JILT 2000 (1) - Philip Leith
Software Utility Models and SMEs
Dr Philip Leith
The European Commission and European Parliament are keen to extend the use of IP rights in smaller enterprises. Patents have been seen to be too complicated and too expensive for these firms and one suggestion - which has been taken up by the EC and Parliament - has been that utility models should be made available. A Directive has thus been proposed to enable these. The Directive in its original form did not include protection for software but this was added on the suggestion of the European Parliament.
The whole question of patent and associated rights is currently under discussion in Europe and certainly merits full discussion. In this paper, however, I concentrate upon the notion that utility model protection for software is an appropriate tool for SMEs and suggest that it is not.
Keywords: Utility model protection, software patents, innovation, SMEs.
This is a Refereed Article published on 30 June 2000.
Citation: Leith P, 'Utility Models and SMEs', 2000 (2) The Journal of Information, Law and Technology (JILT). <http://elj.warwick.ac.uk/jilt/00-2/leith.html>. New citation as at 1/1/04: <http://www2.warwick.ac.uk/fac/soc/law/elj/jilt/2000_2/leith1/>
The European Commission has recently published its amended proposal for the harmonisation of utility model protection in Europe. The amendments which were proposed by the European Parliament suggested strengthening and extending the protection which the Commission had originally proposed. As the explanatory memorandum to the revised proposal makes clear, 'Parliament did not question the Commission's approach and the main features of the utility model as described in the original proposal were retained'.
Importantly, Parliament proposed extending the scope of the Directive to cover software. This was accepted by the Commission and appears in the new proposal.
There has been much discussion over the past few years in the patent world concerning the value and construction of 'second tier protection' since it became clear that the Commission were interested in this and were promising a Green Paper on the topic. The worry that proposals from the Max-Planck Institut required a UK counter led to the CIPA's conference at Brocket Hall in 1994.
The worry from many in UK industry and those concerned with patents, was that the value of patents could be undermined by the free-flow of unexamined utility models. The image of a 'minefield' was frequently cited as a potential result of having easy access to protection, where these protections were hidden from view until they exploded in litigation under the unsuspecting manufacturer. Such protection, rather than helping Europe to revitalise innovation and development, could act as a barrier to innovation since firms would not wish to expend money and develop in fields where the protections were unclear.
However, what was not envisaged at that time was that protection for software would also be part of the utility model debate. The debate which did occur focussed not at all on program protection, because at that point in time it did not appear that the exclusion from patentability arising from EPC 52(2c) could be effectively removed without the substantial effort of reworking the European Patent Convention. As we now know, the combined effects of proposed Directives and EPO Technical Board of Appeal decisions, have shown that such a Diplomatic Conference to revise the EPC was not necessary.
It seems likely to this author that most difficulty with utility models will arise in the field of software. The combination of low requirements as to inventive step, non-examination combined with the nature of software itself will create problems which may undermine the idea of a utility model and certainly cause problems for innovation in the software industry, if not all industries where computer control is used. Legal theory has shown that lower level tribunals can have a 'polluting effect' on higher level procedures and legal concepts, rather than simply being peripheral or subservient to the higher level legal systems. We should not therefore be too surprised if utility models have significant effect upon patent law and procedure.
This is not to say that protection for programs is not 'a good thing' - certainly if inflatable canoes can be protected, then why not something which is becoming central to economic development? But, as we are seeing, programs are complicated entities which differ substantially from inflatable canoes. We should be careful that protection offered to these different categories is properly capable of offering real protection and enabling innovation in both fields, before encouraging SMEs to participate in its use.
Protection under the Directive is towards inventions which are 'industrial'. Rather than the protection being identical to that offered under the EPC, there is an exclusion of protection for chemical substances or processes. A simple outline of the proposal indicates:
In some ways, the general approach being taken is closer to that which was being requested by UK proponents - particularly with regard to level of inventive step. Neither side in the debate at the time seemed to imagine a nationally based system, given the Commission's single-market approach but the problems which the Commission has been having with cost of translation for the Community Patent may well have undermined any attempts to produce a Community-wide system. The Directive obviously hopes to overcome potential problems for the internal market by imposing internal Community exhaustion of rights.
It is clear that there is a substantial goodwill towards the idea of utility model protection, though this is not total. Figures collected over 1994/5 demonstrate the level of this goodwill (for European-wide rather than national-based systems):
Figure 1: Figures collected over 1994/5 demonstrating the level of goodwill for utility model protection (Leith, 119, 1998).
