the most powerful government agency the majority of Americans have never heard of, much less think about on a daily basis. The USPTO is housed within the Department of Commerce, and it is charged with registering trademarks and granting patents. Its funding comes entirely from application fees, and its task as outlined in the Constitution is to promote the progress and dissemination of new technology. Its decisions therefore enhance or impede people’s access to medical care, agricultural inputs, automobile components, and thousands of other things.

Alas, the Patent Office has often neglected to serve the public’s interest—and seldom has it performed worse than under Donald Trump’s choice of director, the corporate lawyer Andrei Iancu.

Two of the biggest supporters of these bills are former U.S. Patent and Trademark Office (USPTO) directors David Kappos and Andrei Iancu. This is no surprise. Since leaving their roles, Kappos and Iancu have committedly opposed efforts to remodel our clearly anti-competitive patent system. What’s more, the former directors advocate loudly for changing the law while remaining silent about the corporate interests they represent. After leaving the USPTO, David Kappos joined the partnership of Cravath, Swaine & Moore, which represents Amgen, Johnson & Johnson, and Novartis, among other pharma heavyweights. Iancu is now a partner at Sullivan & Cromwell, which lists AbbVie, Bayer, and Merck among its life sciences clients. He also founded the Council for Innovation Promotion as an outfit to more systematically support the entrenchment of the current anti-competitive system.

The former directors’ current roles reveal the other primary beneficiaries of the proposed bills: law firms and lawyers. Patent litigation is big business, with costs ranging from $2 to $4 million per case on average. But recently, patent litigation rates have begun to decline sharply. Given the business they stand to lose, it’s no wonder that law firm partners are pushing for changes that would ensure more patents are granted and litigated in court. Lawyers have every right to advocate for their own interests and those of their clients. But when they do, they should be transparent about it, instead of using the veneer of their former roles in the public sector to paint their positions as anything other than advocacy on behalf of their corporate clients’ interests.

At the outset, it cannot be ignored that § 101 comprises a substantial portion of invalidations at the district court. Professor Stephen Yelderman's recent study found that § 101 was responsible for 509 invalidation "events" (any time at least part of a patent was invalidated) or 33% of all such events in federal district courts between January 1, 2011 and June 30, 2017.227 By comparison, §§ 102 and 103 combined for 491 events or 31.8%.228 Even more so, § 101 invalidation events increased significantly after Alice, rising from less than 60 in 2014 to more than 160 in 2015.229 In contrast, §§ 102 and 103 began to steadily decrease after 2014.230 Using more recent data reveals similar results regarding § 101 invalidation events. From June 2014 through December 2022, there were 506 § 101 judgment events, comprising 39.4% of all invalidity judgment events, whereas §§ 102 and 103 comprised only 24.2%.231 As shown in Graph 1, similar trends are observed for at least partial motion success (not including trial results) on invalidity grounds for §§ 101, 102, and 103 between 2011 and 2022. From 2014 to December 2022, there were a total of 1,458 patent eligibility motions in the federal district courts.232 Of those, only 581 were granted and another 167 were partially granted.233 Thus, since its issuance, § 101 has accounted on average for 83.1 successful motions per year that were at least partially granted. The number of § 101 challenges that were litigated to an appellate decision is also quite small: about 360.234 Nevertheless, Graph 1 confirms that § 101 is indeed a powerful tool frequently used (comparatively speaking) to invalidate patents in federal district court.

Importantly, however, as Professor Yelderman acknowledged in his study, those statistics do not provide the complete picture of how frequently parties utilize different statutory grounds to invalidate patents because they do not include invalidity challenges through inter partes review (IPR) under §§ 102 and 103 before the Patent Trial and Appeal Board (PTAB), which began in September 2012.235 With IPR, which became immensely popular after its implementation, challengers may attack patents only under §§ 102 and 103 not § 101.236 So, while there were a total of 491 invalidation events in district courts under §§ 102 and 103 in federal district court from January 1, 2011 to June 30, 2017, there were 1,353 patents that were found at least partially invalid under §§ 102 or 103 through IPR from its start in September 2012 to June 2017.237 Between April 2015 and September 2016 alone, IPRs invalided all instituted claims in 608 patents and some claims in 125 additional patents.238 In other words, IPRs invalidated almost as many patents in 16 months as Mayo/Alice invalidated in the last decade. By comparison, only a small number of patents (approximately 100) were invalidated under § 101 at the PTAB as part of covered business method (CBM) review.239 However, as of September 2020, CBM is no longer available, so it carries no future impact.240

