Draft 5; 2006-09-19
Please send comments and suggestions via the OSI Contact Form.
Preamble
As the Internet shows so clearly, there is great social, technical, and financial benefit that comes from massive interoperability. This interoperability was not achieved in a vacuum, nor was it necessarily intended or happened as expected by the major players of the industry in its early days. But the processes and practices of the IETF, the W3C, the effect of events like the early Interop conferences, and the inclusive participation of virtually any party willing to follow the published protocols ultimately created something broadly useful and valuable.
Metcalfe’s Law predicts that the value of interoperability increases geometrically with the number of compatible participants, and Reed’s Law predicts that the utility of a network (implied by interoperable equivalence) increases exponentially due to the number of possible subgroups that interoperability enables. Both theories have successfully informed the investment of literally billions of dollars of capital investment as the Internet has become mainstream. Whichever law ultimately governs, interoperability is a positive function governing value, and thus any force that diminishes interoperability must be carefully scrutinized as it relates to ultimate and/or total value.
If interoperability is a grand goal as it relates to software, then standards are the critical tools for achieving this goal. Moreover, standards that permit:
- any license (free, open, or closed)
- any implementation
- any implementor
will expand the universe of interoperability compared with standards that militate against specific types of license, specific implementations, or specific implementors. Thus, for all the benefit that a standard can offer, the important question to ask is what the standard does or does not offer based on its implicit or explicit license or technology freedoms or restrictions.
At this point in time, it is has become largely intuitive across the industry and among users that broad and widely accepted standards are a Good Thing, better, certainly, than proprietary standards. In spite of these widely understood benefits, many market participants paradoxically believe that they can obtain a commercial or strategic advantage through unique, i.e., non-standard functionality. Some seek to burnish the perception of their products or technologies by claiming that they implement “open standards” while at the same time adding extensions that are not part of the standard. Others go farther, claiming that their unique implementations are themselves “open standards”, a reversal of standards logic. The result is that the (usually undefined) term “open standard” has become more of an aspirational term than a defining term, a problem that we seek to rectify.
While the Open Source Initiative is not in a position to completely define what is meant by the term “open standard” (a term that may well become enshrined into law in some countries, and hence the perview of legal authorities, not the OSI), we believe that the OSI does have both the opportunity and the obligation to try to constrain the problem. In particular, as custodians of the Open Source Definition, we believe it is essential to help decision makers and developers understand how “open standards” relate to “open source software” and to ensure that a poor understanding of the former does not discriminate against the latter.
The Purpose of Open Standards
The purpose of an open standard is to increase the market for a technology by enabling potential consumers or suppliers of that technology to invest in it without having to either pay monopoly rent or fear litigation on trade secret, copyright, patent, or trademark causes of action. No standard can properly be described as “open” except to the extent it achieves these goals.
The industry has learned by experience that the only software-related standards to fully achieve these goals are those which not only permit but encourage open-source implementations. Open-source implementations are a quality and honesty check for any open standard that might be implemented in software; whether an application programming interface, a hardware interface, a file format, a communication protocol, a specification of user interactions, or any other form of data interchange and program control.
To help industry participants (suppliers, consumers, and regulators) identify and specify standards that permit open source implementations, the OSI has defined a minimal Open Standards Requirement (OSR). The OSI has also created a set of Criteria that can be used to judge whether a standard fully complies with that Requirement.
Problems not solved (because not in scope)
The problem of creating a robust and high-quality software standard, one that meets the needs of vendors and customers, users and implementors, public and private sectors, now and over time, is complex and context-dependent. In some cases, experienced hands are essential to the successful birthing of a standard; in others, a single idea from a single mind ensures the integrity of the problem definition and hence the problem’s solution. The Open Standards Requirements for Software does not prescribe how open standards for software should be created, debated, ratified, and maintained except that they not preclude a viable implementation in open source.
There are any number of reasons why a particular open standard for software may succeed or fail in the market, and meeting the Open Standards Requirements is no guarantee that the standard will magically succeed on that fact alone. However, standards that preclude open source implementation are unhelpful to what is now a significant community of software developers and users, and it would be disingenuous to call any such standards “open” when they are, in fact, closed.
Issues
- How can we be broad enough to be relevant, yet precise enough to be defensible?
- Implementation vs. Specification
-
cf. Larry’s paper is at: http://www.rosenlaw.com/DefiningOpenStandards.pdf
- Everyone should be free to implement open standards in both proprietary and open source software.
- Open standards should be available to everyone on royalty-free terms.
- Open standards should be developed using a collaborative, balanced and consensus-based approval process.
- Open standards should be developed under formal and binding commitments for the disclosure and licensing of copyrights and patent claims.
- Open standards should be made available under reasonable reciprocal licenses that require licensees to share under the same terms their own patent claims reading on the standard.
- The specifications for open standards should be available to everyone on open source copyright license terms.