Google’s VP8 codec license is OK after all

Responding to my article here two weeks ago, the Software Freedom Law Center (SFLC), a law firm that provides pro bono support to the open source software community, has published an opinion on the patent cross-license Google is considering for the VP8 codec. This is the codec used in WebM and proposed as the “mandatory to implement” video format for WebRTC, the real-time communications component of HTML5.

While the opinion broadly agrees with my assertions — notably, if the terms in the cross-license were part of a software license rather than offered separately, they would render the license incompatible with the Open Source Definition (OSD) — SFLC usefully clarifies that there’s no actual incompatibility with open source licensing:

Because the patent license does not restrict those freedoms, but rather affords some new, limited protections to users and developers within the field of use, it improves on the current situation. Without this license, the patent holders would be in a position to threaten those users and developers as well as others.

[ Simon Phipps tells it like it is: Why software patents are evil. | Stay ahead of the key tech business news with InfoWorld’s Today’s Headlines: First Look newsletter. | Read Bill Snyder’s Tech’s Bottom Line blog for what the key business trends mean to you. ]

Re-reading my article, I realize it could be interpreted as stating there was an incompatibility with open source licensing. That’s not my view. The problem is with introducing patents to the discussion at all. I believe Google should put much clearer caveats around the cross-license for open source developers, along the lines of “we don’t think you actually need this, it’s just to stop OEMs and pro-patent low life saying there’s a problem.”

In a long conversation, SFLC’s Aaron Williamson and I found ourselves in broad agreement. SFLC agrees that the patent license has undesirable terms, notably around the need to register individually (non-sublicensability). I agree that the OSD does not apply to this document. I have been educated, though; I had expected to find that GPLv3 prevented someone writing code while relying on a patent license without passing on the protection of that license.

That’s not so. While GPLv3 section 11 has language requiring developers to convey to others the same patent protections they themselves rely on, I’d not fully absorbed that this requirement is only imposed if the source code implementing the patented idea is kept secret. As long as you publish your implementation of VP8, it’s perfectly OK to license under GPLv3 and to rely on patent protection you’re not passing on.

When it comes to sublicensability, I remain unsettled by having a license that one has to trade personal information to receive and that can’t be extended to others. It’s been suggested to me that Google would have preferred a sublicensable cross-license but was prevented from offering it by restrictive terms from MPEG-LA, but I can’t confirm that. However, a clause of Googles cross-license I’d not previously fully absorbed lends assistance here. Clause 4 offers “release from past infringement.” Translation: If you accept the license, you’re released from any claims of infringement of MPEG-LA patent claims. While no one I have spoken to believes MPEG-LA actually has any patents that read on VP8, this is a powerful device.

It means that in the event you’re attacked by an MPEG-LA patent holder for infringement in your implementation of VP8, you can simply accept this cross-license and instantly be absolved of any claims. Moreover, the license is unilateral. To gain a license, you simply have to express agreement to its terms and conditions by mailing a signed copy to Google; there’s no indication they have to approve your acceptance. Thus, while I can’t extend the protection I have to you, you can at any time retrospectively get the same protection yourself.

But what if Google were to try to change the rules? SFLC says:

There is some valid concern that the VP8 licensors could subsequently change or withdraw the offer of retroactive release. The license could be improved by a promise not to do so. But the law doesn’t reward manipulation; if the retroactive release appears in the final, published license, the VP8 licensors may well be estopped from subsequently seeking to enjoin developers or distributors who have relied on it in good faith.

In other words, having led us all to rely on this cross-license as a safe haven for VP8 developers, both Google and the MPEG-LA licensors on whose behalf they are acting and accumulating reciprocal licenses would have a hard time convincing a judge that someone was at fault for relying on the availability of a retroactive license — a doctrine called “equitable estoppel.”

Other experts I’ve consulted add one more point: Companies implementing hardware and embedding video software are very cautious. They aim to avoid all risk their devices may be stopped at customs as a result of the sort of injunctions multiple actors are seeking in the mobile patent wars today. These OEMs like to have a positive indemnity document rather than just a logical assurance there’s no problem.

Several people I’ve spoken with have told me that Google does not think anyone actually needs a license from MPEG-LA for VP8, but fear, uncertainty, and doubt on the subject have slowed progress so much that they had to get positive reassurance for their OEMs. That’s the reason this cross-license exists — not because there’s any known risk from MPEG-LA, but because certain nervous companies demanded assurance in writing.

At this point I’m satisfied. I think Google has done all it can to deal with a difficult problem — the uneasiness of OEMs in the face of a powerful anticompetitive force in the market — while keeping open source implementation possible. Though Google’s problems with Nokia and VP8 are still live, I believe this lays to rest the FUD created by MPEG-LA when it first announced a VP8 patent pool back in 2011.

Several commentators have also incorrectly associated my article with OSI. While I am indeed the president of the Open Source Initiative, I have several other affiliations. Nothing I write here or elsewhere is on behalf of OSI or any other party unless I explicitly say it is.

WIth this publication by SFLC of their opinion about Google’s VP8 patent cross-license, it’s worth revisiting the real problem: software patents. This whole mess only arises because of the invasive application of the business models of nonsoftware industries to the technology of the Internet. Not all business models are welcome or valid; the Web has become what it is today without tolerating “standards” that rely on patents. There’s no reason we should start tolerating them now.

SFLC’s opinion is not a validation of patent licensing like the parasitic Microsoft business around Android. Rather, it is a pragmatic response to OEM concerns, providing a solution most of us shouldn’t need to care about to an attempt by incumbents to veto innovation. As TechDirt says, allowing incumbents to veto innovation is bad for society; perhaps it’s even an antitrust problem. Until we see meaningful patent reform, this sort of situation will keep arising to slow down progress.

This article, “Google’s VP8 codec license is OK after all,” was originally published at InfoWorld.com. Read more of the Open Sources blog and follow the latest developments in open source at InfoWorld.com. For the latest business technology news, followInfoWorld.com on Twitter.