Search
  • Robert Cantrell

How to Lie with Patent Statistics

Updated: Aug 23

June 20, 2022

By Robert Cantrell - Registered Patent Agent


Before becoming a patent practitioner, and before consulting on invention strategy, I ran patent analysis teams. My teams produced patent analysis reports loaded with statistics. As a result, I learned a lot about the good, the bad, and the ugly that can be a patent analysis report. And though this post is about the patent statistics within such reports, it was inspired by a common technique for misleading an audience I observed on television screens while I warmed up on a treadmill at a gym.


As I warmed up, a news anchor read a report about concerns on peoples’ minds. But there was a problem with the statistics. News writers had split concerns about inflation into three categories: the economy, rising prices, and the price of gas. Add the three together, and inflation eclipsed the rest, totaling about fifty percent. However, splitting inflation into three categories allowed concerns the news writers wanted to raise to appear comparably important on the list and made it easier to ignore the inflation.


Lying with patent statistics is much easier than lying to a news audience, so much so that patent analysts may do so without realizing their errors. Common is to compare the number of filings in one country with the number of filings in another country as if the data were equivalent—the classic apples versus oranges error—without appreciating the differences in patent systems. Also common is to unintentionally present unequal categories, much as the above news writers did, especially where classification subcategories blossom to reflect higher-than-average numbers of patents in those categories. Analysts may artificially inflate smaller categories that have fewer subdivisions. Then there are the algorithms that attempt to assign values to patents where people may use those values without understanding the algorithms—especially when the algorithms are proprietary. Often, these algorithms anchor on patent citation analyses, which can be a valuable tool, but can also mislead, especially if used without reading the patents.


Consequently, and as with the news, it is easy to get patent statistics to say what you or your customers want them to say. Two plus two can equal five, six, whatever you want. However, if you need two plus two to equal four, then be wary about your patent statistics. Let the statistics guide your questions more than provide your answers. Ask why one of your patents was cited more than others. Ask why there’s an uptick in patenting in a category. And always question the numbers. Question why you received the views presented and why people drew associated conclusions.


To illustrate, I consulted about a major corporate acquisition planned between European and Asian consumer electronics enterprises about twenty-five years ago. The acquisition would have been a multi-billion-dollar acquisition and front-page business news. It was driven by R&D. A detail hidden in the statistics—misleading if it was intentional—was the average age of the patents within the most sought technology categories of the targeted company. That average age, a number hidden behind the numbers displayed on the charts, pointed to other viability problems in the sought R&D teams. The acquisition never happened.

 

If you would like to know more about using patent statistics, please contact us on our CONTACT page.


4 views0 comments

Recent Posts

See All