There is certainly a suggestion in the UK that support for utility model protection is over-stated and it has been pointed out to me twice by readers of a draft of this paper that at an Intellectual Property Institute conference in London in 1999 the IP Director of Siemens stated that Gebrauchsmusters 'were not worth the paper they were written on'. This may be so, but there are certainly reasons apart from simple protection which make other German attorneys highly positive about them. It does appear that - as with the thrust of this paper - it is those German attorneys who are most closely associated with potential software protection who are particularly concerned about utility models.
Those countries which have it, do find it useful to a greater or lesser extent and it has been those countries, such as the UK, who do not presently have it who have been viewing them most tentatively. As one German attorney has put it:
'It is a question of experience. Those with no experience think that [dealing with utility models] is difficult and everyone who has experienced it thinks it's wonderful. We have lived with the system for many years and have not found any difficulties, even with infringement cases based on utility model registrations'.
The quantitative research carried out during 1994/5 and detailed in the above diagram indicated that around half of the UK respondents felt that there would be little change to protection from utility models, and some 30% felt it would provide additional protection whilst less than 15% thought it would reduce protection. Thus, from the perspective of the European patent attorney, utility models do not seem to be particularly problematical. Marginally more German attorneys felt that protection would change and there would be additional protection for clients. Of course, the question asked in that survey was related to European-wide utility models which are not being offered by this Directive.
The most suggested advantages of utility model protection as against patent protection are that:
It is clear that the European Parliament were agreeable to such SME-friendly arguments, since they suggested that fees should be reduced by 50% for small and medium-sized firms, individual inventors and universities.
However, for those involved in litigation there are also strategic advantages which arise from this form of protection - being able to claim infringement at an early stage prior to patent grant, splitting off utility model applications from patent applications for quicker protection etc... As both a defensive and offensive tool, therefore, there is much to recommend utility model protection. Outside of litigation, there are other reasons for considering utility model protection. In Ireland, one advantage has been that tax relief has been available on research at an earlier stage and at a lower inventive level than that with the patent. And, in a commercial patent framework where patent licenses are sometimes exchanged and traded by size of document pile rather than value of the claims, there is an advantage in having more pieces of paper rather than fewer.
Thus, despite the concentration of debate upon the usefulness of the utility model to the SME, it is unlikely that large businesses will not see advantage in using these devices for the same tactical reasons that smaller enterprises will. Indeed, there is no evidence that in those countries where utility model protection has been available, that SMEs have been their primary users. It therefore does not seem possible that the new Directive will encourage a situation where SME usage will be enhanced and non-SME usage limited: we should expect larger enterprises to make more use of utility model protection in the new environment.
It has been suggested that one reason for the limited use of the utility model protection in Germany, for example, by US and Japanese firms has been the cost of translation. Those non-European firms who have been filing European applications have been doing so in English, and have had no inexpensive route towards Gebrauchsmuster protection. When the Directive is implemented, and the UK is required to provide protection, it will be a relatively simple matter for these firms who file via the EPO and PCT routes in English to prepare an English-language utility model at low cost and file this as a UK Utility model. Trans-national filing is, of course, usually carried out by larger firms, so a benefit will flow to them.
Another example was given to me by a patent attorney for a major computer manufacturer. He suggested that his company produced many more inventions than they would typically wish to patent (due to expense) but were always concerned that if these ideas did not reach the public domain in some way, then they might be protected by the competition. One current method of ensuring that unwanted ideas become part of the prior art is to publish in the journal Research Disclosure. For a modest fee, companies can put their ideas into the public domain, free from future protection. This attorney suggested that utility model protection may be as effective and cheap a method of stopping competitors from protecting ideas, but also provide extra protection to surround his patent portfolio. Such an option is particularly attractive to the larger company with a large body of patents - rather than to the SME.
It is not just that the advantages which accrue to the SME can be used by larger firms, but that to the SME utility models can be dangerous devices. In Germany the DPA has suggested that while the Gebrauchsmuster system is successful and has increased in application and grant number, that the unexamined nature means that holders must exercise care in their use:
'In competition, claiming this right may trigger off claims for damages against the utility model owner by the alleged infringer. The utility model owner must be extremely careful, and is personally responsible for taking into account and appreciating the relevant state of the art as well as estimating inventiveness'.