When considering judicial outcomes, despite the relatively small number of Alice decisions, the substantial increase of § 101 attacks in court since 2014 has been fodder to paint an alarming picture. For example, critics of Alice often highlight the percentage increase in the number of invalidated patents under § 101 pre- and post-Alice: "Compared with the five years prior to Alice, there has been a 1056% increase in the number of decisions finding ineligible claims, and a 914% increase in the number of invalidated patents."224 Such statistics seem to suggest that Alice has entirely overrun patent litigation. Yet, relying on such percent increases presents a distorted view of the world because § 101 decisions in the decade prior to Alice were almost nonexistent. As others have written, the Supreme Court's 1980s decisions on patent eligibility along with the Federal Circuit's jurisprudence through the early 2000s "render[ed] the patentable subject matter [*38] requirement effectively a dead letter."225 The post-Alice percent increase reflects a comparison to the pinnacle of free-flowing patent eligibility that created the immense problem that necessitated an unprecedented four Supreme Court patent eligibility decisions in a matter of four years.226 Thus, the State Street "glory days" of patent eligibility are far from the normal state and should not serve as the appropriate benchmark for judging the "pre-Alice" world with its very few examples of ineligibility. Moreover, the number of pre-State Street eligibility challenges may also be a misleading metric to measure against post-Alice eligibility challenges because it predates the Internet boom and so much of the technology that serves as the target of Alice attacks. Rather than looking to how § 101 challenges have increased from a very different legal regime and state of technology, the better approach is to consider the current impact of Alice in the context of current patenting trends and other grounds for patentability.

Furthermore, it cannot be overlooked that §§ 102 and 103 present their own concerns for predictability to investors and researchers because invalidating prior art can be secret or difficult to find. Thus, even if a patentee puts forth substantial efforts to ensure their invention is patentable under §§ 102 and 103, it might very well be found to not satisfy one or both of those provisions in later litigation. As Professor Yelderman explained in his study, "somewhere between 18% and 35% of all obviousness invalidations relied on patents that were secret at the time of filing."278 Further taking into account prior art references that would be very difficult to find, "somewhere between a third and a half of obviousness invalidations [draw] on prior art not easily accessible to a person of skill in the art at the time of the invention."279 This is an important finding. Up to half the time a patent is invalidated under § 103, the basis for invalidity would not have been known to the patentee when applying for a patent. "An exhaustive search of all worldwide patents and traditional publications combined with perfect appreciation of those and other already cited references would preclude between 52% and 55% of all district court invalidations."280 Thus, even if § 101 was removed entirely, it can hardly be said that patentees or patent decision-makers should feel too confident that their patent is valid prior to being tested in litigation.

The joint letter opposing PERA directly addresses testimony provided at a recent Senate hearing on patent eligibility. For example, it describes a recent ACLU study showing that current patent eligibility law is clear, consistent, and predictable. According to the study, from 2013 through 2020, the Federal Circuit affirmed patent eligibility decisions at a higher rate than the circuit’s average and at a higher rate than validity decisions under other sections of the Patent Act. Some practitioners may not like the outcomes of patent eligibility decisions under existing law, but the evidence leaves no doubt that those outcomes are consistent and predictable.

Moreover, the impact of patents falling to Alice is usually further overstated without taking into account which of those patents would have failed to satisfy other patentability requirements when tested under rigorous litigation. Stated differently, the number of Alice-doomed patents that otherwise would have resulted in valid inventions is an even smaller piece of the small pie noted above. Some of these patents certainly also would have been found invalid under §§ 102, 103, or 112. Thus, in measuring the impact or "harm," as some might characterize it that Alice brought to patent litigation, we should include not all patents invalidated under Alice, but instead, the patents invalidated under Alice that would not have been invalidated on other grounds. As Professor Saurabh Vishnubhakat previously explained, § 101 has been characterized as "a significant doctrinal shortcut to the other requirements for patentability."243 The impropriety of using Alice as a "shortcut" is certainly well-founded and presents concerns about the administration of patent disputes, but its use as such does not result in unjustified harm to the patent owners whose patents were nonetheless invalid on other grounds or at least, those patent holders should not be heard to complain that § 101 deprived them of patent rights that otherwise would have existed.

Undoubtedly, as highlighted here, § 101 became more frequently utilized in court proceedings after Alice and continues to play a meaningful role in patent challenges whether through judicial decisions or voluntarily resolved disputes. But it is only one ground of invalidation upon which defendants typically rely in attacking the validity of a patent. Thus, without taking into account § 101's utilization relative to other invalidity grounds (including through IPRs) and the more limited "harm" when § 101 is the sole basis for invalidity, the true impact of § 101 in patent litigation is likely to be overstated.

