--> [ News | Government Letter | Minutes | Reversal | Urgent Appeal | Unqualified Majority | NL & Swpat | Parliament ]

Freedom to innovate

Reaction to the letter of State Secretary Van Gennip of 16 June 2004 regarding software patents

Vrijschrift

and

FFII - The Netherlands

22 June 2004

On 16 June, State Secretary Van Gennip of Economic Affairs sent a letter regarding software patents to the Parliament. Our conclusions about this letter, which are explained in more detail in this letter, are:

1. The text of the Council does not protect against an avalanche of trivial patents
2. The text of the Council provides for patenting of pure software, independent of any specific device
3. Unfortunately, we cannot but notice that the letter of the State Secretary contains much false information and contradictions
4. Problems of the Council proposal, incorrect allegations about the EP
5. Of the four goals, three are provably not reached
6. Treaties are interpreted in a tendentious way
7. The letter is also interesting for what it does not say
8. There are more remarks to be made about the letter
9. There is a general misunderstanding regarding software patents

1. No protection from trivial patents

Contrary to what is noted several times by the State Secretary, the text does not protect from an avalanche of trivial patents.

The Council does not define "technical" anywhere, causing "the technical aspects of the invention" to mean as much as "the aspects of the invention for which somebody could want a patent monopoly". Johanna Boogerd noted during the 3rd June debate that a strong definition of "technical" is crucial, but there is none.

The only objective criterium in the text of the Council is "novelty". All other criteria are undefined and therefore subjective.

The (generally known) problem of "trivial patents" is not solved by requiring a "technical contribution". Something technical (regardless of how "technical" is interpreted or defined) can be perfectly trivial. The problem is more that the threshold for the so-called "height of inventiveness" is traditionally (too) low in patent law. Solving this problem falls outside the scope of this directive. (page 3, number 3.)

The law quite clearly states that inventions must "not be obvious to a person versed in the arts". If one takes those words literally, then this is a very low threshold: after all, the "person versed in the arts" does not mean anything -of course the reference point is not a layman-, and "obvious" also does really mean obvious! The following reasoning is also heard from time to time: if an invention is new (another requirement for patentability), this means it is consequently not "obvious", since otherwise it would have been invented already! Patent law is only a hair away from allowing one to say that it very clearly and literally allows trivial patents! This once again illustrates that this directive does not change anything regarding this acknowledge weak point of the patent system.

2. Patenting pure software is allowed

The State secretary notes on page 9, first sentece:

• "Additionally, it is once again confirmed that software on its own,
  • independent from an application in a specific device, is unpatentable."

Article 4a(1) excludes computer programs as such. This sounds encouraging, but isn't. There is no clear definition of what "computer programs as such" are. According to article 2, programs executed by computers are patentable inventions. Martijn van Dam noted this as well during the debate on 3 June.

As far as patent claims are concerned, the Council even goes further. After all, article 5.2 allows for claims on literally "computer programs on their own", if these computer programs are patentable when executed by a computer.

Finally, during the debate the State Secretary gave examples of techniques which should be patentable according to her: the (electronic) shopping cart and electronic payment. These techniques are not part of the physical operation of a specific device, but are executed on normal computers. As such, this is about generic software.

3. Incorrect information

Unfortunately, we cannot but notice that the letter of the State Secretary contains much false information and contradictions.

In the previous point we already saw that the State Secretary claims that the directive does not allow for pure software patents, while in fact it does. And since this was already mentioned during the meeting of 3 June, and since the State Secretary appears to be aware of this error given the examples she gave, we find this "error in the word processing" very objectionable. It is plainly obstinate to fall back on previous, incorrect assertions.

The State Secretary states regarding business methods (article 2b of the Council):

• "As a consequence, business methods, that are non-technical in nature
  • and are just implemented using a computer, are not patentable."

Since technical is not defined anywhere, they can be perfectly quite technical. According to the EPO, they are technical if they e.g. provide huge scale benefits.