The proposed Directive (Art 26) locates the legal framework for utility model, insofar as it is not dealt with in the Directive, as being that of national patent legislation. This means that remedies for aggrieved parties who have been recipients of groundless threats of proceedings in patent matters will also be available for those receiving such threats of proceedings in utility model matters. Since it is well understood that - in the absence of professional advice - SMEs have particular problems in reading claims and assessing the value of prior art, a simple system which may encourage filing and accusations of infringement against competitors by SMEs themselves, may produce negative effects upon business health.
That there are difficulties with utility models is clear: they are not solely SME-friendly, nor without danger. However they are being granted in ever increasing numbers and their dangers may sometimes be overstated. They do offer a cheap and cheerful complement to full-term patenting which has strategic, cost and convenience factors. Given a situation where reasonably well-known technical fields are being protected, there is no reason to believe that anarchy will follow implementation of utility model protection in the UK.
4. Why are Software Utility Models problematical?
The situation becomes different when we consider protection of software under the proposed Directive. There is no long-term development of the law of patent protection for software, and there is little understanding by those in industry of what protection might mean in practice apart from a small number of expert players (it is commonly held that protection for software has been available, but that the form of claims has required drafting skills which are not generally available).
There is still the potential for considerable debate over the results of the two Technical Board of Appeal decisions T935/97 and T1173/97 and neither of these legal advances have been further tested in courts. This means that we have an area of law which is undergoing substantial development and discussion and can certainly not be described as 'clear': yet, the Directive on utility models is opening this area to usage by SMEs, and doing so without the support which would arise from proper examination of claims and prior art.
In essence, it appears to this author that the problems which have generally been highlighted as problematical in the granting of software patents will be more so in the granting of utility model protection. These are:
Though some have suggested that the major patenting problem with software is that it is a symbolic system which due to its nature as instruction is better protected by copyright, this is not completely true. Software certainly has a dual nature - as an instruction, but also - in the mind of the programmer - a 'physical' reality. Programmers clearly program by visualising objects and processes operating on these object (which explains why object oriented programming has been seen as such a successful programming methodology) and it should be possible to examine inventions by recourse to these visualised ideas. For example, something like a SELECT statement in a programming language is obviously simply a sequence of higher level JUMP statements, but has a reality which comes from visualising a kind of 'drop through' filter. Expressed in terms of the higher level abstraction, it does not seem difficult to decide novelty and inventive step over the lower and more confused programming instructions using multiple JUMP/GOTO statements. And, if patenting had been available for the first multi-tasking software, then we could equally see novelty and inventive step in the allocation of resources such as 'round robin' etc..
From one perspective, then, it should be possible to consider patentabilty in relatively simple terms. In fact, there are problems with this, as suggested below, but these do not come from the symbolic nature of software, but rather from the programmer's description of the programmed objects.
Setting levels of patent grant ('inventive step') is one of the black arts of the field, and one which no matter how objective the attempt (through, for example, tests of problem and solution) is always difficult to decide. The EPO has managed to impose upon European patenting a relatively harmonised level of inventive step. It has done so by force of application numbers as well as through significant training and internal harmonising procedures. No doubt, after a period of time, levels of inventive step in software will be harmonised and will operate within relatively narrow bounds. We are not, though, at that stage and there must be some worry that the lower levels of inventive step which might be expected through utility model protection must be unclear.
After all, what is the notion of 'not very obvious' to mean in software terms since it is not yet clear what is to be protected as 'not obvious'.
For those who do not have sufficient experience of the patent system (in particular SMEs) it may be that what will be seen as 'not very obvious' (and also worth protecting) are those parts of the program which the copyright system has failed to protect. For example, the 'look and feel' of programs. Certainly, a new program's look and feel can be 'not very obvious' and can have significant advantages - both to smaller firms who wish to protect their ideas, as well as larger companies wishing to develop a brand image and common interface. It seems likely that there will be a host of utility model applications (untested and unexamined) requesting protection which industry has shown it is interested in gaining, but which has not been provided through the copyright system. Such a volume of untested utility models must cause extra costs for SMEs - in terms of applying for protection which is worthless, and also considering the merit of competitor's claims.
