by Jeremy Malcolm2
Patents are statutory monopolies granted to the owner of a new invention and lasting for up to 20 years. Although patents have existed for centuries, they have only more recently become applicable to computer software. This has introduced the risk of some fundamental standards of computing and the Internet becoming encumbered, and has begun to stifle the free development of open source software. Fresh concerns have widely arisen in Australia in light of the Australia-United States Free Trade Agreement. This presentation will look at why software patents seem to be so readily granted over what developers would never consider as "new inventions", and what can be done about it. The implications of the Free Trade Agreement will also be addressed.
Software patents are a hot issue for open source. Linux Australia held its first Open Source Forum on the topic last October,3 and Free Software Foundation General Counsel Eben Moglen has recently said, “the threats posed to software by patents are a serious difficulty for us”.4
Software patents are not a problem for open source developers alone, though open source developers in the main are probably less well-resourced to deal with them than developers of proprietary software. They don't have the money to pay patent licence fees, or to defend patent infringement lawsuits, nor a stock of software patents of their own to cross-licence to whomever is alleging infringment.
Most open source developers don't even have the resources to find out whether their code infringes an existing patent. It's difficult even for a lawyer or patent attorney to find that out. There are so many patents for computer software out there, particularly in the United States, that developing an application without infringing software patents has become a very hit and miss affair. It may be that the first you know of an infringement in your software is when the patent owner comes knocking at your door.
Patent rights were originally developed in England in the 15th century. In simple terms a patent is a government-sanctioned monopoly right granted over a new, inventive and useful device or process. During the term of the patent – normally 20 years in Australia – the patent holder has the exclusive right to make or use the invention, or to license others to do so.
There is no requirement that the patent holder actually use the patent themselves, which is one respect in which the patent system has fallen down. It has led to a situation in which many patent holders don't actually create anything useful themselves, but simply use their patents to earn money from others who want to do useful work in the same area.
If their patents are broad or numerous enough, they can also use them to warn off competitors from attempting to compete with them in a certain field, or they can use their patent portfolio as a bargaining chip to cross-licence with their competitors, allowing each of them to share the market while crowding out smaller competitors.
An Australian patent – over software or anything else – is obtained by drafting a specification that fulfills the necessary criteria (which will be discussed below) and lodging it for examination by IP Australia, the Federal Government's intellectual property registry.
The specification is a highly specialised and technical document that sets out the method of operation of the invention. These are drafted by patent attorneys, who are not lawyers as such; they are required to hold scientific or technical qualifications that enable them to draft patent specifications in that area.
The specification is made available to the public, so that when the patent expires, others can take advantage of the innovation. In theory, the specification can be used by competitors to try to find a way to work around the patent in developing a similar or interoperable product. In practice though, patent specifications are drafted in a peculiar kind of language that makes sense only to other patent attorneys.
To obtain an Australian patent and renew it for the maximum period of 20 years costs around $15 000. If you wish to have patent protection in multiple countries of the world, you will pay considerably more. Unlike copyright protection, patent protection in Australia does not automatically give you similar protection in other countries. It only gives you a one-year head start to file a patent application in other the countries where you desire protection.5
The three requirements of a patentable invention in Australia are that it has a patentable subject matter, that it is novel, and that it involves an inventive step (or is non-obvious). Similar requirements exist in most other countries of the world.
Patentable subject matter
A patentable subject matter is a useful man-made process or product, as distinct from an abstract idea or a purely artistic work. For many years, computer software was considered not to be patentable unless it formed part of a physical process. So for example, software that controlled a physical device such as a robot or an oil rig would be patentable, but spreadsheet software would not be.
Australia was one of the first countries to broaden the patentability of computer software. In a 1992 case6, IBM was found to be entitled to patent a new method of rendering curved images on computer displays. Guidelines subsequently issued by the Patents Office specified that a patent could be granted for software if it had a “commercially useful effect”. The validity of this guideline was confirmed by the Federal Court in 1994.7
Examples of “commercially useful effects” include an improved image as in the IBM case, or a more effective way of performing a calculation, compression or storage routine. In fact, a skilled patent attorney should be able to draft a software patent specification so as to demonstrate a commercially useful effect for almost any novel and inventive software application.
The second requirement of a patentable invention, that of novelty, requires that the invention was not known to the public before the patent application was lodged. The novelty of an invention is determined by the patent examiner conducting a search of literature in the relevant field, to determine what the state of the art was at the date of the patent application being filed.
The novelty requirement also necessitates that the invention has not been disclosed publicly by the person applying for the patent. An example of the application of that rule is that the applicant must not have offered the product for sale before applying for a patent. This limitation has been relaxed since 2002, when a grace period was introduced which allows the invention to be disclosed within up to one-year prior to the application being lodged.
