In 2005, the European software industry was embroiled in the climax of a major conflict over the extension of the patent system to cover software. This fight had started in 1999 with moves by the European Patent Office (EPO) to change the European Patent Convention (EPC) (the treaty that created the EPO as an independent international body) to remove the exclusion of Article 52 which says that "programs for computers" are not patentable "as such".
Moves to change the EPC failed, so the patent industry and large software firms together with the European Commission pushed to create a law at the European Parliament (EP) which would have explictly allowed software patents in the largest economic block in the world, independently of any provisions in the EPC.
The FFII, FSF, and the small-to-medium software firms that make up 80% of the IT sector campaigned on a pure abolitionist platform: No Software Patents. The economic analysis of software patents made it clear, early on, that the benefits of software patents accrued only to their holders (patent trolls and large software firms), and those who profit from litigation, at a large cost to society. The US experience had already shown that software patents created litigation, slowed down progress, blocked competition, and were fundamentally incompatible with a fast-moving industry based on innovation and copyright.
At this time Red Hat had filed around 50 software and business method patents at the US Patent and Trademark office (USPTO), at the EPO, and at the WIPO. For example, EP1691276, "System and method for verifying compatiblity of computer equipment with a software product", filed in January 2005, and EP1659493, "Replacing idle process when doing fast messaging" (by Alan Cox). Today Red Hat as about 350 software and business method patents filed, indicating that their patenting activity has been consistent and growing.
The first Red Hat software patent filed at the EPO was EP1312195, "Method and apparatus for handling communication requests at a server without context switching", filed in 2001. Red Hat was thus at the forefront of filing software patents at the EPO.
The problem for clients of the patent system was that these EPO patents were not upheld by the national courts where infringements could be prosecuted. Specifically the UK upheld the original EPC interpretation that patents on software were not enforceable, while Germany generally tolerated software patents.
The European Commission was fully convinced by the arguments of large software firms, and the patent lobby, that without software patents Europe would be unable to innovate. The logic of the day was, "more patents means more innovation", despite the lack of any evidence to support this, and mounting evidence from the US of the contrary.
With such political pressure to open the doors to full patentability of software, the Abolitionists understood that compromise was not possible, and defined a firm, unambiguous position: No Software Patents. As a theoretical basis, they developed a set of "10 core clarifications" (10CC) designed to resist attack by patent attorneys. The 10CC were the basis of the Abolitionists amendments, which were presented in Parliament and gathered so much support that by June of 2005, at the second reading, the originally pro-software patent Directive had threatened to become a total ban on software patents. The pro-swpat lobby realized the risk, voted to reject the entire Directive, and switched to a number of backup plans, which today are still unrolling. According to parliamentary insiders, the Software Patents Directive was the most heavily lobbied directive of all time.
Of all the firms involved in the conflict, those involved both in the free and open source (FOSS) business, and involved in software patenting, were the most crucial. A large part of the argument against software patents was that it they were dangerous to the highly competitive and open FOSS sector and that this sector was key to European technological progress. IBM, Sun, Novell, and Red Hat were the main firms on both sides of the debate. Red Hat, the only one with a pure FOSS business.
While this climactic debate was happening, Red Hat defined their position with respect to the US and the European patent systems' treatment of software. Red Hat argued - as they have consistently before and since then - that the patent system was not working for software:
The licensing of software patents has become an industry unto itself… Today in the U.S. information industry, we are observing the building of an iron curtain of intellectual property.
The problem, in part, lies in the assumption that patent law, as presently applied to software, is infallible. … We have now established a patent system for protecting software that is failing to promote that progress.
However, when it came to proposing solutions, Red Hat rejected the Abolitionist position and endorsed the position of Microsoft's General Counsel, advocating improvements to the patent system to improve so-called patent quality:
Even Brad Smith, General Counsel of Microsoft, has called for a number of these reforms, as well as greater harmonization around the world. This is an area on which Brad and this author agree. Specifically, this author advocates:
- Non-diversion of PTO fees;
- Third-party participation in the patent examination process;
- Post-grant oppositions;
- Challenges to the validity of patents to be based on a "preponderance of the evidence," not "clear and convincing evidence" standard;
- Restoration of the once-strong non-obviousness standard;
- A higher threshold in finding willful infringement;
- A higher threshold before granting injunctive relief;
- Publication of all patent applications after 18 months;
- Harmonization with Europe to narrow the scope of what is patentable in the software and business method arenas as well as assuring the right to interoperability.
