What are standards and why are they important? This paper tries to examine some of the literature available that studies the interaction between standards and Free Software, trying also to define a category of standards that can be safely implemented by Free Software.
The term 'standard' has different meanings, but in Information Technology we can define standard (using the definition by Krechmer) as 'shared agreement between various actors enabling communication between different products or services'. As the International Standards Organization (ISO) reports on its website, if there were no standards we would all notice immediately. The existence of such agreements implies that each product or service that complies with the standard behaves predictably and independently from the specific implementation. As an example, consider the GSM standard for mobile phones and networks: it is implemented by different phone's manufacturers and different network operators, but we can still call everybody independently, since all interoperate. Interoperability is the ability of information and communication technology systems and of business processes to exchange data and allow sharing information and knowledge.
Standards can be classified according to their process of definition and control of evolution. Our scope is to identify a class of standards that can be safely implemented in Free Software. To evaluate the impact on the adoption by Free Software developers it is important to consider the existence of licensing policies of possible patents registered on the same standards (or parts of them). Patents on processes and algorithms, also known as 'software patents', were not allowed until mid 80s in the USA. In Europe they are not allowed literally by the European Patent Convention, but the European Patent Office allows them. Patents on standards or parts of them can significantly impact the adoption of such standard in Free Software implementations, as we will see later.
We can classify standards simplifying the classes suggested by Fuggetta and Cerri as follows:
Proprietary standards can further be divided into:
It is worth noticing that all these proprietary standards hardly fit into the definition of standard: they are not agreed between different actors, but imposed by the owner of the specifications on its partners often without any possible discussion. Still it is worth mentioning them, since some disclosed proprietary standards are safe to implement in Free Software. Probably this category deserves a different name altogether.
Open standards can also be 'de iure' standards when they are approved by international standards organization recognized by local governments, like ISO, ANSI and UNI. Open Document format is a Open Standard de iure, ISO/IEC 26300.
When standards are covered by patents third party implementations must be authorized by the owner of the patent. Also, users of software that implements the standard need to receive a license to use the
patented technique.
Not necessarily.
The majority of consortia and standard bodies allows open standards also when they are covered by patents. At the moment, none of them has patent policies that allow to safely implement their standards as Free Software. Not even W3C's patent policy is safe, even though it is the only one that at least forces mandatory and royalty-free licenses for the implementation of their standards.
Also, sometimes it is safe also to implement proprietary disclosed standards if conditions on the specifications and its patents don't interfere with the unlimited rights to run, study, copy and distribute Free Software.
For these reasons we can specify another category for standards:
Basically, this would mean that implementations cannot be limited by NDAs or other agreements. Unencumbered open standards, being free and safe to implement for all software developers (Free and non-free software) allow for greater competition on the market, with lower barriers to interoperability and higher quantity of solution providers. The increase in interoperability lowers the 'exit costs' from particular software solutions, also known as 'freedom to leave'.
Consider the recent case of Public Administrations wanting to start using the ISO 26300 standard for its office related tasks and the high costs that these PAs are facing only to migrate the existing archives under undisclosed proprietary standards to a real unencumbered standard.
Free Software and unencumbered open standards are not the same thing and must not be confused. Although it is likely that an unencumbered standard will have its reference implementation released as Free Software (as it's the case of OpenOffice.org or kerberos) nothing impedes that there are also non-free software implementations.
Fuggetta, A., Cerri, D., Open Standards, Open Formats and Open Source,
July 2006 (submitted for publication,
http://www.davidecerri.org/en/doc/openness.pdf)
Krechmer, K., The Meaning of Open Standards, The International Journal
of IT Standards and Standardization Research, Vol. 4 No. 1, January -
June 2006, http://www.csrstds.com/openstds.pdf
Lecocq, X, Benoit D., Open standard: role of externalities and impact
on the industry structure, 2002, MIT, su
http://opensource.mit.edu/papers/lecocqdemil.pdf
Perens, B., Open standards, principles and practice,
http://perens.com/OpenStandards/Definition.html
Shapiro C, Varian HR., The art of standards wars. 1999, California
Management Review
EU IDABC, Documentation on the Promotion of Open Document Exchange Format, 2003, on http://ec.europa.eu/idabc/en/document/3439/5585
EU IDABC, European Interoperability Framework, 2005, su
http://ec.europa.eu/idabc/en/document/3761