The problem of searching prior art is the most widely acknowledged one with regard to software patents. Programs are by their very nature obscure and difficult to penetrate, for example:
Patent offices are currently arguing that they do have the competence to carry out such searching and examination, but this is difficult to accept. Patent offices acknowledge that depth and quantity of documentation are necessary for good search and examination, and their Annual Reports usually carry indications of this depth and quantity. Yet there can be little available to them in the necessary forms of program code (and indexing of this code) and program description to make the task easy.
No doubt this documentation will be built up, but it will require significant resources and also significant input from those involved in software patenting (in, for example, opposition procedures) before the documentation is sufficient and reliable. We will thus have to live with a period when bad patents will be awarded due to limited prior art being available and being accessible. It seems a dangerous tactic to bring utility model protection into a context where there is so much unreliability.
4.3 Discovering and proving infringement
For all those considering using the new European software protection regime, there must be concern about gathering of evidence and ensuring that infringement is not occurring. In most fields the techniques are well developed - in engineering the machine can be stripped to its fundamentals and in chemistry the substance can be analysed. This is not the case with computer software.
The 1991 Directive on the Legal Protection of Computer Programs provided software builders with an extra right which was not open to other fields - that of the right to prevent decompilation except under specific circumstances and towards specific goals. In short, that analysis of a program structure and operation could only be fully undertaken (i.e. through decompilation) when it was necessary to obtain information for interoperability.
This extra aspect to the copyright regime was controversial since it extended rights for software in a way which was not extended for books or other publications. Thus a reader can analyse a text to enable a better competing text to be prepared, but cannot do so to prepare a more efficient or better program. The extra right is even more controversial when the patent system is taken into account: the infringer who utilises protected inventions within a piece of software may be enabled to hide that infringing use, but particularly when the invention is not related to the outward appearance or obvious functionality of the program. A litigant who suspects infringement may himself have to carry out copyright infringement in order to determine whether this has actually taken place or not.
The route around this evidence collecting problem is, of course, to go outside Europe and to decompile in countries which do not have this reverse engineering right. This is a solution which is particularly suitable for larger enterprises, and less suitable for SMEs.
There is already some lack of confidence in the matters of evidence handling in patent examination. Patent offices, of course, have no method of carrying out independent tests and must rely to a great extent on the information supplied by patentees and opponents. However, examiners do have a substantial background in their technology - the requirement that they have a scientific or engineering degree is the most obvious one. This means that they can bring expertise to bear on the applicant's claims, which is based in scientific training and methodology. We might say that the situation in testing evidence could be very much worse than critics suggest than it is at present, because of the generally high standard of patent examination in the major patent offices.
In the analysis of software patents, though, there is no clear expertise available. Those who have expertise in programming methodologies such as operating systems, program design etc. will not have usually entered the patent examination profession (computer science, anyhow, has problems with method and philosophy which makes it different from other sciences). There will be an 'expertise deficit' in this area until examiners are properly trained in the various elements of software patentability which will make the likelihood of incorrect assessment of patentability much higher than in other areas. And expertise and experience are important in the process of examination in order that, for example, overly broad claims are properly noted. As one patent attorney told me:
'when preparing a case you have to convince the examiner of the merits of the invention. And that is not a point of technical ability but of psychology'.
So far, the expertise in these areas lies mainly with a small number of examiners in the EPO and one of the Technical Boards of Appeal. If a flood of software applications appears in the next few years (as is likely) there will be substantial problems in ensuring that examination standard is harmonised across the software field. Adding to this flood with national floods of software utility model requests for search reports is likely to either produce problems in harmonisation or increase the delays in examination, opposition and appeal which are already complained about vociferously by the patent profession.
There is a final and major problem which relates to the point about what is being protected. If it is 'software on a disk', it is not the larger technical solution to a problem, but something intrinsic to the software itself. This means that the invention must lie, not in the 0s and 1s of the object code, or in the higher level statements of the source code, but in the description of the action which is being carried out by the program. This is the obvious thing to protect and is really the only thing which can be protected. The nature of the programming process leads to difficulties in attempting to protect anything else - particularly trying to protect invention at the level of moving data into and out of registers and accumulators.
The problem is that programs can frequently be described in one way, but function in another. The classic example of this kind of problem is to be found in the field of Artificial Intelligence - much academic work was funded in the 1980s which would have led to a significant number of patent applications had protection then been available. There was much debate over the success of the claims which were being made for these AI programs, a debate which was not helped by the lack of clarity which flows from programming. One early example of the debate concerned that of a program, called AM, which made mathematical 'discoveries' and which was unusual in that its critics had undertaken a detailed analysis of the program, matching the published descriptions of that program against the coding. The critics considered that the description did not match the coding, and that discoveries were not being made. Most programs never see this level of analysis, primarily due to cost and time factors, and the programmer's descriptions have to be taken as read. Particularly in patenting, this must lead to difficulties in considering whether infringement has occurred.
