John Gray's original comments (19/08/2004)
Paul Hayes' response (26/08/2004)
The following comments are from John Gray, patent attorney.
I am perhaps a little nearer to understanding why we cannot even agree whether the draft Directive is a liberalisation or not. The exclusion of computer programs under European law dates from 1973 and that wording excludes patents for computer programs "only to the extent that the invention relates to [a computer program] as such". The same form of words applies to business methods, mental acts etc. This kind of wording requires interpretation, and it is self-evident that somebody framing such a law in 1973 could have little idea of the range of activities that would come one way or another under software control these days.
In the 1980s, 1990s, etc. the courts and boards of appeal have had to tread a difficult path in an effort to honour the spirit of the software exclusion, without removing many areas of technology, especially in electronics, from the scope of patent protection.
In saying that the draft Directive aims to preserve and clarify the law as it stands, I refer to the interpretation of the statutory wording that has been reached by the courts of the EPC member states and EPO Boards of Appeal.
Evidently Mr Hayes would have wished a different interpretation in the first place, and I can accept any amount of argument about what the law should be. But Mr Hayes' wish as to how the 1973 statute should have been interpreted is only a wish, not the law as it stands. For this reason I am adamant that certain lobbyists are misleading the public and especially the programming community when they claim that the draft Directive will radically liberalise the patent law.
Mr Hayes is concerned that industry is seeking to legalise retrospectively many invalid patents. From where I sit, his wishes would invalidate retrospectively a far greater number of patents granted validly for very technical ideas that happen to use software in their implementation. The EU Commission has balanced these options in its deliberations.
I have no idea why Mr Hayes thinks the aim of the draft directive is harmonisation with the USA. For those who bother to read it, the Commission's original proposal document (22-page PDF) plainly recounts not only the consultations they went through and the various opinions they received, but also the rationale for their decision on the best way forward.
Clearly they state that harmonisation within the EU is a priority, because the EPO and the different member states currently can have different interpretations of the exclusion, some more liberal than others. Equally clearly, the Commission has rejected harmonisation with the USA with regard to the topic of business method patents. Harmonisation generally makes the patent system cheaper and simpler not only for patent owners but also for third parties. This is especially important for SMEs. The draft Directive is expressly designed to avoid any sudden change in the law, while clarifying it and harmonising it across the EU, roughly where it stands today.
So far as I know, Japan also has no wish to follow the liberal approach adopted in USA. For all we know, the US Supreme Court itself may turn back from or clarify its current position. In Europe the draft Directive seeks to formalise the "technical contribution" approach established in the EPO case law specifically to exclude business method patents. Read the results of the Government's consultations on these issues.
It is also unsupportable to claim that the law has drifted to the extreme end of the spectrum. Speaking from the coal face I can assure Mr Hayes that it has not. Far from colluding in this liberalisation, my experience is that the authorities (UK Patent Office, EPO) have taken seriously their duty to uphold the exclusion of computer programmes etc.
Patent applicants have certainly not been pushing at an open door for 20 years, and I suspect that many truly technical innovations have gone without protection as a result, particularly for SMEs whose capacity for pushing at sticking doors is lower.
The UK Patent Office alone has taken a string of cases to hearings and even the Court of Appeal (Merrill Lynch, Fujitsu, Raytheon, Gale), spending a great deal of public money reinforcing the barriers in fields where, in my own humble opinion, they risk serious damage to our national capacity for commercialising innovation in the important era of the "knowledge economy".
Similarly, I have accepted in all my writings that many "bad" patents may be granted by authorities unfamiliar with new technology. It does me no credit if patents get granted that shouldn't, but the system is designed to cope with that, and this is not a phenomenon unique to software. In any case, it is one of the many issues that the Commission, the UK government and no doubt the others too have taken into account before arriving at the draft Directive.
"The patenting of software inventions is entirely without empirical or theoretical justification or of any conceivable benefit to the economy or society..." – Paul Hayes
The empirical justification for patents in technology generally was evident to advanced economies like Venice over 500 years ago. So you might as well start from saying the same about the exclusion of software inventions from patentability. All the explanations why software is so different from anything else need analysing in detail, not taking at face value.
"A pure software invention is an abstraction, easily separable from any true invention that teaches something new about the use of controllable forces of nature." – Paul Hayes
Mr Hayes, nobody wants to patent such "an abstraction", and who but academics would buy software if that were true? Back in the practical world, I assure you the line is not so easy to draw.
My reference to the "nuts and bolts" was intended to convey to a lay audience something of the flavour of the issue, not to represent my entire understanding of computer technology. Mr Hayes' assertion that "memory management algorithms [are] purely abstract concepts unrelated to the physical nature of the hardware" might apply to some memory management algorithms, but only if they are sitting on top of other memory management algorithms that provide an infinite memory space with perfectly uniform performance. This merely reveals Mr Hayes' ignorance of the technical end of the software spectrum, and ignores the reality inside of cache memory systems, virtual memory systems and the like, without which his computer would not be the tool it is today. (Actually I more inclined to believe Mr Hayes is aware of these things, but prefers to ignore them for the sake of his argument.)
The kind of software which Mr Hayes describes, that can be written by a programmer at an abstract level (i.e. without considering the technical limitations of what lies beneath), can be excluded from patentability without too seriously disturbing the status quo.
