The problem with standards, patents, and defining 'open'
It is about more than purists versus pragmatists
Published 14:25, 25 April 12
Currently a debate is raging in the UK to define an “open standard” for government procurement purposes. It is a debate which bleeds across European procurement discussions and will likely effect North American debates. At its heart is whether patents that read on a formally defined standard and are essential to implementation are licensed in a “fair, reasonable and non-discriminatory” manner or must be royalty-free.
The concern raised is that if we do not define “open standard” to mean free from any financial encumbrance (typically meant to mean patents), then the government would not be in a position to acquire an implementation of a standard that was licensed as free or open source software.
Large vendors with large patent portfolios are painted as terrified that such royalty-free procurement policy would crack open their patent portfolios and they would lose control of their R&D investments and the attendant expensive intellectual protections. In a world where vendors are entangled in expensive patent litigations (e.g. in the mobile space) and spending billions on acquiring patent portfolios in bidding wars (e.g. Microsoft and AOL), it’s a complex issue.
Debates involve government-enacted patent legislation, economics and innovation arguments wrapped up in national flags (during tough economic times), and corporations competing with one another using every tool at their disposal. No amount of yelling or logic will reassure ultra-conservative lawyers and lobbyists from vendors with very deep pockets that royalty-free licensing on a definition of open standard for government procurement isn’t the beginning of the end of the world as they know it.
But there’s a deeper problem. It’s not just the R&D investment — vendors are worried they’ll lose their product implementation investments as well. If vendors spend enormous time implementing “open” royalty-free standards and possibly certifying them as governments often desire, and an outside actor can invalidate the “open” definition of the standard by presenting a patent, then things would be very grim indeed for all participants.
There’s a fundamental problem. The problem relates to the differences between patents and standards. Patents exist in a legislative system to protect a single invention. Standards on the other hand exist to encourage many implementations. They are polar opposites in the world of economics and there is tension between them because of their very definitions.
One can even argue that historically patents and standards serve different parts of the technology adoption life-cycle. Patents are government-enacted tools designed to encourage lots of new innovations by benefitting the inventors. Standards come later in the technology implementation life-cycle to encourage lots of implementations of proven solutions and serve the customer.
Successful standards tend to be based on existing practice and experience. When the IETF went to work on the original Internet standards, there was lots of experience with computer networking that encompassed IBM SNA, DECNet , Banyan, and a host of other technology implementations.
When folks start developing a standard, they often start with a model technology that has a well understood licensing environment. For example, the UNIX system was the model for the POSIX standards. It wasn’t royalty-free, but every vendor had an implementation and everyone understood their licensing obligations with respect to the copyrights and trademarks.
As standards often occur later in the technology implementation life-cycle, there was historically a generally well-understood patent landscape as well. Standards development organisations such as OASIS and IETF have always needed intellectual property rights (IPR) policies to have a way to discuss what happens when one discovers patents in the midst of a specifications debate.
They are designed to give the participating members doing the work as much warning as possible about potential patent problems. Indeed, the OASIS IPR policy is very well designed to ensure participating members are well behaved. While any individual vendor may love the idea of developing a formal industry standard that embeds their patents, thereby creating a tax on the specification, in general standards development organisations don’t want to create standards that are instantly taxable by one of their members.
This is key. The participating members of a standards development organisation are acting on their own individual behalf to ensure everyone is well behaved. These organisations are investing their time and energy (and money) in defining the standard and then implementing it. They want to ensure what they create is as level a playing field as possible, and they understand they can only police themselves through their membership and IPR policy. They can’t police outside actors from upsetting the process.
This is not the same as the Open Source Initiative creating a definition of attributes that apply to software licenses. That was a relatively new discussion around a new idea that could become self-policing. The current debates around “open standards” involve enormously entrenched ideas about long held beliefs in standardisation and intellectual property protections. It will be impossible to come up with a simple well-meaning definition that can be defended.
So that’s the challenge: Define an open standard in such a way that all the hard work and investment developing the standard, implementing it (and possibly implementing multiple standards in a product), certifying the products against the standards if necessary and other related policy approval processes, and customers procuring and installing it, cannot be destroyed by an actor outside that group presenting a patent that reads on the standard previously defined as “open”.
Glyn Moody pointed out through an excellent investigation that much of the current vendor lobbying is actually unnecessary as there are two clear statements in the Open Standards document that allow encumbered standards to be used. Unfortunately, that doesn’t remove concerns about creating a definition of “open standard” that can be trivially destroyed by outside players.
I have developed and balloted standards, put my money where my mouth was and created a company to implement the standards in products, and survived certification arguments for government procurement. This was before patents had become the big issue in software they’ve become over this past 15 years. As well as freaking out vendors, the current definition of “open” meaning free from financial encumbrance doesn’t serve the end consuming customer. I think the best thing the UK government could do today would be to spell out their preference for unencumbered, then royalty-free, then FRAND standards in the document long-hand, rather than coming up with a concise but vulnerable definition of “open standard”.
Posted by Stephen R. Walli
Stephen is technical director at the Outercurve Foundation. He has worked in the IT industry since 1980 as both customer and vendor. He was most recently a consultant on software business development and open source strategy. His customers included Microsoft, the Eclipse Foundation, the Linux Foundation. He's an adviser to Ohloh (acquired by SourceForge), Bitrock, Continuent, and eBox.