Statement made on 19 Feb 2009
Introduction
Based on the background provided in Analysis on balance: Standardisation and Patents please allow me to bring up the issue of patents on standards that are licensed in ways that discriminate against the Free Software (Open Source) paradigm and software implementations inside this market segment.
Some standardisation bodies like the W3C have adopted patent policies that prevent discrimination against any market segments. However, while other standardisation bodies like ITU-T, ITU-R, ISO and IEC follow a Common Patent Policy that is based on the principle that "a patent embodied fully or partly in a Recommendation | Deliverable must be accessible to everybody without undue constraints", they contradict this policy by accepting RAND conditions which are manifestly incompatible with the Free Software paradigm and disrupt software models used by it.
This situation is regrettable, because though it is already common practice for participants to assign Copyright to the standardisation bodies in order to free resulting standards from copyright encumbrance, no such practice exists for patents.
It is possilbe that governments or the European Commission would find it difficult to enforce such a practice against the expressed interest of some companies and organisations, but the European Commission is at liberty to decide which standards it accepts as sufficiently accessible to meet its requirements for transparency and non-discrimination.
This issue also came up in the discussions of WG2. Please find included the relevant section that was initially submitted by myself:
Exclusion from standards implementation
The bias against OSS in procurement is aggravated by discrimination against OSS in the licensing conditions for some IT standards. Over the past years it has become clear thatspecific patent licensing schemes, most importantly the so-called 'RAND' terms, discriminate against OSS implementation. This issue complicated the recent antitrust cases in Europe and was subject of a specific workshop on 'IPR in ICT standardisation' organised by DG Enterprise.
The workshop revealed a fundamental incompatibility of RAND models with OSS implementations, as well as a very controversial debate around this issue. From the perspective of OSS adoption, it could be said that RAND conditions fall short of the Common Patent Policy of ITU-T, ITU-R, ISO and IEC, which states that "a patent embodied fully or partly in a Recommendation | Deliverable must be accessible to everybody without undue constraints."
Examples of such exclusions can be found in various areas. One of these areas are the MPEG standards in multimedia, where innovation has been dramatically reduced before the recent development of the Dirac codec by the BBC as OSS provided a high-quality modern alternative that is not patent encumbered.
The resulting Action language was:
Reducing barriers
The barriers to entry are particularly harmful in the area of interoperability, where inability to implement standards leads to increased cost and reduces the reuse and recombination factor, which will be essential for the future IT industry.
The European Union therefore needs to which extent it can bring European standardisation bodies into line with the stated goals of the Common Patent Policy of ITU-T, ITU-R, ISO and IEC.
It would probably be helpful if this group came up with a suggestion for language to be used by the group working on procurement that would tie acceptance of standards in procurement to meeting the non-discrimination principle also in the licensing of the standard.
It is important to ensure that all software paradigms and the software models fostered therein can compete effectively when using standards, and that no distortion or legal uncertainty exists in such situations.
Statement made on 24 Feb 2009
Several concrete proposals have been made in the political debate over the years, some of which were also explained in the reference that was sent around with my initial email, http://fsfeurope.org/projects/os/ps
The idea to make the end-user subject to patent payments does not work, as it would make distribution illegal for 80% of Free Software and it does not help anyone to make customers liable.
As to the solutions that would solve the problem, not all of these items are usable in the context of a European Software Strategy. So we should probably begin by focussing on:
- Minimum definition of RAND to include all of industry, as the Common Patent Policy already states it should be.
- Recognition of standards bodies such as W3C and OASIS, with blanket approval only for bodies that have binding policies which follow the above principle.
- Procurement of solutions only if the standards meet #1 and #2
Statement made on 6 Mar 2009
There are fundamentally different goals for patents and standards, as illustrated by Mr Karsten Meinhold, chairman of the ETSI IPR Special Committee, when he stated that "IPRs and Standards serve different purposes: IPRs are destined for private exclusive use, Standards are intended for public, collective use." This tension has previously been managed through systems such as RAND, including RF-on-RAND ("Royalty Freeon RAND"), RAND-RF ("RAND Royalty Free") or RAND-Z ("RAND with Zero royalties"). While this may work for certain software models, it does not facilitate fair market access and competition for all software models currently competing in the market. The most notable issue arises with Open Source, a model that is hindered in multiple ways by the application of patents to interoperability and standards; this ranges from incompatibility regarding the application of royalty payments to Open Source through to (F)RAND terms discriminating against the model because they do not permit sub-licensing.
Proponents for systems such as RAND suggest that some Open Source licences may not be hindered by such terms, and therefore (F)RAND or similar is perfectly adequate toprovide fair market access for Open Source and proprietary models. However, while a limited subset of Open Source licences exclude patents from their explicitly addressed concerns, the majority of Open Source is not licensed in a way that permits theapplication of patent restrictions. According to Blackduck research, 66.57% of projects use GPL family licences that explicitly prohibit the application of patent restrictions on covered software. Exclude these licences from a standard, and you are excluding 2/3 of the Open Source model participants from accessing a standard.