Another possibility is using the backdoor which was explained by Ms Van Gennip herself, regarding the double-click patent of Microsoft: if you state that because of your business method less mouse clicks are required, then this method is suddenly transformed into a technical invention according to the Council, because you make use of a mouse and supposedly improve something to it.

Regarding article 5.2 of the Council:

• "However, the Council text says that that program on that medium is only
  • part of the patented matter, if it is intended to be loaded into the invention and exercise its function in there."

Using the TV example: suppose you have a program that processes a video signal inside a TV and that this processing is patentable, then article 5.2 allows that one can at the same time also obtain a patent on the computer program on its own (without any context at all).

This article also causes the functionality of the program to be monopolised as an object instead of as a process, which causes not only the use, but also the publication of such programs to become an infringement (just like a patent on a hammer not only covers the use, but also the selling/distributing of such hammers). In this case e.g. an internet provider could be sued for infringement if one of his customers publishes a program which uses the patented technique.

Finally, the "context" also has very little limiting significance. Nobody will ever ask for a software patent on a method to improve the image quality of a TV, but to improve the quality of "a video displaying device". In other words, it is trivial for a hardened patent attorney to word a software patent always in such a way that all useful contexts/applications are covered.

Further, the text claims:

• "Patent law that is generally in force however, considers three
  • requirements for patentability."

This is wrong, there are four requirements: one must 1) have an invention which 2) is new, 3) contains an inventive step and 4) which is industrially applicable. Not everything can be an invention, such as findings concerning mathematics and computer programs (as stated in article 52 of the European Patent Convention).

Just like the Council does not see the "technical contribution" as an additional patent requirement, but as a more specific definition on how to assess the inventive step, the EP considers the condition of "technicality" to be a more specific definition of assessing whether or not something is patentable subject matter (all things that are excluded from being an invention in article 52 of the EPC, turn out to be not technical according to the EP's definition; this definition comes from a.o. the German jurisprudence and the examination guidelines of the EPO before it started granted software patents).

The State Secretary also writes:

• "The inventiveness of the invention must lie in the technical aspects of
  • the invention. Therefore, inventions which are only inventive regarding the non-technical aspects, are not patentable."

This is incorrect, the inventiveness can lie entirely in the non-technical aspects. Article 2b of the Council proposal states:

• "The technical contribution shall be assessed by consideration of the"
  • difference between the state of the art and the scope of the patent claim considered as a whole, which must comprise technical features, irrespective of whether or not these are accompanied by non-technical features."

As such, "the patent claim considered as a whole" must indeed contain technical aspects (that's what "which must comprise technical features" refers to), but a computer is almost per definition technical. A claim like "a computer on which a computer program is executed that does this and that" perfectly fulfills this definition and thus is patentable, even in the inventiveness lies in the non-technical aspects (the software) of the invention.

4. Problems of the Council proposal and unjustified accusations of the EP

4.1 Article 2 - Definitions

The reference to "forces of nature" as proposed by the European Parliament is not ridiculous as the letter insinuates. In the German jurisprudence, the application of forces of nature has been a central criterium for patentability for decades already. In The Netherlands, the High Council already decided in 1950 that (process) patents must cause a "change in nature". The EPO also used a similar criterium in its judgements until 1984 (until from 1985 on, they started granting software patents).

The letter states:

• "All processes take place by forces of nature."

The EP states that the invention (which must be novel, contain an inventive step and must be applicable) must lie in a (consequently novel) application of the forces of nature, not that that the application of your invention must involve the usage of natural forces. A computer program running on a computer is not a novel usage of natural forces, because that computer was built 100% to be able the execute the instructions in that program (so this is not a new, but a known usage of forces of nature, regardless of how new the program itself is).

The EP states that innovations in the realm of data processing are not patentable, because that is the only thing you can do with a computer program. However, it does not state that if a technical invention also contains data processing, it suddenly becomes unpatentable.

An example:

Suppose you have a completely computer/software controlled weaving machine. You then invent a new way to position the knives under a certain angle which enables them to cut the threads faster. The only thing you have to do to implement this invention, is make a small change to the computer program that controls the machine. This (how those knives have to be positioned and moved) is perfectly patentable under the proposal of the EP, because your (technical) invention is not the modification of he computer program, but how you have to position those knives in order to cut more efficiently.

