At EFF, we often criticize software patents that claim small variations on known techniques. These include a patent on updating software over the Internet, a patent on out-of-office email, and a patent on storing data in a database. Now, Google is trying to patent the use of a known data compression algorithm - called asymmetric numeral systems (ANS) – for video compression. In one sense, this patent application is fairly typical. The system seems designed to encourage tech giants to flood the Patent Office with applications for every little thing they do. Google’s application stands out, however, because the real inventor of ANS did everything he could to dedicate his work to the public domain.
Jarek Duda developed ANS from 2006-2013. When he published his work, he wanted it to be available to the public free of restrictions. So he was disappointed to learn that Google was trying to patent the use of his algorithm. In his view, Google’s patent application merely applied ANS to a standard video compression pipeline. Earlier this summer, Timothy B. Lee of Ars Technica published a detailed article about the patent application and Duda’s attempt to stop it.
This week, the Patent Office issued a non-final rejection of all claims in Google’s application. The examiner rejected the claims on a number of grounds. First, he found the three broadest claims ineligible under Alice v CLS Bank, which holds that abstract ideas do not become eligible for a patent merely because they are implemented on a generic computer. The examiner rejected all of the claims for lack of clarity and for claiming functions that are not described with sufficient detail (applicants are often able to overcome these kinds of rejections with an amendment).
The examiner also rejected all of Google’s claims as obvious in light of Duda’s work, in combination with an article by Fabian Giesen and a 20 year-old patent on data management in a video decoder. Duda had made a third-party submission to ensure his work was before the examiner. Notably, this is a non-final rejection (and even final rejections at the Patent Office are not really final). This means Google can still amend its claims and/or argue that the examiner was wrong.
It is time for Google to abandon its attempt to patent the use of ANS for video compression. Even if it could overcome the examiner’s rejection, that would only reflect the failings of a patent system hands out patents for tiny variations on existing methods. It may be that Google is seeking the patent solely for defensive purposes. In other contexts, Google has worked to make video codecs royalty free. But that doesn’t make it okay for one of the world’s biggest companies to get a software patent on a minor tweak to someone else’s work. Perhaps it is unlikely that Google would assert an ANS patent in the short or medium term. But many once-dominant companies have turned to their patent portfolios as their star has faded.
ANS should not belong to tech giants willing to push applications through a compliant Patent Office. ANS should belong to all of us.