When Charles Townes and Arthur Schawlow filed for a patent on the laser, the US Patent and Trademark Office rejected the application repeatedly for more than three years. The invention combined ideas from optics and electromagnetism in ways examiners had rarely seen, and its strangeness was held against it. That story hints at a broader pattern: inventions that mash together ideas from unfamiliar corners tend to have a harder time getting through the patent system, even when they later prove transformative.
A new analysis published in PNAS asks whether this hurdle falls unevenly on men and women inventors, and if so, why. The answer that emerges is that the widely discussed gender innovation gap in patenting is not really a story about conventional inventions at all. It shows up almost entirely when women try to patent the boundary-pushing kind.
A question about who gets credit for offbeat ideas
Women earn about 60% of college degrees but account for only around 21% of patented inventors. Prior work has often attributed this imbalance to stereotypes and bias against women. But researchers noted that the gap tends to appear in patenting outcomes rather than uniformly across other measures of scientific work, which suggested the machinery of the patent office itself might be part of the story.
Tara Sowrirajan and Brian Uzzi of Northwestern University’s Kellogg School of Management, along with Ryan Whalen of the University of Hong Kong, set out to separate inventions by how much they departed from conventional thinking, then examine what happens to each type at the patent office.
Measuring how “unconventional” an invention is
Patents are tagged with technology classification codes that group inventions into topics like “Soil Working in Agriculture” or “Shaping or Joining of Plastics.” Most patent applications carry more than one code, since inventions often draw on multiple technical areas. The researchers used the frequency with which specific pairs of codes appear together across all previous patents to gauge how surprising a given combination was.
If a patent combined codes that rarely coexist, it received a high unconventionality score. If it combined pairs that commonly appear together, it scored as conventional. As a sanity check, unconventional patents in the dataset were 52% more likely to end up in the top one percent of most-cited patents, and their owners were more likely to keep paying maintenance fees, consistent with the idea that these are higher-impact inventions.
The team then examined 6.2 million USPTO patent applications filed between 2001 and 2018, along with smaller datasets of Canadian and UK patents used to check whether findings held up in other systems. For each application, they had information on the inventors, the invention, the examiner assigned to review it, the examiner’s cumulative experience, and whether the patent was granted. Inventors’ and examiners’ genders were estimated from first names using an algorithmic tool.
The gap appears only when inventions push boundaries
For conventional inventions, men and women had roughly equal chances of getting a patent granted. As inventions became more unconventional, the two lines diverged. At the far unconventional end, the gap between men’s and women’s grant rates widened to nearly 15 percentage points. The pattern held after controlling for team size, technology area, examiner experience, applicant type, and filing year, and it replicated in the UK and Canadian data.
The authors estimate that if women had matched men’s grant rate for unconventional inventions in the US, roughly 2,238 additional patents would have been granted, with an implied economic value of more than $234 million in 1992 dollars.
Why women examiners weren’t the safe harbor
A common assumption is that bias by male reviewers drives outcomes like this, and that pairing women applicants with women reviewers should help. The data pointed the other way. Women inventors had their highest grant rates when assigned to men examiners and their lowest when assigned to women examiners. Women examiners were about 8% less likely than men examiners to grant patents to women inventors.
The researchers interpret this not as evidence that women examiners are biased against women inventors, but as a signal that something else about the assignment process is driving results. When they broke examiners down by experience level, men and women examiners with similar amounts of on-the-job experience granted unconventional patents at nearly identical rates. The correlation between the two groups’ behavior across experience and unconventionality levels was 0.98.
Experience, dropout, and the assignment problem
Experience turned out to matter a great deal. In appeals of rejected patents, decisions by less experienced examiners were overturned more often than those of veterans, especially when the invention was unconventional. Across the board, more experienced examiners granted patents at higher rates and were particularly more open to unusual combinations of prior ideas.
Two institutional patterns then combined to disadvantage women inventors. First, women examiners on average had less experience than men examiners, a difference the authors link to higher dropout rates among women at the USPTO. Second, women inventors were assigned to women examiners about 17% more often than would be expected by chance. This is partly because examiners are routed to specialized “art units” that match the invention’s technical field, and the gender mix of examiners in a field correlates strongly with the gender mix of inventors filing there.
The upshot: women inventors’ applications were disproportionately reviewed by examiners with less of the experience that helps reviewers recognize the value of unconventional work. An all-women inventor team assigned to a low-experience woman examiner had a 37.1% chance of being granted an unconventional patent. An all-male team assigned to a high-experience male examiner had an 81.8% chance, a 44.7 percentage-point gap.
The researchers estimate that if half of unconventional applications from women-majority teams had been assigned to experienced women examiners, unconventional patent grants for those teams could rise by roughly 13%.
What the authors take from this
Sowrirajan and colleagues argue that their findings shift the emphasis away from stereotypes as the main driver of the patent gap for unconventional work and toward institutional practices that are, in principle, easier to change. Retention policies that help women examiners build longer careers, training that exposes junior reviewers to unconventional inventions, and adjustments to how applications are routed could all be tested.
They also raise a caution about a well-intentioned instinct: matching women inventors with women reviewers. If those reviewers are systematically less experienced because of retention problems elsewhere in the organization, gender matching can backfire for the applicants it is meant to help.
The study has limits. Gender was inferred algorithmically from names, which introduces some noise. The analysis focused on patent offices, and the authors note that the USPTO’s centralized structure may make it more amenable to procedural reform than more decentralized creative fields. And while unconventional patents are associated with higher citations and maintenance fees, patenting itself measures only one slice of innovation, not invention writ large.