The changes to the computer program (and possible mathematical formulas which describe the trajectory of the knives) are however not patentable. Summarising, one can thus say that software itself is never patentable, not even when it is part of a technical invention. A technical invention is however always patentable, whether it involves the use of software or not (and one can also patent the technical invention + the software, if one would want to do so).

As such, a computer program that improves a video signal by processing it using some mathematical formula, is indeed not patentable according to the EP, not even when it is used in a TV. In fact, there is not even a practical difference between a TV and a computer monitor. This is exactly the kind of software patents which are harshly criticised in the US, and which the EP does not want to inflict on Europe. An improvement to e.g. the electron cannon inside the tube (which can be computer-controlled), remains patentable however.

It is further stated that:

• "Finally the EP proposes, with regard to article 2, via the amendments
  • 38, 44 and 118, to reserve patent rights to branches of industry that produce material goods. It is unclear what is intended by this."

That paragraph puts words in the mouth of the EP and contains strange reasonings. The amendments mean that the patent must be "susceptible of industrial application", and that "industry" in this context must be understood as "the automated production of material goods". This does not allow for the conclusion that a patent holder must perform this manufacturing himself and whether or not Microsoft produces mice (which it in fact does do), is completely irrelevant.

4.2 Article 4 - Limitations

The articles of the EP here are more or less irrelevant if its article 2 is kept. After all, in that case the term "invention" from the EPC and the term "technical contribution" from the guidelines of the EPO are defined in such a way, that software patents become impossible.

The version of the Council however does not limit anything, contrary to what might seem to be the case at first sight.

Article 4 bis 2:

• "Accordingly, inventions involving computer programs ... and do not
  • produce any technical effects beyond the normal physical interactions between a program and the computer ... shall not be patentable."

The "normal physical interaction between a program and a computer" is a strange expression. Nobody outside the EPO talks about a "physical interaction" between a program and a computer. It means more or less the same as "the normal physical interaction between a cook and a recipe": a cook follows the instructions in a recipe, just like a computer follows the instructions in a program.

The EPO itself even wants to get rid of this unclear wording, and notes it was only necessary while waiting for the exclusion of software patents to be removed from article 52 of the EPC (this however did not happen, which was the original incentive of the Commission to introduce this directive):

• "This scheme makes no mention of the 'further technical effect' discussed
  • in T1173/97. There is no need to consider this concept in examination, and it is preferred not to do so for the following reasons: firstly, it is confusing to both examiners and applicants; secondly, the only apparent reason for distinguishing 'technical effect' from 'further technical effect' in the decision was because of the presence of 'programs for computers' in the list of exclusions under Article 52(2) EPC. If, as is to be anticipated, this element is dropped from the list by the Diplomatic Conference, there will no longer be any basis for such a distinction. It is to be inferred that the Board of Appeals would have preferred to be able to say that no computer-implemented invention is excluded from patentability by the provisions of Articles 52(2) and (3) EPC." http: //www.european-patent-office.org/tws/appendix6.pdf (point 1 under "Notes")

The fact that the exclusion of "computer programs as such" is confusing, is also noted in the following citation:

• "From the legal part of the study it emerged that the present ban on
  • patenting "software as such" leads to problems. The present statutory
  exception for "computer programmes as such" not only does not allow reasonable interpretations, but also in practice leads to a tremendous amount of arguments which is incomprehensible, not only for outsiders." Micro- and Macroeconomic Implications of the Patentability of Software Innovations (Max Planck Institute and Fraunhofer Institute, 2001)

This clearly demonstrates that instead of diminishing the legal uncertainty and unclarity, the "limitations" of the Council only increase them by insisting on expressions such as "computer program as such" and "further technical effect". On top of that, from articles 4a(1) and 4a(2) of the Council, it follows that the Council interprets "computer program as such" as the source code or executable (object code) form of a certain computer program. However, no-one ever asks for a patent on a certain computer program, but always on a certain algorithm (e.g. to improve a video signal), since the written program is already protected (for free and much stronger) by copyright.