All current business models, including those incorporating Open Source, are legitimate. This raises the question whether it can be considered Fair, Reasonable and Non-Discriminatory to exclude this legitimate part of economy by choice of patent licensing terms. The European market must support innovation and inclusion, and it cannotallow the distortion of access to competitive technology in favour of any one modelin the field of software development. Conversely, the inclusion of models does notmean discrimination against previous models. Instead it promotes competition and innovation and as a consequence leads to a more healthy market for all participants.
This paper must be neutral with regards the software models available, be they proprietary, Open Source or mixed model, and it must facilitate access to the standards underpinning a competitive market. This is a fair approach, and does not run the risk of pre-empting policy by implicitly or explicitly positioning any one model more favourably. That could be premature and exclusionary with regards competitive flexibility.
Statement made on 20 Mar 2009
Re "What works":
I believe this section is a bit too optimistic.
The rather turbulent discussions and procedural irregularities in ISO, many of which were raised in regards to behaviour of European National Bodies, caused a loss of faith in the operational quality of several National Bodies, and ISO as an organisation.
As any crisis, this provides both challenge and opportunity. There is an opportunity for European bodies to take more leading roles, and there is a chance that European National Bodies will improve operational quality as a result.
FSFE considers standardisation a field of regulation in the public interest. And like other fields of public interest it should adhere to certain minimum principles of transparency, accountability and due process, which in our experience are not always maintained.
The technological aspects of standardisation should certainly be guided by principles of being voluntary, market led and industry driven. But procedural and public interest considerations do apply, and cannot be left for voluntary inclusion by industry.
Re "Cooperation between organizations"
I find this section quite interesting, as it describes the strategic abuse of standardisation as leverage to distort markets.
From a systematic perspective, I would describe this as a function of the value of individual standards. The growth of the IT industry overall in combination with the extraordinarily strong (and growing) networking effects have dramatically increased the value of standards, and thus the incentive for abuse of the system.
Mechanisms that previously made manipulation sufficiently costly to prevent large scale abuse no longer provide an effective deterrent when compared to the value derived from protecting markets that can deliver up to 10bn USD yearly revenues through standard abuse.
This is where additional procedural safeguards and rules of due process can help and where governments will indeed need to build up more competency in particular in the public benefit impact of standards.
Re "Relevance"
This section is interesting, and might possibly be expanded, e.g. with some evidence that shows which are the factors that make standards win, which I agree are not financial. Evidence is provided for instance by the early attempts to create internet protocols, or the image format debates around patent-encumbered GIF that led to the creation and wide deployment of PNG.
Re "Open"
Buzzwords are part of this industry for as long as I can think.
The document uses a couple of other buzzwords in other places without similar digressions in the buzzwordyness of a certain term. So I would suggest to either create an appendix in which all terms that can be called buzzwords are treated equally, or to focus on the substance of the debate.
Standardisation is open to a large extent by definition, e.g. BSI's definition of a standard, which describes it as "a published document that contains a technical specification or other precise criteria designed to be used consistently as a rule, guideline, or definition. [...] Any standard is a collective work. Committees of manufacturers, users, research organizations, government departments and consumers work together to draw up standards that evolve to meet the demands of society and technology. [...]"
So the collective and inclusive creation as well as common usage are part of the definition of what constitutes a standard.
An "Open Standard" would therefore have to meet higher criteria for openness, which in my experience are usually connected to lack of encumbrance from private exclusive rights, but can also be connected to procedural issues, as you describe.
It might be worthwhile to provide reference to what different groups consider good criteria for an Open Standard, e.g. the definition used by the SELF EU Project, the 2008 Geneva Declaration on Standards and the Future of the Internet or the Document Freedom Day, which FSFE decided to follow, as well: http://fsfeurope.org/projects/os/def.en.html
Re "Interoperability"
I believe the ECIS statement was initially made by IEEE, which defines interoperability as "The ability of two or more systems or components to exchange information and to use the information that has been exchanged." (see http://en.wikipedia.org/wiki/Interoperability)
Re "Interoperability outside standards"
Statements such as "Real world cases that go beyond some agreement on exchange formats are rare in the commercial as well as in the open source corner" need to be rephrased.
Commercial Free Software is ubiquitous today, as non-commercial proprietary software has always been. "Commercial" and "Open Source" are false antonyms, and other groups (namely WG7 and WG2) have discussed this issue in some detail, their work can provide the basics for the discussion in this group, I believe.
The section also forgets to mention that Free Software (a.k.a. Open Source) also enables a new form of interoperability on the basis of actual code, instead of standards, which is a meta-description of properties of the code to write.
By allowing direct inclusion of the code responsible for reading and writing certain file formats and protocols, a high level of interoperability can be achieved in practice at comparatively low cost.