The requirement of novelty is not actually as strict a limitation as you might suppose. If an invention simply combines two well-known pre-existing products or techniques – say, a method of encryption and a method of compression – the combined invention is still likely to be considered as novel.
Such an invention may, however, be found to lack an “inventive step”, which is the third and final requirement that a patentable invention has to fulfill. An invention that lacks a sufficient inventive step is one that would be obvious to a suitably skilled person, such as another programmer who had access to technical literature in the field.
When determining whether an invention that builds on pre-existing knowledge is sufficiently inventive or not, a patent examiner tries to avoid the benefit of hindsight, because some of the best inventions ever devised are also some of the simplest. However many would contend that the bar of inventiveness has been set far too low in the case of software patents, as will be discussed later.
Patent infringement actions
A person who infringes a patent by making or using a patented invention without permission can be sued by the patent holder for damages. Perhaps the most distinctive feature of patents, in comparison to copyright law, is that it is possible to infringe a patent unintentionally, and that this makes no difference to your liability for damages.
One of the few grounds on which a patent infringement claim can be defended is to argue that the patent should never have been granted to begin with, because its subject matter was not patentable, or because it was not novel, or did not contain an inventive step. If you can demonstrate these things, you may be able to have the patent annulled. However, the cost of this exercise should not be underestimated.
Encumbrance of Internet standards
Most of the controversy that has arisen over software patents has been incited by the opportunistic attempts of many patent holders to lay claim to some of the fundamental building blocks of the infrastructure of the Internet. Unisys, for example, only began to enforce its patent for the LZW compression algorithm used in GIF format graphic files after those files became a de facto standard image format for the World Wide Web. (That patent has since expired in the United States in June 2003.)
British Telecom went even further, attempting to lay claim to the concept of hyperlinks that are fundamental to the Web. Its claim, based on a 1989 patent that was originally applied for in 1976, was rejected by a New York District Court in 2003.
Increased risk to open source developers
In the course of an interview for ZDNet last month about the dangers of the extension of software patents in Europe, I stated:
There is no question that some of the open source software that is out there – such as the Linux kernel itself – has got patent violations in there. That is acknowledged. There is more danger that those potential violations will be litigated.
The resulting furore on Slashdot was deafening. But my statements aren't actually terribly controversial. Last year, Open Source Risk Management published a paper by Free Software Foundation lawyer Dan Ravicher in which he claimed to have found 283 potential patent violations in the Linux kernel.8
In 2002 Linus Torvalds himself said, “Everybody in the whole software industry knows that any non-trivial program (and probably most trivial programs too, for that matter) will infringe on some patent. Ask anybody. It's apparently an accepted fact...”.9
Admittedly, many of the patents are held by “open-source-friendly” companies such as IBM, which has pledged10 not to enforce its patent portfolio against Linux. And many of the other patents may not have been granted correctly in the first place, and hence may be vulnerable to attack by a well-resourced defendant. But even so, there do remain many apparent patent violations in open source software including the Linux kernel, and those who would shoot the messenger are simply sticking their heads in the sand.11
That is not to say that open source developers have done anything wrong by transgressing into patented territory. In all likelihood they didn't even know that they were doing so. What it does mean, though, is that they need to be careful to avoid violations to the extent that it is within their power to do so, and I'll provide some suggestions in that regard later.
Software patents for obvious inventions
It is not only the opportunism of patent holders that has given rise to controversy, but also the poor quality of the patent examinations that is perceived to have allowed some rather obvious or widely-used inventions to be patented. Amazon.com's infamous patent for its “one-click” on-line ordering system is the most commonly cited example.
One of the reasons for the poor standard of assessment of software patents is that traditionally the state of the art in a particular technical field is assessed by reference to journal articles and papers. Software inventions are not typically documented in those forms, but more often in source code or software manuals. Patent examiners do not make reference to these sources, which means that they are commonly unaware of what the state of the art is.
Patent examiners also tend to be less well paid than the professionals whose inventions they are examining, with obvious implications for the quality of personnel who are attracted to those positions (notwithstanding that Albert Einstein was a patent examiner in his early working life!).
Software patents in the United States
The largest number of software patents are those registered in the United States. Under United States law, it was decided in 199812 that a method of doing business (or a software program) will be patentable if it produces a useful, concrete and tangible result, rather than just being an abstract idea. As with the Australian test, a skilled patent attorney can very easily draft a patent specification for almost any software application so that it demonstrates a useful, concrete and tangible result.