With regard to the impact at the Patent Office, another recent empirical research study found that there was no reason to conclude "Alice had a causal effect on receiving § 101 rejections among software patent applications in general" and no evidence that "Alice increased the barrier of patentability for [many] technology areas in software."93 A different study by the USPTO analyzed the impact of Alice in patent prosecution and found that "the USPTO's efforts to clarify the Alice standard have substantially offset the uncertainty created by Alice."94 For example, the USPTO found that after issuance of its 2019 patent eligibility guidance document, in "Alice-affected technologies," applicants were 25% less likely to receive a first office action rejection under § 101, which amounted to a lower rejection rate than before Alice issued.95 A comparison between 2019 and 2014 may not be appropriate, however, since the set of patent applications after the 2019 guidance is different from the 2014 set of applications prior to Alice, and the relative decrease in rejections may have been influenced by factors other than the guidance, such as a decrease in Alice-susceptible applications.96 Nevertheless, the data shows a marked drop in rejections immediately after the guidance document issued as well as an overall lower § 101 rejection rate compared to before Alice issued.97 Moreover, the USPTO's research also found that the guidance document reduced examination uncertainty measured by the variability of § 101 first office action rejections across different patent examiners98 by 44% for Alice-affected [*15] technologies, which also was substantially below pre-Alice levels.99 This evidence suggests that patent applicants can expect more predictability in § 101 examination determinations than prior to the issued guidance, and perhaps even prior to Alice.

I. The current patent eligibility jurisprudence results in predictable and appropriate outcomes

The current test for patent eligibility is both clear and in line with historical and global standards for patenting: to be eligible under § 101, a patent must reflect the practical application of an advance in technology. This notion traces its origins back to the U.S. Constitution itself, in which the Framers authorized the grant of patents for the purpose of promoting the progress of the “useful arts,” which has always been understood to mean technology.1 Shortly after the Constitution’s adoption, this understanding was confirmed in one of the very first Patent Acts: the Patent Act of 1793 defined patent-eligible subject as “any new and useful art, machine, manufacture or composition of matter”—in other words, the means for the practical implementation of an advance in technology.

d. In other words, how much is the current uncertainty costing our economy in terms of jobs, innovation, and development?

The premise of this question is simply false. As discussed in my written testimony, there is no reliable evidence of any significant uncertainty or economic cost. I cited USPTO’s own rejection rates, the USPTO’s own internal quality metrics, an empirical analysis by respected academics, and various other objective indicators that suggest there is no problem. In contrast, those contending that substantial uncertainty exists have relied on statements of opinion, press articles, and an outdated opinion survey. Given that they have had a decade to do so, I would assume that these advocates would have produced objective evidence supporting their position if such evidence existed.

As I noted in my testimony, the USPTO has published a study examining the application of section 101 that considered both patent-eligibility rejection rates and the variability of examiners’ eligibility determinations.3 This study found by 2020, both rejection rates and examiner variability were lower than they were before the Alice case was decided.4 I also cited an academic study that examined all 368 patent eligibility decisions that were made by the Federal Circuit between 2012 and 2022.5 The authors found that patent eligibility decisions by district courts and the Patent Trial and Appeal Board are affirmed at the high rate of 87%, indicating that eligibility determinations are (at least according to the Federal Circuit) overwhelmingly correct. Perhaps more importantly, the study also found that there was relatively little evidence of disagreement among Federal Circuit judges regarding how to apply the Supreme Court’s patent-eligibility jurisprudence. As the authors noted in a summary of their study, “under one of the most well-established metrics for measuring the predictability in the law, § 101 proved to be more predictable than other areas of patent law over the past decade.”6 In my oral testimony, I further noted that the USPTO’s own internal quality metrics indicate that Section 101 is the most accurately applied by examiners of the major statutory requirements.7 I also cited a number of additional studies that indicated, e.g., that Alice resulted in increased R&D investment8 and that there was “a positive association between Alice and both R&D spending by software firms and patenting by firms that held relatively more software patents prior to the Court’s opinion.”9 Finally, in previous testimony, I cited additional data and studies.10

In contrast, Director Kappos’s testimony cited a trade press article, a student note, and an academic article reporting the results of an opinion survey conducted in 2017.11 Notably, Director Kappos and others continue to cite this opinion survey prominently, despite the fact that venture capital investment data are now available for the seven years since the survey was conducted. Presumably, the proponents of PERA would cite the actual data rather than an outdated opinion survey if those data supported their contentions.

One of the concerns frequently raised with regard to a narrower scope of patent protection for AI as with any software is that investors will be hesitant to provide capital without powerful monopoly rights to exclude others from practicing the inventions, particularly with regard to larger companies that can rely on their market dominance for competition.263 Although there is no shortage of theoretical arguments and opinions on Alice's impact on investment, as Professor David Taylor remarked in his recent study of the issue, "[e]xisting literature provides surprisingly little data even to begin to answer [the question]."264 Generally, the argument is fundamentally sound and well-founded: the less power a company has to exclude others from the marketplace, the less attractive the company is for investment, since any other market player can swoop in to overtake their place. Existing empirical evidence also supports this proposition, generally, at least with regard to some industries.265 However, factors other than patent protection often play a key role, and thus, details very quickly complicate the analysis.