The Abolitionists viewed these "reforms" to be ways of reinforcing the patent system, and making software patents more, not less dangerous to the software industry. This echoes a common division of opinion. While the four firms cited invested in improving the patent system (to prevent devaluation of their own portfolios), the Abolitionists fought for the total end of software patents.
To clarify why: a single patent can demolish an small company, and if the patent is "higher quality", it becomes harder, not easier to fight. Eliminating junk patents leaves space for more resistant patents which can do more damage.
Red Hat endorsed the work of the European Commission, which was a forceful proponent of software patents, with little sympathy for small-to-medium sized firms, nor the FOSS economy:
There are, however, solutions available, both immediate and long-term, that are applicable in the U.S. and abroad. An early lead is being taken in Europe as the European Commission and European Parliament work to establish a uniform system for software patents (more properly, computer-implemented inventions) across the European Union.
And in this statement they endorsed the "proper" term "Computer Implemented Invention" (CII). The term is a clever circular definition designed to circumvent the EPC restrictions on software patents. An invention is by definition patentable, the terms "invention" and "patent" are largely interchangeable. The words "computer implemented" mean that the invention is not software (as such), it just happens to run on a computer.
Thus any party that accepted the terminology would accept inevitably that computer programs could be patented. "Software" became relegated to source code, and the EPO was able to argue that inventions that ran on computers (which could be patented) were fundamentally different from software (which as source code was adequately protected by copyright). By splitting off the "invention" from the "software" the EPO was able to bypass its own rules and allow patents on software.1
One of the great successes of the Abolitionists was to recapture the terminology by getting the term "Software Patents Directive" into common use. In the propaganda war, whether one said "CII" or "software patent" largely defined one's position as for, or against, software patentability.
Regarding the details of the Directive, whatever its name, Red Hat did not argue for abolition. Instead, they argued for amendments to eliminate business method patents and promote software patents with a higher standard of obviousness. They also argued for protection for interoperability:
32 While the final result of this legislative process remains uncertain, what is certain are some of the amendments that are likely to be incorporated into the legislation. First is the definition of the term "technical contribution" as it is incorporated into the legislation. A narrow definition of this term, along the lines of its interpretation to date by the courts of Germany, will eliminate the vast majority of business method patents and will restore a substantial non-obviousness test to software patents. The second is the addition of a statutory protection of the right to create interoperable products. Should those amendments be adopted, the European legislation will have gone a long way toward addressing some of the more severe problems inherent in the U.S. patent system.
The Abolitionists regarded this as betrayal from a key firm that should have been firmly in the Abolitionist camp. Red Hat may have been arguing from a pragmatic position in which it seemed the Directive would pass. However, their statements provided ammunition for the pro-swpat camp and demonstrated that significant FOSS firms explicitly supported software patents.
Since amendments are voted on individually, there was no guarantee that the interoperability amendment would be voted, if their first also was.
While promoting amendments to harmonise software patents on the German model, Red Hat did also argue for changes to the patent system as it applied to software. They said:
At the same time, we should not assume that, even as modified and improved, the application of patent law to software is inherently the best or most logical means of protecting the interests of the developing party. Perhaps we should consider a new paradigm that is specifically designed for software and incorporates the best elements of both patents and copyright. Some characteristics of such a paradigm could include:
- a shorter term of protection (5-7 years would maintain the speed of innovation);
- protection only for complete systems or features, not components (avoids the trivial);
- strong protection (reward first movers);
- published source code (allows knowledge to advance); and
- interoperability.
It is a central tenet of the patent industry that there is a single patent system that covers all subjects. Any attempt to create domain-specific patent models is fiercely fought by the pharmaceutical industry, which fears the weakening of the patent system. So the above proposals were highly speculative and irrelevant to the Directive (and were not proposed in any amendments), with the exception of interoperability.
Even so, the Abolitionists regarded these as unacceptable. A 5 or 7-year patent, following several years of examination delay, would mean a 10-year block on competition in emerging areas of software technology. The "strong protection" cited, is code for "faster and more effective litigation". The avoidance of "trivial" components is meaningless since terms like "complete systems" and "features" cannot be firmly defined. And interoperability, while a valid goal in itself, does not protect independent re-invention.