The Australia-United States Free Trade Agreement
Our Free Trade Agreement with the United States was settled between each country's trade ministers in Washington DC on 8 February 2004. The implementing legislation was passed by our Federal Parliament on 13 August and 9 December 2004.
The most significant changes that the Australia-United States Free Trade Agreement introduced were not to do with patent law, but rather copyright law, which covers computer software and also music, literature, movies, and television. Of the changes that the agreement made to copyright protection, there were two that are most relevant to computer software.
The first of these changes was the extension of the term of copyright protection from 50 years from the date of the author's death (or 50 years from the date of first publication in the case of a corporate author), to 70 years.
The second change of note was the introduction of an equivalent regime to that provided by the United States Digital Millennium Copyright Act for the resolution of disputes between those who are, or claim to be, copyright owners, and Internet service providers who are hosting allegedly infringing content.
Software patents and the FTA
Ironically the only change to patent law made in the wake of the Free Trade Agreement was actually a new ground of opposition to the grant of a standard patent-namely, that the invention is not useful or has been secretly used.
However the text of the Free Trade Agreement foreshadows moves towards the harmonisation of our patent law regime with that of the United States (which effectively means bringing our laws into line with theirs). Clause 14 of article 17.9 states:
Each Party shall endeavour to reduce differences in law and practice between their respective systems, including in respect of differences in determining the rights to an invention, the prior art effect of applications for patents, and the division of an application containing multiple inventions. In addition, each Party shall endeavour to participate in international patent harmonisation efforts, including the WIPO fora addressing reform and development of the international patent system.
Whilst this has not been translated into any concrete form in the enabling legislation, it does suggest the long-term intent of both countries to create a uniform patent system. Unlike the copyright system, our patent system is presently quite separate to that of the United States. Patents granted there are not automatically recognised here, nor vice versa.
Moves towards unifying the patent regimes would not altogether be a bad thing for Australian IP owners, because it would cut both ways: it should allow for the streamlining of the recognition of Australian patents in the United States as well as the reverse, and could obviate the necessity to search two separate patent databases.
However it is a bad thing for open source developers, and for those calling for the reform of the patent regime, as it restricts Australia's ability to foster uninhibited free software innovation, and to legislate on patent law independently of US interests.
Software patents in Europe
The European Patent Convention, a pre-EU instrument dating from 1974, actually expressly excludes “computer programs ... as such” the classes of patentable subject matter, on the ground that patents are directed towards technical inventions, not commercial methods. Even so, the European Patent Office has managed to interpret the qualifier “as such” in such a narrow way, whereby software patents would be granted so long as they contained an inventive step with a “technical effect”.
In order to give some meaning to the limitation in the European Patent Convention on the grant of patents for computer programs “as such”, the Patent Office judged that the technical effect must be additional to the technical interaction that software will inevitably have when run on a computer, but this still does not set the bar particularly high. For example, a program that improves the efficiency of a certain operation or makes it easier to use would have a sufficient technical effect to be protected.
It is only the most useless of programs that would not have such an effect – and who would want to patent such programs anyway? In consequence, software patents have been granted almost as readily in Europe as in the United States and Australia.
The Computer Implemented-Inventions Directive
Why, then, the big fuss about software patents in Europe over the last few years? Because not all member countries are prepared to enforce the patents that the European Patent Office has granted, or at least there are inconsistencies in the extent to which various countries will do so; for example, England will enforce them, but Germany will not.
To introduce some uniformity in the treatment of patents thoughout Europe, a directive was introduced into the European Parliament intended to codify the existing practice of the European Patent Office, titled Patentability of Computer-Implemented Inventions. There have in fact been three versions of this directive put forward since 2002, and as at the date of writing none of them has been finally accepted.
To understand the differences between these and how it is that different versions can be volleyed back and forth, it is necessary to understand a bit about the quite complicated political structure of the EU.
The procedure by which the patent directive is being developed is called “co-decision”, and it requires agreement to be reached between the European Parliament and the Council of the European Union. The European Parliament is a little like the General Assembly of the United Nations, except that it has more representatives from the more populous countries, and they are directly elected by their constituents. The Council of the European Union on the other hand contains the actual politicians of the member countries from their national parliaments.
To make things more complicated, there is also a third body, the European Commission, which is like a public service, as it is not elected but appointed by the Council with approval of the Parliament. The first revision of the patent directive presented on 20 February 2002 came from this European Commission.
This first revision would have essentially codified the existing practice of the European Patent Office in applying the European Patent Convention, which would have had the effect of legalising software patents throughout the EU.
The second version of the patent directive was presented by the European Parliament on 24 September 2003. It attempted to reign in the practice of the Patent Office by requiring that computer-related inventions have a physical form.