Although Professor Taylor's study found "negative impact of the Supreme Court's eligibility cases generally on investment," it also found that "most investors do not report changing their investment decision making based upon the Supreme Court's eligibility decisions," though a significant portion of those investors were also not informed about the relevant decisions.266 One additional key finding from the study was that "[t]he availability of patents... was not the most important consideration to the investors."267 Instead, "[t]he quality of a target [*47] company's people ranked as most important, followed by the quality of the company's technology and the size of the potential market for the technology."268 Only 6% of respondents identified the availability of patents as the first or second most important factor in deciding whether to invest.269 Additionally, Professor Taylor's second key finding was that "reduced patent eligibility correlates with particular investment behaviors in particular industries."270 As one example, in contrast to biotechnology, "[i]nvestors overwhelmingly indicated... that the elimination of patents would either not impact their firms' decisions whether to invest in companies or only slightly decrease investments in companies developing technology in... software and Internet (80%)."271 In light of Professor Taylor's important findings, perhaps it is not surprising that other evidence also shows that investment and private equity funding for AI, in particular, has continued to increase substantially, notwithstanding the concerns over the patentability of software and AI raised by Alice.272 More so, some evidence shows that when AI startups are acquired there appears to be little focus on their existing patent portfolio,273 suggesting that a lack of patent protection in the AI space, at least at this time, is not as detrimental as might be the case in other industries. Further empirical research in the context of AI companies is necessary.

As the Supreme Court is the only court that can overturn its precedent, absent intervention from Congress, the Alice framework will continue to govern the analysis of whether an invention is patent eligible. However, lower courts led by the Federal Circuit have and will continue to refine the boundaries and application of the Alice framework. The USPTO also has taken steps to provide clarity in this area of law by promulgating iterative guidance to those seeking a patent, including example AI claims that are patent eligible.84 Importantly, such guidance while helpful to practitioners and examiners comes with an important disclaimer, as it is not "the law of patent eligibility, does not carry the force of law, and is not binding on" the courts.85

Despite the criticisms, and the need for further clarity regarding the application of the Alice framework, some evidence suggests that the law is much more predictably applied now86 (and perhaps always has been),87 including with regard to AI patents.

4. One of the key concerns from innovators is that, absent additional clarity in this space, we’re going to start seeing American companies start developing their inventions overseas in jurisdictions which have broader standards of patent eligibility. Do you agree with that concern and, if you do, what evidence have you seen to suggest that technological inversion is already occurring?

The TRIPS agreement requires that all signatories treat foreign inventors and patent owners exactly the same as domestic ones. This means that an American company would gain absolutely no legal advantage from moving its development activities to another country because it would be treated no differently by the U.S. or any of the other 163 signatories. So, to the extent that this could theoretically be a problem, there would have to be a concern that American companies were moving development to one of the very small number of nonsignatory countries, which is simply not the case.

• Patent eligibility law does not systematically result in the rejection of patent applications in the U.S. that are granted in Europe and China.

A 2017 essay by Kevin Madigan and Adam Mossoff,34 which purported to report study results supporting this claim has been resoundingly debunked.

Nor does technological “leadership” guarantee military advantage. Other factors determine military success, the most important being political will, leadership, and strategy. An opponent that has an advantage in these areas will be able to resist and thwart a more technologically advanced opponent (the Taliban, for example). Advanced technology in the service of flawed strategy will not change outcomes. There is an assumption that technology provides an advantage and, in a contest, where other factors are equal, technological leadership can be critical, but in most situations this technological advantage is only one factor among many in determining effectiveness. Specific quantitative measures may not actually measure what we want to assess. It may be better to ask what nations want (wealth, international influence, military power) to determine the contribution of a basket of technologies.

These issues are not binary – an outcome where China makes all the shoes, and the United States makes none. We would need to consider both the number of shoes produced and their quality and type, and then ask how shoes contribute to wealth, power, and military strength and whether there are alternative technologies that can substitute for them (since under duress, nations are inventive). Answering these questions requires a broader view than that taken in many reports that looks at the overall economy and its technological base. The best predictors remain overall national revenue and tech market share.

The keys to technological leadership are at the most fundamental level, access to capital and to ideas (both the ability to create new ideas and to use them). Better measures of success include market share (meaning someone actually buys what you are making), the number of startups (particularly ‘unicorns’), revenue, and the long-term trend in national income. These revolve around the central importance of the ability to commercialize research and innovation. This is an area – entrepreneurship - where the United States has had an advantage over Europe and China, both of which score well on proxy innovation metrics but less well for outcomes. The biggest risks to the U.S. innovation system are political dysfunction and regulatory ideologues. These are domestic problems, not the result of international competition, but is easier to fault China than in the mirror.