Consequently, the Council here excludes something from patentability which no-one would ever want to ask a patent for (and until now never asked one for), and which as such has nothing to do with the problems of software patents at all. In fact, such patents are not possible in the US either. It does however demonstrate that the term "software patent" is a badly chosen term: it is not about patents on software, but about patents on rules for mathematics, presentation of information and business methods. One can however only obtain patents on these unpatentable rules, if one states that they are executed by a computer (introducing so-called "technical character"), from which the name evolved.

4.3 Article 5 - Form of the claims

The claims of a patent from the part of the patent which stipulates on what one wants a monopoly in exchange for publishing his invention. Point 3 already showed how article 5.2 allowed patents on computer programs on their own (and as such does not only forbid usage, but also publication).

The request of the EP to publish source code (which in fact has the support of a.o. FENIT), is not in any way a trade of copyright protection in exchange for patent protection. One never patents an entire program, but only specific algorithms (e.g., to quickly calculate a square root). One must then only publish a reference implementation of this algorithm without limiting copyright clauses, and of course this does not mean that one has to publish the source code of each program which uses this algorithm.

If software forms part of a patent, the availability of source code is essential. The description in the patent offers little of concrete value. Additionally, it is difficult to make a program unreadable in the way that can be done using superfluous juridical wordings. In comparison: patents on more tangible subject matter often also contain concrete examples, such as drawings.

4.4 Article 6 - Interoperability

It is strange that the Cabinet stresses decompilation so much, since this term denotes the meticulous studying of the internal working of a program to find out how it works. If a particular technique is patented however, this should not be necessary at all, since all necessary information should be found in the patent in that case. Of course, it's possible this is an acknowledgement of the fact that most patents can only be understood by juridical experts and not by software developers.

In either case, this does not help at all: decompilation simply cannot be forbidden using a patent. A patent gives a monopoly on the use (or production/distribution) of something, not on the studying of something. As such, this guarantee is an empty promise. Additionally, if you then afterwards want to use the patented technique for interoperability purposes (e.g., to read a Microsoft Word document), it suddenly turns out that there is no exclusion from patentability for such usage. As such, this remains forbidden en you're left in the cold.

A few recitals of the Council (17 and 18) state that one can never use a patent to infringe on the anti-trust law, in particular when interoperability is concerned. This is worthless for several reasons:

1. One is never allowed to infringe on the anti-trust law anyway;
2. This concerns European anti-trust law, and (intellectual) property law falls under the jurisdiction of member states. This means that the European anti-trust authorities must first indication that there is a big influence on an area where they do have jurisdiction before they are allowed to take action;
3. The EU is not going to start an anti-trust case for each complaint about interoperability problems;
4. Anti-trust law is too blunt a tool (it took years to get a judgement in the Microsoft case, and will take several more years if Microsoft appeals);
5. Interoperability is a basic building block of the informatics world, like the Cabinet confirms. One needs strong ground rules which explicitly guarantee this right.

The Cabinet justifies the removal of the EP amendment by claiming that it violates article 30 TRIPs, which states that limitations regarding enforceability may not "unreasonably conflict with a normal exploitation of a patent". It does not define what is reasonable and what isn't, however. We think it is reasonable to allow others to use a certain technique, if this is necessary to obtain interoperability.

As a matter of fact, articles 7 and 8 of TRIPs confirm this interpretation is legitimate. After all, they state a.o. that all TRIPs rules must be interpreted so that they encourage technical innovation, do not conflict with public interests and do not allow for abuse by owners of intellectual property rights.

5. The goals are not attained

Of the four goals (noted on page 3 and repeated on page 16), three are demonstrably not attained:

1. Creating legal certainty in Europe by means of harmonisation and clear rules

See point 4.2 of this document.

2. Keeping the patentability of computer-implemented inventions.

This is a political choice which is presented as some kind of dogma. The claim that treaties force us to do this is cannot be substantiated.

