Clarification of European Law on Patentability of Software

In the Autumn of 2003, the European Parliament drastically amended a proposal, prepared by JURI (the European Parliament's Committee for Legal Affairs and the Internal Market), which purported to clarify and harmonise the status of software in relation to patents. The democratically elected Parliament's amendments actually made it achieve that goal, which it singularly failed to do before until amended. The oligarchically appointed European Commission promptly withdrew the proposal and has now (Spring 2004) resubmitted it without the Parliament's amendments.

I have worked as a programmer in the European software industry for over 20 years and it has always been clear to me that European patent law precludes the granting of patents on software. As the present crisis has loomed, it has become clear to me that there are two areas of unclarity in the law:

• The law was phrased to preclude patenting of software as such, a form of words which allows that a patent might involve software in some sense, yet still be valid.
• The law's limited clarification of the as such wording is expressed in terms of allowing the patenting of a technical effect notwithstanding its implementation in software.

The first and simplest thing that a clarification of this situation should do, and do explicitly, is to state that a computer program, written to run under a general-purpose computer operating system on general-purpose computer hardware, is in fact software as such and so does not, under any circumstances, infringe any patent. This would protect the European software industry from the fate which we currently see engulfing our peers in the U.S.A.: when they write software, they can have no assurance at all that they will be able to bring that software to market and recover the cost of its development by the honest process of trade.

Now the existing law contains these technical effect and as such qualifications, which surely cannot have been included without the authors of the law intending that they imply some cases where an invention involving software may lawfully be patented. Let me give, for the sake of illustration, an example of a case where I think all can agree such an exemption might apply.

In 2003, I heard of an invention by NASA (I do not know whether they have patented it) comprising an aeroplane with flexible wings (like the Wright brothers' original aeroplane, as it happens) combined with a control system (making extensive use of software) which enables the aeroplane to operate more efficiently than conventional aeroplanes (with rigid wings and ailerons) yet to fly with that degree of stability for want of which all aeroplanes since the Wright Flyer have had rigid wings.

It is my understanding of the exemptions above that they (were intended to) relate to cases of this kind: where an invention intimately involves software - yet involves substantial inventiveness aside from the software - its involvement of software should not render the invention inelligible for a patent. To grant a patent on such an invention, without allowing it to become a patent on software, it is necessary to make clear - as the law presently does not - that the resulting patent should cover the hardware - wing and control system - but not the software.

Thus if I were to write a flight-simulator program in which a flexible-wing aeroplane is modelled, the software my simulator would include to control the (virtual) wing would not be covered by the patent, even if it was amenable to separation from my simulator and use by an aircraft designer as the control system for a real physical flexible-wing aeroplane. Of course, that designer's work would be covered by the patent, which would come into play at the point when my software is deployed as the control system of the flexible-winged aircraft.

It should, in particular, be noted that an invention should not be deemed to warrant a patent on the above grounds unless it genuinely makes some substantial contribution to the technical advancement of some (non-software) field. It must actually have warranted a patent for the effects it actually produces in the (non-software) field of endeavour to which it is a contribution. Thus, if some process (such as selling goods to one party but delivering them to another) has been successfully implemented without the aid of computer programs (as, for instance, by InterFlora for many years) prior to the filing of a patent on a mechanism for doing the same thing with the aid of a computer, that patent should not be granted.

As it happens, the EPO has granted Amazon exactly the patent described in the InterFlora illustration I just used. It has also granted a great many other patents on software, many of them scarcely more reasonable. Patent lawyers have stretched the tiny amount of unclarity in the existing law beyond all possible credible bounds in order to justify such flagrant breaches of the law - which serve no-one but patent lawyers. The EPO and its friends are now trying to get these unlawful actions legitimised by having the law changed.

If it is the Commission's intent to clarify the patentability of software, and if their intent is really to serve the best interests of the European software industry, the Commission will draft a proposal which encodes and fleshes out the above sketch of the difference between software and patentable inventions only rendered feasible by the use of software.

Now various parties have asked, since the present law exempts software from patents, why should software be different ? and, in closing, it seems fair to address that question.

• First, I note that the reason generally given for providing patents is that they reward inventors, thereby stimulating innovation. In the software industry, there is no lack of innovation - indeed, our customers might with good cause beg us to slow the flood of innovation, which swamps the users of our products in permanent upheaval. Stimulating innovation in the software industry is unnecessary.
• Second, it suffices to examine the patents that the European patent office has been granting - such as the Amazon patent I allude to above, which would appear to cover a business method InterFlora have been using for decades - to see that the patent office is incapable of assessing whether a claimed innovation is indeed novel and non-obvious to practitioners in the relevant field of endeavour: the industry has no faith in the patent examiners' ability to assess these issues (and this applies to every patent office which has granted patents on software, some of which have been doing so for years - it is not just that the EPO's examiners are inexperienced). The essential cause of this is that the software industry is innovating so fast that it is impossible for an examiner to be well appraised of the state of the art - it is, indeed, hard even for full-time practitioners to keep up. Regardless of the reason, the inability of patent examiners to win the confidence of the industry is a material difference between software and those industries which are well-served by the patent system.

The software industry, without any interference from the patent system, innovates at a break-neck pace: this makes patents both unnecessary and unworkable in the "software as such" industry.

I've said it before: to [ EPO | UKPO ]
I'm not alone: [ FFII | EuroLinux | CDR ]
FFII disects McCarthy: [ MEP reply | factsheet ]