It is worth understanding why the Abolitionists were so absolute, and so successful, in demanding "no software patents" while pragmatic firms like Red Hat aimed for a softer, more "realistic" middle ground. There are two insights which drove and united the Abolitionist movement.
First, the understanding that the patent system was fluid, and adapting to new opportunities in an organic fashion. From the point of view of a small software developer focused on new products and clients, or an economist, the patent system looks fundamentally parasitical. The growth of the patent system into software and business methods happened slowly, but methodically, through the redefinition of language and the infiltration of the political establishment with ideologies such as the perfection of the patent system, and "more is better". Once understood, it is easily predicted: the patent industry can overcome any barriers that are not absolute. There are no lines that can be drawn between the "good" and the "bad" software patents, that a determined patent attorney cannot cross, given time and money and a cooperative patent office.
This is why the Abolitionists unanimously and without compromise rejected all attempts to build such barriers out of terminology like "technical". And it is why they regarded with hostility anyone, especially lawyers, who proposed them.
The second key insight which drove the Abolitionists was the understanding of the economics of the patent industry. In its marketing, the patent industry claims to promote innovation and help small inventors. In reality, the patent industry is overwhelmingly tuned to the needs of its "clients": large firms and patent trolls. For both these classes, patents are sensible and economically worthwhile. Large firms mostly fight smaller competitors. Patent trolls build and defend a patent licensing business. The small-to-medium software developer finds themselves at the sharp end of the stick, with respect to larger patent holding firms. Trolls tend to attack large end users, and firms with too much money. But they can also attack SMEs, in markets where it is profitable.
Given that it is the large firms, and the lobbyists, who exert the most political influence, especially within the closed circles of the EPO, it seems inevitable that the patent system would and will evolve to provide the widest possible patentability, the strongest possible litigation, and the biggest profits for its clients, at society's expense. The US, often ahead by several years, provided a clear scenario of what could happen in Europe.
By 2001 Red Hat along with others like Microsoft and IBM were pushing the envelope on what could be patented in Europe, investing in lawyers to undermine the exclusions of EPC Art 52. Red Hat's statements must be understood in this context. As they took patents at the EPO level, they argued against abolition, and in support of "reforms" that would improve the value of Red Hat's own growing patent portfolio. While they pointed out problems with software patents, they did not turn this into political action. Instead, they endorsed the propaganda term "Computer Implemented Invention" and they lobbied for amendments that would legislate for, not against, software patents across Europe where the letter of the law still forbade them.
The FFII President Benjamin Henrion was in the European Parliament in 2005, collecting signatures for the amendments that would ban patents on software. He recalls, "I was in the EP. They did not helped us collecting signatures, going for interop instead."
The evidence thus shows that Red Hat, along with IBM, Sun and Google, did not campaign for an end to software patents in Europe, neither when that was within reach in 2005, nor today when it remains a reachable goal. Instead, they focused on:
- Trying to define distinctions between "bad" (business method) and "good" software patents, despite the fact that no such distinctions have ever successfully been formalised and upheld;
- Aiming to "harmonize" US and European patent law, which has invariably meant extending, not restricting, patentable areas;
- Aiming to "improve" the quality of issued software patents, despite evidence that "good" software patents do as much damage as "bad" ones;
- Lobbying for "interoperability" exemptions.
While interoperability seems a noble goal, in the context of these US firms, it means "Linux-to-Windows interoperability". Red Hat was thus working, together with Sun, Oracle, Google, and IBM, to establish US-style patenting of software while carving out enough space to protect their own Linux-based businesses.
Red Hat, instead of working with the Abolitionists to end software patents in Europe, was working to rewrite the law and remove the protection from patents that Europe's small software sector had enjoyed until then. While this seems normal and obvious for a firm that needs to answer to shareholders and investors, it is Red Hat's own statements that baffle us. If Red Hat seeks patents "to discourage patent lawsuits by giving us the ability to retaliate against potential patent aggressors by asserting counter-claims as a defence", why did they not work for abolition in Europe in 2005 and why are they not working for abolition in 2009?