There Ain’t No Such Thing as a Free Video Codec

Google, Inc recently achieved something that perhaps no other company or organisation on the planet could have done – it secured an agreement from most of the major corporate owners of patents relating to digital video coding standards, which will enable developers and users to access digital video on the web without payment of patent royalties.

At least, this is what will happen if the World Wide Web Consortium (W3C) takes up the opportunity that Google has created.  However, Google’s recent publication of the proposed terms of a royalty-free cross-licence for implementers of it’s digital video technology has drawn predictable criticism from a number of sources, including Open Source Initiative President Simon Phipps (‘proposal closes door on software freedom’) and FOSS Patents blogger Florian Mueller (effectively accusing Google of hypocrisy and disingenuousness).

It is easy to be cynical about the motivations of a company as large and ubiquitous as Google.  But it is worth having a closer look at exactly what Google has achieved here, regardless of its level of self-interest.  In my opinion, the deal that Google has struck is a significant step forward for royalty-free and open standards on the internet.  And while it is never going to satisfy those individuals and organisations at the more radical end of the free software and/or anti-patent movements, as a practical matter the deal will give the vast majority of internet users an assurance that there will be an open source and royalty-free video standard available for use on the World Wide Web.

Google’s Push for a ‘Free’ Video Standard

Back in March, Google reached an agreement with the video coding patent pool operator MPEG LA in relation to Google’s VP8 video codec – technology which enables the coding/decoding and compression/decompression of digital video content.

The VP8 codec, which forms part of the WebM media file format which Google is putting forward as a potential web standard, is the result of work initiated by a company called On2.  The objective of this work was to develop a codec which would be free of any liability to pay royalties to the owners of patents relating to existing industry-standard video coding formats.  Back in 2010, Google acquired On2, and made the VP8 codec technology available under an open source licence.

Almost immediately, MPEG LA put out a call for companies with patents they considered to be essential to implementation of the VP8 codec to come forward, with a view to forming a new patent pool.  Apparently, at least 12 prospective licensors did so. 

Google has always maintained that the VP8 codec is unencumbered by IP rights belonging to any other party.  I do not believe this, for at least three reasons.

First, since the VP8 implementation is open source, there is nothing to prevent patent holders from evaluating the codec to assess whether or not it implements any patented methods.  While it would not be surprising for one or two patent trolls to come forward asserting dubious claims, the likelihood that around a dozen substantial operating companies would do so without good cause seems slim.

Second, there have been so many researchers, in both public and private organisations, working on video coding technology for decades now, that the idea that any one company could develop a truly independent technology of comparable quality simply beggars belief.  Even the Chinese (not always known for their respect of foreign IP rights) were forced to concede that they could not develop a video coding standard of their own without employing at least some existing and patented techniques.  In the end, they took the pragmatic approach of incorporating the cost of licensing, in addition to technical specifications such as quality and bit-rate, into their criteria for an acceptable standard.

Third, although Google’s deal with MPEG LA and a group of 11 patent owners does not involve any public admission on either side that any patents either are, or are not, essential for implementation of the VP8 codec, it is difficult to believe that Google would have done any deal if it felt that it had a very strong case.  The cost of litigation to settle the question would not be any barrier to Google (although the uncertainty that this might cause for the duration could be a concern).  The licensors – all reputable operating companies – would have no desire to be cast in the role of ‘patent trolls’, effectively extorting money from VP8 implementers on the basis of bogus patent claims.

There is therefore good reason to believe that VP8 is not ‘royalty-free’.  Indeed, Google may well be paying, on behalf of the entire global internet community, for access to the relevant patents.  (Of course, we will be paying them back, with interest, through every transaction we conduct via Google services, but that is a separate issue.)  And Google may have further work to do here, because it seems that VP8 is not out of the woods yet – Nokia, which is not a member of the MPEG LA group, has since come out claiming patent rights of its own.

The ‘Problem’ With Google’s Proposed Cross-Licence

So Google has negotiated the deal, and is footing whatever bill there may be for access to the MPEG LA group’s patents.  So what is the problem?

Reaching a workable agreement with a patent owner is not the same thing as making the patent ‘go away’.  The patent holders do not want their patent rights to become available, without payment of royalties, for just any purpose.  The agreement is that Google can sublicense the right to implement the VP8 codec (and its successor, VP9).  In doing so, Google is obliged to impose certain conditions on sublicensees.

Most significantly, the licence is limited to implementation of the VP8/VP9 codecs.  A sublicensee has no right to use the patented technology for any other purpose (including non-standard variations of VP8).  In order to ensure that every implementer is bound by this obligation, all must be licensed directly with Google – no further sublicensing is permitted. 

These arrangements are incompatible with most open source principles, which generally eschew any restrictions on use, and are predicated on automatic sublicencing to enable redistribution of code, in source or executable form, without the need to seek further licence grants.  And that is why dedicated open source and free software proponents are never going to be happy with this kind of arrangement.

On the other hand, the proposed cross-licence has a grant-back clause: any party which takes a royalty free-licence through Google agrees that it will contribute back any patents it may own or obtain which are essential to implementation of the VP8/VP9 codecs.

Market-Based Development of ‘Open’ Standards

The digital video technologies we enjoy today, including broadcast TV, DVD Video discs, Blu-ray discs, dowloadable content and online streaming video, are the result of over 30 years of research and development, most of which has been conducted by private sector companies, at the considerable collective expense of their shareholders.  Owners of IP relating to the H.264 (or MPEG-4 AVC) video coding standard, which is used for most high definition video applications, include Philips, Sony, Ericsson, Toshiba, Apple, Cisco, Dolby Labs, Hewlett-Packard, Hitachi, LG Electronics, Microsoft, Panasonic, Samsung, Sharp, Fujitsu and ZTE.