The field of AI is more extreme, perhaps, than most of the examples which we might expect to be arrive in software patent or utility model applications. But many software patent applications will be of this nature. Applications in recent years have been to do with architectures for neural networks which systems have, in my view, not proven in practice to substantiate the claims made.
This problem of code vs description does indicate that there is something different in the nature of a program to that of a technical invention. My own belief is that proposing 'intelligence' in a computer program is rather like proposing perpetual motion in mechanics - despite there being a considerable number of respected academics who would argue that 'intelligence' can indeed be located in programs.
The difference in nature between program and other technical device is often to do with the more complex context in which computing is placed: much programming research work is done by cognitive psychologists, for example, and is related to philosophical debates about the nature of information and reasoning. This is not a context in which the inflatable canoe usually appears.
The Directive has obviously grown from the current Commission and European Parliament philosophy which suggests that more protection is better. The pro-protection philosophy may well be correct, but there is little evidence that more protection necessarily must translate into more pronounced European innovation. The inception of the EPC has increased the European protection available for US and Japanese industries by producing a patent granting system which enables easy grant throughout Europe. With around 50% of patents being awarded to non-European sources, one might suggest that more and easier protection may simply retard European innovations, particularly in the SME field, and also particularly in the field of software which is currently driven by US research and development.
The decision to award software patents has been made and will be impossible to hold back. This may well be for the best, and we should therefore work towards creating a sound software patenting philosophy which ensures that invention is rewarded and that 'bad software patents' do not become a factor in inhibiting development and research.
It is not so clear that we should accept that such a software patenting philosophy which is robust enough, or has developed a community of practitioners and examiners who have the ability to ensure that the balance between rights owner and non-owner, is already here. The complaints in the US indicate that there are difficulties in such patents, and we must expect that the difficulties will flow here, too. Given this, it is a serious mistake to consider that our ideas of software invention are at such an advanced state that we should consider allowing utility model protection, and encouraging SMEs in software production to look for protection to this kind of device.
Chemistry has been excepted from those areas receiving utility model protection. There are substantial arguments that software should also be excepted, at least until the working of the system is carried out in the required review.
1. COM (1999) 309, 12.07.1999.
2. CIPA, Second Tier Protection: report and proceedings of a symposium, 6-8th July, 1994.
3. Which includes agriculture - Art 7(1).
4. These industries were unhappy with the idea of utility models. See Recital 13.
5. These - Art 3(1a-1d) - are excepted 'as such' by Art3(2).
6. These are non-protectable by Art 4.
7. These are not considered of industrial application under Art 7.
8. See a 1995 view of the rival proposals in M. Llewelyn, 'Proposals for the Introduction of a Community Utility Model System: a UK Perspective', Web JCLI  2
9. Art 21.
10. 1998, P. Leith, Harmonisation of Intellectual Property in Europe: a case study of patent procedure, p120, Sweet and Maxwell.
11. Note 10.
12. Though it is not clear how this idea would have been policed. It would not be difficult for large enterprises to set up smaller, independent companies to control their utility model portfolios. Whether Universities - already funded significantly by the public purse - should be further subsidised is a further question. This proposal was not accepted by the Commission.
13 . Where it is titled, the 'Short-term patent'.
14. The German Patent Office now allows filing of non-German patent and utility model applications, but translations must be provided within 3 months.
15. Research Disclosure, Emsworth Design, Inc. New York.
16. DPA, Annual Report, 1995, p32.
17. Patents Act 1977, s70.
18. EPO Examination Guidelines, Part C, Chapter IV.
19. I shall discuss this in a project report, being prepared, on the Boards of Appeal of the EPO. It is, of course, frequently noted as a problem with the EPO.
20. Ritchie G.D. & Hanna F.K.,  'AM: A case study in AI methodology', in Artificial Intelligence , Vol 23, No 3. This is a particularly interesting case since it is one of the few examples in the literature where a program's code was discussed in any detail by anyone apart from the author of the program.
21. P. Leith, 1990, Formalism in AI and Computer Science, Ellis Horwood.
22. Art 28.