My advice to clients with such inventions has always been that the law on these is at best unsettled. Where I have real problems with the proposed amendments is in their effect on "software engineering" which does need to take account of the physical limitations of memory, or instruction types or whatever, and where a seemingly trivial re-arrangement of an algorithm can have a radical improvement in performance.
If Mr Hayes is ignorant of this area of programming, he owes a debt of gratitude to those who aren't.
Take a very real, everyday example. If I want to buy a PCI card dial-up modem in a computer shop, I can today buy one where all of the signal processing is performed in hardware, or one where the identical processing is performed to a large extent with the help of driver software running on the host CPU (my PC).
The purchasing decision for the educated customer is based on lower hardware cost versus the impact on the availability of my CPU to do other things while the modem is dialling. Most users, including most programmers I am sure, are completely unaware of the trade-off, and see only the lower price for the "soft" modem. Similarly, the "software radio" is on the horizon.
Software drivers of all kinds nowadays perform digitally functions that would previously have required built-in hardware in a printer, scanner, network card etc., or embedded software within the device itself replaces more costly hardware. The function is not different, it is merely implemented in a different blend of hardware and software as a routine engineering choice.
Now, if the signal processing techniques that allow the modem to operate at 56k rather than 33k, for example, are patented, but the "soft" type of modem does not infringe the patent under the amended Directive, then the "software" industry will have won its battle at the expense of other industries who believe in the benefits of patent protection, for all its faults and limitations. Maybe it would all be for the best, but those industries should be heard and not dismissed out of hand. It is not always boom time in the telecoms hardware industry, and continued R&D is in jeopardy if the traditional incentives are removed!
Mr Hayes may not be aware of these technicalities, but they underpin all the devices he uses in his work. If he is aware of them, it is dishonest to ignore or deny them when debating where to draw the boundary between "pure software" and technical inventions.
It is grossly misleading for Mr Hayes to imply to readers that a combination of copyright and the "sui generis" chip protection ("semiconductor topography" or "maskwork") provide any substitute for patent protection, for signal processing technologies that can be implemented either in dedicated circuits or in software.
I might as well say "You don't need a Fire Station in your neighbourhood, because there's a blanket and a bucket under the stairs." Those types of protection may have their place, but they are both limited to actual copying of the detailed implementation, not to its novel principles.
Further, the semiconductor topography right is useless for digital processing functions that can be laid out in an infinite variety of ways on a generic "programmable logic array" type, where no particular layout is necessary. Given that those types of protection in the UK are both free to obtain, the fact that chip makers continue to spend money on patents proves that they are not a substitute.
"The EPC contains arbitrary exclusions? Perhaps it was drafted by a couple of slightly tipsy legal experts one afternoon in a pub but I was under the impression that the exclusions were very carefully considered and deliberately placed there for very good reasons." – Paul Hayes
I bet Mr Hayes never quite bothered to look up what those reasons might be. Perhaps the dominant computer hardware manufacturers in the 1960s simply couldn't stand the thought of being beholden to a mere programmer with a patent in his hand?
"We are the people who realise how extraordinary and anomalous it is that 99% of the owners of personal computers stare at a screen mostly filled with the patterns produced by the software products of just one enormous company." – Paul Hayes
Yes indeed, but this monopoly arose without patents. Microsoft is a Johnny-come-lately into the world of patents, albeit a big one. With your retrospective crystal ball, can you prove to me that that this global monopoly could have arisen if, for example, Apple/Xerox, WordPerfect or Lotus had thought to obtain a few key patents in the field of GUIs, word processing or spreadsheets? Little-guy Stac's disk compression patent forced Microsoft to bin a whole edition of MS-DOS in the olden days, so don't tell me it can't be done.
"The EPO has since 1986 granted more than 30,000 of these pure software patents, including many covering business methods and trivial and basic functionality." – Paul Hayes
There is no shortage of cases being rejected by the EPO and UKPO on the grounds of being software or a business method. The latest state of the EPO law is defined by the "Pension Benefits" case which was rejected. IBM had appeal cases in document processing rejected years ago. The cases rejected by the Boards of Appeal are only the tips of the icebergs. Not only that, but currently any application that even has a whiff of software or business method about it is being held up without the EPO even performing a search, as a matter of deliberate EPO policy. This affects especially applications from the USA, because of the way they tend to be drafted. Harmonisation with the US is simply not on the cards.
What of the Open Source community?
The achievements of Linux and Open Source are great, but it should be possible to deal with the existence of patents within that, rather than throw the option of patents out of the window. For one thing, participation in open source remains voluntary, so far as I know, and it would surely be too oppressive to deny non-participants the established licensing models.
Secondly, it is not that long since your average open source advocate thought that open source was an alternative to copyright laws, and did not realise that an open source licence is in fact an application of copyright. The friction here comes because patents have a different nature, I accept. I also say open source advocates should not automatically have the right to dismantle the existing forms of protection across a whole load of other industries.
"This is an insult to [...] programmers" – Paul Hayes
Let me assure programmers that their views are not disrespected. If I did not care about their view I would not be sticking my head above the parapet to correct misinformation and give a flavour of an alternative view.
John Gray is a patent attorney with Glasgow-based patent and trade mark agents Fitzpatricks.