3. The prevention of trivial patents by requiring inventiveness regarding the technical aspects of the invention.

Point 3 of this document demonstrates that the inventiveness can lie entirely in the non-technical aspects. Additionally, the version of the Council does not even define "technical", and there is no reason why something technical could not be trivial.

4. The preservation of the possibilities for designers to take note of the computer-implemented inventions of patent owners so that via the way of decompilation and interoperability they can design new software products which connect with patented invention.

See point 4.4 of this document.

6. Treaties are interpreted in a tendentious way

The letter regularly refers to obligations regarding the TRIPs treaty. Whether the obligation of article 27 to allow patents for "inventions" in "all fields of technology" also means that software must be patentable, is however quite controversial. Prominent juridical experts such as prof. Verkade and prof. Hugentholtz do not agree with this interpretation.

After all, two conditions must be met: an invention must be present (software, mathematical methods etc. are not inventions in Europe) and additionally, the invention must lie in a field of technology (which depends on how one defines "field of technology", since TRIPs doesn't specify this). Further, the previously mentioned articles 7 and 8 always allow exceptions in case the rules are considered not to encourage the economy or the general innovation climate.

In fact, the TRIPs treaty even explicitly forbids software patents. The WTO states on its website regarding article 10.1 of TRIPs:

• "The obligation to protect computer programs as literary works means e.g.
  • that only those limitations that are applicable to literary works may be applied to computer programs."

Patent protection is not possible for literary works. One cannot work around this by stating that this only applies to computer programs "as such" and not to their execution by a computer, because article 4 of the EU Copyright Directive regarding Computer Programs from 1991 states that a computer program as literary work includes the following:

• "... the permanent or temporary reproduction of a computer program by any
  • means and in any form, in part or in whole. Insofar as loading, displaying, running, transmission or storage ..."

The execution is thus also covered by this. Additionally, there is still the WIPO copyright treaty, which states in its article 10:

• "(1) Contracting Parties may, in their national legislation, provide for
  • limitations of or exceptions to the rights granted to authors of literary and artistic works under this Treaty in certain special cases that do not conflict with a normal exploitation of the work and do not unreasonably prejudice the legitimate of the author."

The normal exploitation of a computer program includes the usage and selling of it, and both can be forbidden by a third party if it possesses a software patent on which the program infringes.

Accordingly, Europe does not have to fear international sanction, and could even (if it wanted to) starte a procedure against the US because of infringements.

7. The letter is also interesting for what it doesn't say

The State Secretary nowhere rebuts the argument that a "computer program executed by a computer" fulfills the definition of "computer-implemented invention".

A critical report of the Federal Trade Commission on (software) patents: http://www.ffii.org.uk/ftc/ftc.html

Economists speak out against software patents: http://www.researchineurope.org/policy/patentdirltr.htm

Dutch Companies speak out against software patents: http://wiki.vrijschrift.nl/VerklaringBedrijven

The EPO and national governments earn money from each granted patents. Te ability to grant more patents is thus financially interesting.

The quality of the patents granted by the EPO is declining: http://news.zdnet.co.uk/business/legal/0,39020651,39157016,00.htm

Software patents have an effect outside the software branche. Companies with internal software development are hit as well, and also enterprises with an online Internet store. The 30,000 already granted software patents are mostly trivial and especially very broad. Een simple web shop can easily infringe on 23 patents: http://webshop.ffii.org

In its opinion on the original Commission proposal (which is almost identical to the Council version), the European Economical and Social Committee almost literally burns the text to the ground: http://eescopinions.esc.eu.int/EESCopinionDocument.aspx?identifier=ces\int\int145\ces1031-2002_ac.doc&language=EN

8. Additional remarks

• On page 15, point 3: "Then it will be clear to what degree the EP stays with its previous opinion ..." -- During the second reading an absolute majority is required to still change anything (of the number of deputies, not of the number of voters!). Presenting the situation as if the EP still has a full participation possibility is thus misleading. The vote in first reading more accurately portrays the intention of the Parliament.

