While small firms often have trouble getting patents, large firms are under fire for getting too many patents—for example, covering different versions of the same pharmaceutical drug. AI-enabled innovation has also generated concerns about a flood of patent applications for items that are obvious variations of existing innovations. This February, a Senate committee heard testimony about how marginal patents are being used to delay competition and drive up prescription drug prices.
All of this prompts a question that Mark Lemley, Doug Lichtman, and Bhaven Sampat asked in the title of a Regulation article nearly 20 years ago: “What to Do About Bad Patents?” (Winter 2005–2006).
Quality vs. access / If the problem is low-quality patents crowding out high-quality ones, the solution would seem straightforward: increase the quality of patents before they are issued. And indeed, the USPTO is working to do that. The agency has begun to devote more time to the upfront examination of patents and, in a 2022 Federal Register Request for Comment, signaled its exploration of new responsibilities to ensure that applications filed are “robust and reliable” through greater clarity and support for submitted claims.
But there’s a catch: there is a tension between high patent quality and broad access to the patent system for first timers and entrepreneurs like Parrott. That’s because improving patent quality takes time and money. It takes effort on behalf of patent applicants to define the terms they’re using and clearly delineate what they consider the invention. It takes time to do prior-art searches and ensure the claimed subject matter is truly novel and nonobvious. And solutions like the one Lemley et al. proposed in Regulation of “paying for a thorough examination of their inventions” take, well, money.
Unsurprisingly, members of the patent bar, through their responses to the Federal Register notice, have expressed concern over the prospect of having to do more work for an initial patent filing while getting paid the same price of around $10,000 that has been in place for over 20 years. Some have lamented the disproportionate burdens expected on first timers and under-resourced inventors. The data show that women, underrepresented minorities, and self-represented applicants already experience less success in converting their applications into patents. Raising the examination bar risks further exacerbating these disparities and the extreme skew in the distribution of who gets patents. In 2020, 50 percent of new US patents went to the top 1 percent of patentee.
Raising additional hurdles to patenting by first-time and underrepresented innovators is problematic given the stakes. Patenting by new entrants unlocks hiring, financing, and social mobility. And the real possibility of patenting incentivizes invention in the first place. If women, underrepresented minorities, and people of low-income backgrounds invented at the same rate as wealthy white men, researchers estimate, there would be four times as many inventors in America as there are today. For those reasons, the USPTO and its counterparts around the world are focused on increasing access to the patenting process and, in the words of Vidal, getting people off the bench and bringing innovation to impact.
Harnessing AI / To do so, Congress recently increased the discounts on USPTO application fees for small and “micro” applicants that fall under a certain threshold. Unfortunately, the bulk of the cost of filing for and successfully pursuing a patent is in the form of legal, not administrative, fees. To achieve the twin goals of patent validity and equity will require a different approach than USPTO discounts. Nor do we think it is tenable, as Lemley et al. argued, to just weaken the presumption of validity for the majority of filers.
We suggest a new approach: to dramatically promote the quality of patents, the USPTO should dramatically promote access to patent-quality technology and make it available to all who need it.
Patent law, and in particular 35 U.S.C. §§102 and 103, requires innovations to be truly novel and nonobvious to be eligible for patent protection. But too often patent applications are filed without adequate attention to what similar innovations came before, and what will be required of the present application to qualify. These prior-art searches are important because they save time and money downstream by reducing the risk that an application ultimately won’t pass muster.
Patent claims must also, under 35 U.S.C. §112, be supported by the language of the body of the patent application. But too often, it is less than clear what words in a claim actually mean, especially when they don’t otherwise appear in the specification. Other errors abound, such as invalid claim numbering, grammatical errors, and inconsistencies between the drawings that are part of the patent specification and the words describing those drawings. That wastes examiner and applicant time. Unfortunately, previous work by one of us in the Iowa Law Review demonstrated that such errors disproportionately befall small and micro entities.
But what is hard for humans is easier for technology and those who have access to it. In particular, tasks like conducting prior-art searches and ensuring consistency and clarity are increasingly being assisted by technology. A growing number of third-party vendors like ClaimMaster, Harrity, PatentBots, and others are utilizing artificial intelligence in its various forms to automate patent application drafting and error-check the results. Tools on the market can also assist in distinguishing filed patent claims from what came before, flag mismatches between the specification and the claimed invention, provide first-draft specification language, and place a patent application in a better condition for being granted.
The idea of providing access to “technology” to level the patent playing field is not new. The earliest patent system allowed applicants to submit their applications by mail, so that rural inventors would not be at a disadvantage. The USPTO eventually allowed submission of patent applications by fax and later facilitated e‑filings, electronic records access, and even remote interviews to broaden (and reduce the expense of) participation.
