For the most part, Google has actually been one of the good guys on patent issues. Unlike some other Silicon Valley companies, Google has long resisted using its patents to go after others, instead only using the patents defensively. It has also fought for patent reform and experimented with new models to keep its own patents out of the hands of patent trolls. But it’s been involved in an ongoing fight to patent something that an earlier inventor deliberately released into the public domain, and it reflects incredibly poorly on Google to keep fighting for this.
A Polish professor, Jarek Duda, came up with a new compression technique known as asymmetric numeral systems (ANS) years back, and decided to release it to the public domain, rather than lock it up. ANS has turned out to be rather important, and lots of companies have made use of it. Last summer, Duda noticed that Google appeared to be trying to patent the idea both in the US and around the globe.
Tragically, this happened just weeks after Duda had called out other attempts to patent parts of ANS, and specifically said he hoped that companies “like Google” would stand up and fight against such attempts. Three weeks later he became aware of Google’s initial patent attempt and noted “now I understand why there was no feedback” on his request to have companies like Google fight back against attempts to patent ANS. In that same thread, he details how there is nothing new in that patent, and calls it “completely ridiculous.” Despite noting that he can’t afford to hire a patent lawyer, he’s been trying to get patent offices to reject this patent, wasting a bunch of time and effort.
While a preliminary ruling in Europe appeared to side with Duda, accepting his evidence of prior art, Google is still fighting against that ruling and is continuing its efforts to patent the same thing in the US. This is getting new attention now after Tim Lee at Ars Technica wrote about the story, but it’s been covered elsewhere in the past, including getting lots of attention on Reddit a year ago and Hacker News soon after that.
Google’s response to Lee at Ars Technica are simply ridiculous. First, it claimed that Duda’s invention was merely “a theoretical concept” while it is trying to patent “a specific application of that theory that reflects additional work by Google’s engineers.” But if you read through the analysis by many people who understand the space, that doesn’t appear to be the case. There’s very little that appears “new” in the Google patent, or non-obvious based on what Duda and others had already disclosed.
Google’s second response is even more nonsensical:
“Google has a long-term and continuing commitment to royalty-free, open source codecs (e.g., VP8, VP9, and AV1) all of which are licensed on permissive royalty-free terms, and this patent would be similarly licensed.”
While that’s true, that’s no excuse for locking up what’s in the public domain and promising to treat it nicely.
The thing is there is simply no reason for Google to continue down this path. Again, the company has almost never been an aggressor on patents preferring to use them defensively. And it can still do that here — by just pointing to the public domain to invalidate anyone else’s attempt to patent this. The fact that Google is being slammed in various forums over this (and has been since a year ago) should have clued the company in to the fact that (1) this isn’t necessary and (2) harming its own reputation with engineers just to secure a patent it doesn’t need is not a good idea.
Google has tons of patents. It doesn’t need this one. If it really thinks that it’s own invention here goes beyond what Duda did — and Ars Technica notes that Google ignored multiple requests to explain what is different in its patent application — then the company needs to be much more transparent and upfront about what is different from Duda’s work and the company can just as easily release the same information to the public domain as well. Yes, that would be giving up on one patent, but Google can survive donating a patentable idea to the public domain if it actually has one.