ITHIS IS THE WILL to human inventiveness that 50 million patents have been granted worldwide. But overall, much of the collection looks like an intellectual dumping ground. Included are plausible ideas that no business has ever wanted to pay for, plausible ideas that failed, and nonsense. A patent on the crustless peanut butter and jelly sandwich, for example, was not renewed in 2007.
Narrow the list down to those that are both sane and legally in effect, which means a fee is paid to a patent office to keep them alive, and there are 16 million patents that matter. Last year 1.6m were awarded.
Most are owned by businesses, but balance sheets and conventional accounting are poorly suited to capturing their value. Using the acquisition cost and then depreciating it doesn’t work. Instead, lawyers provide subjective numbers based on factors such as a patent’s likely validity, royalties, and litigation history. Many companies find that it’s not worth paying the tens of thousands of dollars it costs for an assessment.
In 2008, an intellectual property exchange opened in Chicago to do for patents what other exchanges have done for stocks, bonds and commodities. Its backers were blue chip companies like Hewlett Packard and Sony, but it closed in 2015. Patents cannot be treated as commodities, the Cornell Law Review. A subsequent effort to evaluate them used software to read and rate the documents. Until now, however, even machine learning techniques have failed to allow code to penetrate the opaque legal language in which patents are written.
Today, a startup called PatentVector, founded by a law professor, an information science professor, and a software engineer, is trying something new. It uses a variation of a method started in the 1960s that evolved to calculate how often individual patents are cited (a similar citation-based process is used to assess academic research).
Rather than trying to understand the patent, PatentVector uses artificial intelligence to comb through 132 million patent documents held by the European Patent Office in Munich (the largest collection in the world). Then, it assesses, on the one hand, the frequency with which a patent is cited and, on the other hand, the frequency with which it is cited by patents themselves frequently cited. This gives an indication of importance which is then multiplied by an average patent value based on an estimate by James Bessen, an economist at Boston University, which has become a benchmark. A number of companies, law firms and institutions (including the Canadian Patent Office) purchase the product from PatentVector.
The results contain valuable information about the invention. Frédéric shelton IV (pictured) isn’t among the prominent innovators of the 20th century, but he probably should. He works at Ethicon, a medical device subsidiary of Johnson & Johnson, and PatentVector values his inventions at $ 14 billion, which puts him forever in front of anyone. Its first three are for a mechanical surgical instrument, surgical staples and the cartridge for staples; in short, tools for cutting fabrics and binding them.
Ethicon itself, a manufacturer of medical devices, holds 95 of the world’s 200 most valuable patents, according to PatentVector. The company also employs Jerome Morgan, who is ranked second with $ 5 billion in patents (many of which overlap with those of Mr. Shelton). Only one other person is part of the $ 5 billion club: Shunpei Yamazaki, president of Semiconductor Energy Laboratory, a Japanese research and development company. Mr. Yamazaki’s most important patent covers screens for computers, cameras and other semiconductor devices.
PatentVector found 65 other people each responsible for patents valued at over $ 1 billion. Only 14 of the top 650 DIY enthusiasts are women. Highest ranked is Marta Karczewicz, who works for Qualcomm, an American chip designer, and was instrumental in inventing the video compression technology that powers Zoom and other video services.
Almost all valuable patents can be found in a few large industrial groups: biopharmaceuticals, software, computer hardware, medical devices and mechanical equipment. Over the past 40 years, the importance of specific categories has increased and decreased slightly, but biopharma and information technology (THIS) dominated and their importance increased. The companies with the highest overall patent value are in THIS, surmounted by IBM, Samsung and Microsoft.
PatentVector’s figures on the patents held by countries are also revealing. America has the most active patents of any country at 3.3 million, followed closely by China with 3.1 million. But there is a world of difference in how often they are cited and their imputed value. The American library is estimated at $ 2.9 billion, compared to $ 392 billion for the Chinese collection.
Of course, the methodology of PatentVector will be subject to careful scrutiny. Naturally, the startup patented its own technique. Information about patents, which are essential parts of invention, has never been so important. Perhaps it was inevitable that innovation would apply not only through patents, but to them as well. ■
For a more in-depth analysis of the biggest stories in economics, business and markets, sign up for Money Talks, our weekly newsletter.
This article appeared in the Business section of the print edition under the headline “Billion Dollar Plans”