Over the past few months, there was a kerfuffle in the open source world over Google’s licensing of WebM, a video codec license — resulting in a revision by Google of the codec in early June.
What happened raises some very complex issues in open source licensing — surrounding implied patent licensing and license proliferation.
Google’s original release of WebM contained a patent termination clause that terminated both the patent license and the copyright license. Google then revised the license terms, and now everyone is happy. (Maybe.)
When discussing media formats, royalty-free refers to patent claims; in comparison, using formats like MP3 and MPEG-4 requires the licensing of patent rights from patent holders like Thomson and representatives, such as MPEG-LA, of the various patent holders.
When it comes to media codecs, many people are confused by the label “open source,” because the term is often used erroneously to mean an open standard. “Open standard” has no universally accepted definition, but is usually refers to standards with no known patent claims, or where all known claimants have agreed to license their patents royalty-free.
A particular media format may have many different implementations, and in many cases some implementations are open source and some are not. The implementation is code; the standard is a description of how the code should process media files in the format.
WebM, a royaty-free media file format, embodies the Vorbis audio format and the VP8 video format — the latter of which was developed by On2, a company that was officially acquired by Google in 2010. The software codec implementing WebM is an open source software project stewarded by Google, which is licensed under a open source BSD type license. Reporting of the re-licensing in the press was the usual tangled web of misinformation and inaccuracy that always seems to attend any patent subject, particularly in the open source world. But apparently Google, like many companies, was concerned about the effect on its patent position of releasing code under BSD alone.
The problem with open source licenses, from the patent point of view, is that none of them contain what lawyers call “reservations of rights” or disclaimers of implied licenses. So, even a license like BSD, which has no express patent grant, may be interpreted to grant some patent rights. See FSF’s slides about this issue here. This issue is not necessarily resolved by those licenses with express patent grants, like Apache 2, because they do not contain reservations of rights not expressly granted — a clause ubiquitous in proprietary licenses. So, some posit that additional implied rights may be granted, beyond the express grant. (GPL v3 actually states this expressly in Section 11.) Others would argue that it is highly unlikely that additional licenses would be implied when an express grant exists. The law on implied patent licensing is unclear, and when the law is unclear, companies react conservatively. So most companies releasing code under open source licenses are concerned that by doing so, they have unwittingly granted rights in patents beyond those covering inventions embodied in the code.
Contrary to popular paranoia in the open source community, most companies who release open source code do not worry about reserving rights in their patents because they want to sneakily sue the average hobbyist, or even open source users, for patent infringement. They are more concerned about getting sued themselves, and not having any patent weapons to protect themselves. Most companies use their patent portfolio substantially for defensive purposes– as a tool for mutually assured destruction to keep their competitors from suing them. Competitors acting in similar markets have a hard time avoiding each other’s patents. Thus, BadGuy sues GoodGuy for patent infringement, based on GoodGuy’s open source software release. (Similarities to the Jacobsen case are not coincidental.) GoodGuy tries to countersue for infringement of its own patents, but BadGuy claims that BadGuy has a license to GoodGuy’s patents, because they are implicitly licensed under the open source license. Because the extent of the implied license is unclear, this creates a monkey wrench that BadGuy can use to make the patent litigation more complicated and expensive — forcing GoodGuy to settle and pay for a license.
This issue is particularly thorny for permissive licenses like BSD, because their requirements are so limited. So it is easy for BadGuy to claim to be operating under the benefit of an implied license — BadGuy is not even required to abide by copyleft to claim this benefit.
While patent termination clauses exist in almost every open source license that contains an express patent grant, most only terminate patent licenses.
Placing this all in context, Google released WebM under a BSD type license, but included a patent termination provision similar to that in Mozilla. After much hue and cry, Google then released the codec under pure BSD, with a separate patent grant.
Chris DiBona’s blog about the re-release is here. He states, “Using patent language borrowed from both the Apache and GPLv3 patent clauses, in this new iteration of the patent clause we’ve decoupled patents from copyright, thus preserving the pure BSD nature of the copyright license. This means we are no longer creating a new open source copyright license, and the patent grant can exist on its own.”
Google’s licensing FAQ on WebM is here.
The blog says this means Google has not created a new license, but that is a philosophical question. Does an additional patent grant outside of the scope of a copyright license in fact create a new license? It is also unclear whether a new license is necessarily a bad thing. The corporate world of open source enthusiasts seems to be demanding a license with a reservation of rights to clarify the scope of its patent grant, but no such license exists. While there is much furor in the open source world about the problem of “license proliferation,” that problem means different things to different people. Is the sheer number of licenses a problem? The fact is that very few of them are actually used widely. Proprietary software, in comparison, has a nearly infinite supply. Or is it a problem that open source licenses tend, as a rule, to be opaquely drafted and difficult to interpret? Or is it a problem that every license is not GPL? Well, OK, only FSF thinks the latter.
Author’s note: Apologies for the lateness of this post. Now I know how to use the “Draft” feature.