Of course, these companies engaged in the standardisation activities for sound commercial reasons – they wanted to ensure that their own technologies would have an opportunity to be adopted and that they could get in on the ground floor with the development of products compatible with the standards.  One further financial benefit gained by companies which elect to contribute proprietary-developed technologies to a standard, is the right to collect royalties from future implementers of the standard.

Viewed from a free-market perspective, all this is a good thing.  The market – including the market for IP rights, as well as compatible products – provides the incentive for many companies, which would otherwise be competitors, to get together and develop a common standard.  During the standardisation process, they compete to put forward the best solutions to technical problems, in the hope that they will be adopted into the standard.  The best solutions (in terms of practical commercial implementation as well as technical merit) emerge from the process.

Modern standard-setting activities are therefore, first and foremost, market-driven.  The reason this is possible is because the IP developed by participants can be owned, which means that it can be bought, sold and licensed for commercial gain.  Furthermore, in the early stages of adoption of a new technology standard, the fair cost of accessing the IP is also determined by the market.  If the price is set too low, the companies contributing to the standard do not receive adequate compensation, and may therefore decline to participate.  On the other hand, if the price is set too high, then the standard may not be widely-adopted, which is to the detriment of all of those who contributed to its development.

(I should note that this balance of market power shifts once a technology standard has become widely adopted, but that is a topic for another time.)

By and large this process has worked well.  The global proliferation of interoperable devices and software is proof-positive of this.  It need not have been this way.  History is littered with examples of markets fragmented by a lack of agreement over standards: VHS vs Beta VCRs; MiniDisc vs DAT vs CD; HD-DVD vs Blu-ray; PAL vs SECAM vs NTSC analog TV broadcasts; and different power supply standards (AC vs DC, 110V vs 204V, 50Hz vs 60Hz) to name just a few.  Many of these resolved themselves through one technology eventually dominating, while others have been solved by the development of various types of adapters, or multi-system compatible products.  Consumers pay for these solutions, either because they backed the wrong horse, or because they end up paying for products which support multiple systems which many do not need.

‘Are You a Socialist?’

So if you want global standardisation and interoperability, but you do not want private corporations to ‘own’ any of the adopted technology, you need to put forward an alternative mechanism to the free market approach which currently operates.  In short, if companies do not pay for the research and development, and the costs of participation in the standards-setting processes, who does?  Governments (i.e. taxpayers)?  Philanthropists?  Perhaps all the labour could be crowdsourced through the open source movement?

Any of these approaches might work on some occasions, for some technologies.  Development of the internet protocols was essentially paid for by US taxpayers through funding to the US Defense Advanced Research Projects Agency (DARPA).  Richard Stallman’s ability to devote himself to the GNU project was given a significant boost in 1990 by the award of a $240,000 grant by the John D. and Catherine T. MacArthur Foundation.  Tim Berners-Lee developed the Word Wide Web while working at CERN (the European Organization for Nuclear Research), which is funded by the governments of its member countries.  And, of course, open source developers have built a pretty solid computer operating system in Linux (although open source advocates rarely acknowledge the achievements of proprietary models, such as the release of Microsoft’s first GUI-based, fully pre-emptive, processor-independent, multiprocessing, multiuser operating system – Windows NT – way back in 1993).

But ultimately the solutions have to be not only technically sound, they also need to be commercially sound.  Something as complex as a digital video standard is not just about whether a smart kid with a PC can implement a software codec.  It is also about whether it is commercially feasible to build the kind of hardware required to deliver high-definition live global broadcasts of huge events such as the Olympics.  The reality is that there are some things that the open source community can do really well, and other things for which it is simply not equipped.

If you believe that Google is, indeed, pressing developers into a deal with the devil, then I think there is a question you need to seriously ask yourself: ‘Am I a socialist?’  Do you really believe that the current market-driven system of technology standardisation would be better replaced by some form of centralised bureaucracy, doling out funds, resources and whatever other forms of control and management may be required to get the job done within an acceptable time frame?

Alternatively, if you think that the free-market generally does a reasonable job of delivering technology advances that benefit society, then you might also need to consider one of the fundamental tenets of the capitalist system – that if you want the benefit of something that somebody else has paid to develop, then one way or another, you are going to have to pay for it. 

Conclusion

The fact is that when, for example, you buy a smartphone – which is, if you stop to think about it, an absolutely astonishing piece of technology for an extremely reasonable price, thanks to the competitive marketplace – you are potentially contributing something to every company that has invested in that technology.

Video codecs (which are, of course, themselves incorporated into smartphones) are just one of the many technologies which have come about as a result of significant investments by private corporations.

So, whether you are paying cash, or agreeing to some reciprocal obligations as a condition of your use, the fact is that there ain’t no such thing as a free video codec.

The bottom line is that somebody has got to pay.  It is about time that advocates of ‘free’ and ‘open’ standards for the internet acknowledged this.  There may well be viable alternative funding models, but they are not going to come about through advocating the abolition of relevant IP rights. 

For all its faults, at least Google is pragmatic enough to move this process forward.  If left up to the ideologues within the free and open source software movement, this would never happen.

Tags: FRAND, Google, Open source, Technology standards