Professor analyzing decades of data to determine patent value

LAWRENCE — For more than two centuries, patents have been considered a key governmental policy tool for economic innovation. And for just as long numerous assumptions have been made about what they mean to an innovation’s value, where the most important ones are litigated and numerous other questions. A University of Kansas law professor is part of a project that is providing definitive answers to these and other patent questions for policy makers through a unique, big-data approach.

Andrew Torrance, the Earl B. Schurtz Research Professor at the University of Kansas School of Law, and colleagues have developed an approach to analyze mountains of detailed U.S. patent data from 1976 to the present day. One application of their research, commissioned by Canada's Ministry of Innovation, has been a comprehensive analysis of how patents having either Canadian inventors or owners compare with those without such connections. One of their most striking findings is that patents listing at least one Canadian inventor are more than 15 percent more valuable, on average, than other patents.

In a separate study, they have shown that litigated patents tend to be much more valuable than those that avoid court, and that federal courts in the southern midsection of the U.S. play host to litigations involving the most consistently valuable patents. Current studies involve comparisons of patent values of so-called “patent trolls” and companies whose goods or services are covered by their patents, an exploration of which parts of the U.S. give rise to inventors of more valuable patents, and which areas of technology give rise to the most valuable patents.

The U.S. Patent Office recently made decades of patent data available online. Torrance and colleagues Jevin West and Carl Bergstrom of the University of Washington used this data to build a huge database which they can use to analyze the data from a myriad of perspectives. Through this approach, they hope to test many questions arising from the perceived wisdom about patents.

“We’ve put that data together in a giant database and added other data to it as well that includes information on every U.S. patent from 1976 until last Tuesday (the day new patent data is released by the United States Patent & Trademark Office each week),” Torrance said. “We have transformed it into an easy-to-use form that allows us to run many different types of analyses.”

The Canadian Ministry of Innovation approached Torrance to learn more about the value Canadian inventors add to American patents. Their goal was to learn more about how Canadian inventors and companies perform in the U.S. patent system. The data provided a number of fascinating insights possible only through a big data approach, including one that should make Canada quite happy.

“We found that, when you add a Canadian to a U.S. patent as an inventor, that patent tends to increase in value by more than 15 percent,” Torrance said. “When you add a generic, non-American from another country, the average patent value actually tends to go down. This raises intriguing questions about how Canada fosters more successful inventors.”

What’s not clear is why Canadian inventors tend to increase a patent’s value. It could be due to the particular technology fields in which Canadians tend to invent, characteristics of science and technology education in Canada, or Canadian skill at collaborating with other talented inventors, Torrance said. But he and colleagues are beginning to analyze the data to calculate the average values of patents generated by inventors from every other country to compare them all.

The findings are unique because the data they are drawn from was largely unavailable for decades, which forced people to make assumptions about the patent system and value of patents it issued. Additionally, because the data accessed is comprehensive, the analyses can provide objective answers based on all the data rather than just small random samples.

Torrance compared it to polling: Political polls ask a sample of people questions such as which candidate they plan to vote for, then report who has a lead, based on the representative sample of people they polled. That method, widely used in research for many years, can provide a good idea of the answer to a question, but it comes with built-in error margins. The method Torrance and his colleagues are using, however, gives definitive answers because it relies on all the data. It is akin to being able to access every voter and get a definitive answer on whom they voted for.

“Having these gigantic data sets finally allows us to answer questions about which, until now, people could only speculate – and often speculate wildly,” Torrance said. “Now we can formulate a question about patent law, such as, ‘How valuable do design patents tend to be compared to utility patents,’ write a software script to analyze our huge data set and then see what answer the data give. That simply was not possible before the era of big data.”

Torrance and colleagues have already submitted their preliminary analyses of Canadian inventors and patent owners to the government of Canada, which then hopes to use the resulting insights in future policy decisions regarding the Canadian patent system and how it influences innovation. Torrance and colleagues plan to publish these findings and plan to carry out many more analyses using their data.

“Our big patent data research should keep us busy for a while,” Torrance said. “There are myriad basic questions we can now answer.”

Two projects they’ve already begun are looking at the value of patents that are litigated and where litigation of the most valuable patents takes place. In the former case, there has long been a school of thought that holds patents that are litigated in court are not inherently more valuable than those left unlitigated but are acquired by companies with the resources to hire teams of attorneys to assert those patents against others in legal proceedings.

In the latter, it has long been assumed that most patents are litigated on the coasts, and that the middle of the country is a “patent flyover country” of sorts. Contrary to this assumption, the big patent data analysis has showed that both assumptions, though long-held, are extremely inaccurate. For example, the highest concentration of valuable-patent litigation occurs in the southern middle of the country, with the coasts and the north lagging behind. Publications are forthcoming on both topics.

Torrance was also recently named a senior fellow with the Center for International Governance Innovation, or CIGI, International Law Research Program. The international, nonpartisan think tank focuses on improving international governance through research on the global economy, global security and politics, and international law. The organization brings scholars from around the world together to provide governments information on innovation and how it can address problems such as human rights, avoiding war, fighting terrorism and poverty, improving development and the standard of living for people worldwide.

Torrance hopes his ongoing research, both into patent systems and user, open, collaborative and free innovation, through CIGI, will be valuable in contributing to CIGI’s goals and to questioning assumptions that may not survive rigorous scrutiny.

“It’s great to be able to ask basic questions, then look at the data and see what they say, compared to what the assumptions are,” Torrance said. “We’re already in the age of data and are increasingly able to answer questions that were infeasible to tackle before. My background is in science, and it’s gratifying to be able to apply the scientific method to legal questions, especially when the answers upend long-held, but unjustified, assumptions. This is a great way to improve the law.”