• The State Secretary expects a lot from the evaluation possibilities (page 12). What can one expect from an evaluation? While preparing this directive, a lot of studies were already carried out (and disregarded). And although evaluating afterwards is easier than evaluating before, countless studies already show it is extremely difficult to get a grip on the quite diffuse problem of software patents. If we now open the flood gates, the harm will already have been done by the time the evaluation occurs. As the European Economical and Social Committee stated in its opinion regarding the Commission text: "It is hardly plausible to have us believe that the directive would only be a sort of reversible three-year experiment, at the end of which an assessment would be made."

• On page 1: "Without protection, companies will invest less in innovation." -- This is an assumption which cannot simply be taken for granted. Economical research shows that software patents have lead to diminished investments in R&D. An extensive overview can be found at http://swpat.ffii.org/archive/mirror/impact/index.en.html

• On page 2: "I am convinced that, if we want to encourage the knowledge economy in Europe, the protection of such inventions must be done in a good and univocal way." -- The content of a number of important articles in the directive is vague and open to multiple interpretations, which is a showstopper for this goal. Whereas the amendments of the EP mostly try to create clear conditions, the Council text seems to attempt to create wiggle room for broad assumptions regarding the patentability of software, without clearly stating this.

• On page 3: "The creation of legal certainty in Europe by means of harmonisation and clear rules;" -- The legal certainty is mainly created for the owners of (many) patents. Whether the situation will improve for smaller software developers is another question, because of the risk that they could be using software-solutions which later turn out to be patented. A software product with some degree of complexity can easily infringe on tens of patents, and it is impossible to keep track of all these patents. The effect of the Council version of the directive goes a lot further than plain "harmonisation". Whereas currently in most member states software patents would be considered invalid, they would become valid in the whole EU according to the directive. The term "harmonisation" suggests that the situation would no change significantly on the whole, while in fact it would.

• Further, on page 3: "Preserving the patentability of computer-implemented invention" -- This formulation also suggests that the introduction of the directive (Council version) would change little to the current situation. In practice, the patentability is extended significantly and codified.

• On page 14: "There is a discussion about whether or not the EPO is sufficiently critical when evaluating patent applications. If the quality of the examination by the EPO is lacking, the previously mentioned trivial patents can be granted. This is an unwanted situation which is bad for innovation in Europe. Article 8 states that the Commission will diligently observe the practice of the EPO. The EC (sic) can however not directly influence the EPO." -- Exactly because there are doubts regarding the evaluations by the EPO, it is important for the directive to clearly state what conditions computer-implemented inventions have to fulfill. Instead, the articles of the Council leave much open to interpretation and as such does not offer abundant protection against the appearance of pure software patents and their abuse. See also: http://news.zdnet.co.uk/business/legal/0,39020651,39157016,00. htm

• Regarding article 4: "The in this article posed requirement of technical contribution must also prevent so-called trivial patents." -- The problem of trivial patents falls entirely outside the scope of this directive, and can also not be solved in a whim by stating that inventions must make an almost completely undefined "technical contribution".

• By stating that a "computer program as such" is not patentable, the Council does not remove any uncertainty, since a large part of the discussion revolves around the question of what is exactly the difference between a "program as such" and a "computer-implemented invention".
• The "invention" of Microsoft regarding the computer mouse: this is not about "a patent on an improvement to a mouse". Nothing is changed to the mouse. When you press a mouse button, the operating system on the computer receives a signal (that is the current situation as well). When you release it again, another signal is received. The "invention" lies in the fact that depending on the time between the first and the second sinal, a different application is launched.

This also nicely illustrates why the EPO considered the Amazon 1-click selling as a technical invention: you click 1 time with a mouse instead of several times, and since a mouse is a technical device, this was considered to be a technical invention (reduction of the number of necessary mouse clicks). This patent claim was not granted in the end however, thanks to the fact that (fortunately) someone else had already done the same thing before. Otherwise, we would have had bad luck.

9. General misconceptions regarding software patents

Not yet translated.