Re "Things in common"
On patent trolls:
Patents are created as exclusive rights that can be traded, e.g. a University can do some research, file a patent, and then sell the rights in that patent to a third party in order to raise funds for the University. This practice is in fact being promoted heavily in some countries.
The third party that bought the patent now needs to bring in sufficient return on investment that should a) help cover cost for patents that turned out worthless, and b) maximise profits for its owners. So it will do what it can to maximise the patent revenue, including litigation.
If it is a company that specialises on this kind of activity, it is generally described as a "patent troll." They are in other words a logical and unavoidable consequence of a system that grants exclusive rights on ideas as items of trade for which profit should be maximised.
Trademark like provisions of "use it or lose it" might help to mitigate some of these effects, but could also lead to a dramatic increase in patent litigation fueled by the desire to maintain patents valid.
So the systemic cost (cost of patent fees, legal counsel, research, and litigation) of patents which is already described as unsustainable by some could skyrocket in response.
Re "The Open Source Royalty Free Camp"
I would prefer if the document could avoid usage of terminology that is laden with implications, such as "camp."
The final paragraph of this section mixes several unrelated issues and is based on incorrect assumptions, including the false antonyms "commercial" and "open source".
Free Software (a.k.a. Open Source) is a diverse ecosystem that Gartner expects to include most of our economy in one form or another by November 2009. Parts of this ecosystem are non-commercial, but the larger part tends to be commercial in usage and production.
Some licenses (including the GNU LGPL and the Apache License) indeed allow for code to be included in proprietary works, but this is unrelated to the issue of patents and standardisation.
So is the issue of business models, which is orthogonal to actual software models, or development models. See http://blogs.fsfe.org/greve/?p=260
Portraying the discourse of software patents as a Free Software issue is biased and incorrect, however. The resistance against software patents included all sorts of companies, proprietary software companies as well as Free Software companies, and many other SMEs that were users of software.
Since this debate is likely to open a can of worms, I would suggest that we remove the generalised statements entirely, or we are likely to repeat many of the discussions that took place around the software patent debate over the past years.
There is also another suggestion that was added, which is a general exception that renders patents unenforceably against interoperability.
This was agreed-upon position for the vast majority of participants in the debate on the software patent debate years ago, including the majority of proprietary companies, so it should be added elsewhere, not in the section specific to Free Software/Open Source.
Re "OSS and Standards"
The reference to "Cathedral and the Bazaar" is unfortunate as that text confuses development models and software models with its use of ambiguous terminology (see reference above). Aside from the historical irony that cathedrals were one of the best examples for application of what is called a "bazaar model" in this paper, there is Free Software / Open Source that is developed in a very closed fashion, and proprietary software that is developed in a highly collaborative way.
So generalisations drawn from a particular development model are likely to apply to all software developed according to that development model, and will not be correct for the software model differentiation that is provided as context by the heading of the section.
The section furthermore reflects use of false antonyms ("commercial" instead of "proprietary") and parts that violate the basic premise of the software model.
This is in particular true for "Private use would be excluded and commercial use would then trigger royalties." which is in violation of the basic fact that the software model is defined by freedom of unlimited use for any purpose, commercial or non-commercial.
In other words: a solution in which non-commercial use would be treated differently from commercial use would mean that the software no longer qualifies as Free Software/Open Source and would be proprietary in its entirety.
Re "Open Source and Venture Capital"
The statement "But the search was in the field (although human readable) of the core mission of the companies financed by VCs" implies contradiction to the previous sentence, but it does not contradict it, as there are no strict parameters that this field should faithfully represent the question of Free Software usage, and in which way.
Please avoid derogatory terminology, e.g. "viral GPL licensing", the neutral term for this is Copyleft.
Re Google: Although details are not public, it is known that Google uses a lot more Free Software than just GNU/Linux. The statement that patents exist does not relate to the basic fact that Google is an enterprise that was enabled by Free Software and releases a great deal of Free Software. Other Free Software companies (e.g. RedHat) also hold patents.
These are typically held for defensive purposes only, see
http://www.redhat.com/legal/patent_policy.html
But is this section really primarily relevant to standardisation?
It seems that this is an issue that would really belong into WG7, which has been dealing with these issues in much more detail. That said, I agree with the action item conclusion that additional discussions on this issue would be highly useful.
Re "Issues around Standardization and Procurement"
It was consensus among the Working Group on Open Source Software (WG7) that it was legitimate, although not strictly recommended, for governments to mandate Free Software in order to gain strategic control over their own infrastructure.
This is not primarily an issue of cost, but infastructure control, and does not discriminate against any particular vendor or technology.
Re "Trends / Open source models"
If there is a counter argument, it would be good if that argument were stated, and not just its existance presumed. The paragraph preceding the statement is based on the fact that proprietary software can have Free Software enclosures, which is true, but does not relate to the issue of excluding Free Software from standards implementation in any way, as it relates to proprietary, and not Free Software.