Continuing this trend, the USPTO recently adopted the DOCX format for patent applications, which uses XML that is both universally used and easily machine-readable. This format allows the USPTO to analyze the data and provide instantaneous feedback on common mistakes. Rather than waiting a year or more for a human patent examiner to kick the application back to the inventor for non-compliance, DOCX immediately notifies the inventor, regardless of sophistication or attorney representation, of specific errors and even provides guidance for remediation. Time is money. In future iterations of the tool, the USPTO could flag not only ministerial issues (e.g., forgetting to put a period at the end of a patent claim) but substantive ones (e.g., ensuring the disclosure requirements under §112, mentioned above, are met).
And to help applicants meet their burdens under §§102 and 103, the USPTO has floated the idea of using newly introduced AI search tools to help micro-entity applicants get their application in better shape before they start the evaluation process. Generative AI language processing systems being used in an increasing number of domains provide another set of potential tools for improving patent quality.
Treating expanded and equitable access to technology as an instrument for patent quality reform has at least three major benefits.
First, it works. Technology can improve patent application quality. Technology can provide not only pre-submission validation of the application’s contents, but also recommendations on how to correct an application’s shortcomings. The early assessment of the novelty and nonobviousness of a patent’s claims, before legal and filing fees are spent, should improve the quality of the applications filed.
Second, technology can improve the quality of patent examination. Automation via a universal document format like DOCX allows examiners to spend less time on identifying technical errors and resolving ambiguities, and more time on substantive patent questions such as novelty and nonobviousness. AI systems, as they become stronger and more reliable, can assist in finding the most relevant and salient prior art, ensuring all applications get a thorough review.
Finally, and most importantly, broad access to technology facilitates equity and efficiency. While enhanced patent quality benefits all, leveling the playing field through technology benefits the most those who have the least experience and are therefore likely to need the most help conforming their applications to the intricacies and technicalities of the law.
Going further / To be sure, the idea of the government developing and deploying technology may feel fraught. For example, the USPTO’s move toward DOCX and pre-application review has had its hiccups, prompting doubters. But technology improves. While change is hard, the USPTO’s past record, for example, of overcoming initial headaches to move to digital records through its Electronic Filing System (EFS-Web), demonstrates the benefits of pushing through. Lower costs, greater access, and the dramatic expansion of the ability to systematically analyze the patent system’s performance followed the transition from paper to e‑records.
Further, the USPTO need not necessarily become a technology vendor to dramatically increase access. Following the lead of the Internal Revenue Service in providing tools for tax preparation, for patent application preparation the USPTO could enter into public-private partnerships with software companies to make their products available at a discount or for free (as noted in the Iowa Law Review article listed at the end of this article). Or, like the IRS, the USPTO could explore certifying or authorizing certain vendors that provide assistive technologies whose services meet a certain quality threshold. It could also subsidize or support discounts like those offered by the USPTO to small and micro-entities, in effect supporting “means testing” between applicants. Price differentiation and sales to smaller filers who wouldn’t otherwise be able to afford access would enable technology firms to expand, rather than risk cannibalizing, their markets. A small-scale pilot undertaken with technology firms to test the value of doing so would be an excellent first step.
Another step the USPTO could take is to facilitate the development of customized technology solutions to meet the needs of less experienced filers. Rigorous analysis has shown that targeted assistance for novice users works and can reduce disparities in the patent grant process. The USPTO’s network of pro bono providers also reaches a distinct set of filers with specific needs. Development of effective assistive technology drawing upon what has worked for these populations in the past may go even further than general purpose technology to close the patent grant gap. Those interested and invested in equity could incentivize such special purpose technology through an advanced purchase commitment or prize mechanism.
At the very least, as the USPTO rolls out new requirements or technologies, it could pay more attention to how the changes affect new and small inventors given their unique needs. Rigorous evaluation and iteration can help ensure that the potential of patent quality technologies to improve the system for all is realized. As computer automation and AI create efficiencies in the patent system, the patent system should seize this opportunity to advance the twin goals of quality and equity through broad access, for the benefit of all inventors and the public at large.
Readings
- “Rigorous Policy Pilots the USPTO Could Try,” by Colleen V. Chien. Iowa Law Review 104(1): 1–30 (2019).
- “USPTO Pilot Program Reduced Disparities in Patenting.” Issue Note 102, Office of the Chief Economist, U.S. Patent and Trademark Office, November 2022.
- “What to Do About Bad Patents?” by Mark Lemley, Doug Lichtman, and Bhaven Sampat. Regulation