The third and current revision of the directive came from the European Council on 18 May 2004. It was very similar to the original version prepared by the Commission, and again would have legalised software patents throughout Europe.
On 28 February 2005, the European Parliament requested the European Commission to restart the legislative process, allowing for an entirely new directive to be drafted. The Commission refused, and sent the Council's directive back to the European Parliament for a second reading.
The Parliament may now accept, amend or reject the directive, but since any motion to reject or amend requires an absolute majority, the default is to accept the directive. If it passed, all member countries would be required to pass national legislation to give effect to it.
Work within the system
The solution for large and well-resourced developers is to run their application past a good patent attorney or lawyer who will determine exactly what patents would be infringed by the software, and to seek to license the use of those patents from their owners, hoping always that the number of patents required to be licensed and the cost of doing so will not make it unviable to take the new application to market.
For open source developers, searching the patent database can be daunting. Instead, consider asking on relevant mailing lists and newsgroups if anyone knows of patents that your project might encroach upon, and read as much literature in the field as you can to see if any patents are mentioned. The licence agreements and documentation for similar proprietary products are also likely to mention any patents in the field.
If you do find an existing patent that you think your project may infringe, your first response should be to attempt to work around the patent to accomplish the same outcome by non-infringing means, as the GNU project did when creating gzip as an alternative to compress. If that is not possible or if you don't want to risk it, you can instead contact the patent owner and see if they will give you a royalty free patent licence that is compatible with your project's licence.
Overthrow the system
There have been many proposals for patent law reform, but they all run into the problem that we live in a world in which international trade requires a degree of uniformity in each country's intellectual property law regime. These trade imperatives are reinforced by international instruments such as the Free Trade Agreement, and in the case of copyright, the Berne Convention.
What this probably means is that the impetus for change will have to come from the United States, not from Australia. And there are no shortage of pundits wishing to be heard on the issue. Jeff Bezos of Amazon.com (ironically, in view of their infamous “1-click” patent) is one of those who has pushed for the reform of the patent system in the United States. He has proposed that the term of US patents should be reduced from 17 years to a 3—5 year term, and that there be a one-month public comment period before a patent is granted. Bezos states, “Bottom line: fewer patents, of higher average quality, with shorter lifetimes. Fewer, better, shorter. A short name might be 'fast patents.'”13
Promote alternative technologies
If we can't beat software patenting in the legal arena, the open source community can certainly go head to head with patented technology in the technical arena. The community should continue to promote non-patent-encumbered standards and formats, such as the Ogg-Vorbis format for audio compression (in place of the patent-encumbered MPEG Layer 3 format), and the Portable Network Graphics (PNG) format for image compression (in place of the GIF format).
The World Wide Web Consortium (W3C) introduced a new policy last year requiring that any technology submitted for acceptance as a standard be released with a royalty-free licence consistent with open source software licensing terms. Unfortunately the Internet Engineering Task Force (IETF) does not yet have an equivalent policy that extends to non-Web-related Internet standards such as routing protocols.
Use patent-unfriendly licences
Leaders of open source software projects should also select open source software licences which contain clauses unfriendly to patent holders seeking to copy or modify the software.
The GNU General Public License provides in clause 7 that if patents or other legal obligations would prevent a person from freely distributing the software, then they may not distribute the software at all.
Licences such as the Open Software Licence go further by requiring contributors to open source projects to grant perpetual royalty-free licences to their patented technologies to all users of the open source software. This in fact introduces a considerable risk to patent holders that they might lose their patent protection if they use their patented technology in an open source project inadvertently.
Slightly differently again, the Mozilla Public License provides in clause 8.2 amongst other things that if a licensee of software brings a patent claim against any developer or contributor of that software, then unless they can agree with the developer or contributor on a reasonable fee for their own past use of the software, their licence to use it will be terminated and their right to damages will be limited.
This is far from a complete solution to the problem, however because it only affects patent holders who are themselves involved in the distribution or modification of the software that is alleged to infringe the patent.
Whether or not you consider that software patents should exist at all, nobody can seriously dispute that they have been granted far too readily to date, particularly in the United States. Although Australia's patent registry does not have a record of granting so many dubious patent rights as its American counterpart, open source developers operate in an international legal environment in which the effect of changes to American and European law reverberates worldwide.
As open source users and developers, our response should be to avoid using patented software wherever possible, and to promote alternative unencumbered technologies. Developers should keep themselves informed about the scope of patented technologies in their field, and ensure that any contributors to their projects waive any patent rights they may have. Finally, we should all be vigilent for moves by Australia to further harmonise its patent laws with those of the United States, and make sure that our voices are heard in other fights such as that taking place in Europe.