Patents on computer-implemented inventions and pure software patents:
what's in a name?
(Note: there is also a three-page and a two-page PDF version of this text available. The text of both versions is identical. The font of the two-page version is simply somewhat smaller and it contains less whitespace)
This text tries to explain why allowing patents on
"computer-implemented inventions" (as defined in the proposed directive) is no different from allowing "pure software patents", and as such why opponents of the directive on the patentability of computer-implemented inventions always talk about a directive on software patents.
What do you understand under "computer-implemented invention"?
- The most straightforward meaning of "computer-implemented invention", would
be a solution that is actually implemented by a computer (after all, that's what the term literally says). Computers only implement logical/mathematical solutions. That is why they are called computers: they
compute, they do simple math very quickly. So with only a computer you can only
implement logic, math and rules of organisation. These are not inventions
according to Art 52 EPC. So the straightforward meaning is contradictory.
- What is hopefully meant with the term (but which is contrary to the
definition in art 2a), is gadgets which implement an invention, and which also happen to have a computer inside.
- I'll use mobile phones and washing machines as examples on this page. If
there are other things you would like to see covered, please don't hesitate
to mail me!
How do patents normally work, for things that are not
- Mobile phones and washing machines were invented a long time
ago, so you can't invent them again. Since patents are only granted
for inventions, you cannot patent a mobile phone nor a washing
- You can however create a mobile phone with a new, better antennae,
or a washing machine with a better motor. So although the mobile phone
and the washing machine are not new themselves, they can contain
inventions, and these inventions can be patentable.
- You can then get a patent on "a mobile phone with a new type
antennae that is created such and such", or on "a washing machine with
a new motor that is constructed like this". Since you actually invented a
new type antennae and motor however, you could just as well get a
patent on "a new type antennae that..." or on a "new type motor that..."!
(except when the only advantage of your invention is in conjuction
with that specific application)
- Conclusion: when you invent something, you can patent this
invention. You can say in your patent claim that the invention is used in
one or other device, but that merely makes your claim more specific; it
does not influence whether or not your invention is patentable.
So what about computer-implemented inventions?
- Example 1: suppose we create a program for a mobile phone which dramatically
improves its reception, or a revolutionary washing program for a
washing machine that gets all stains out.
The above innovations do not lie in the fact that you use a computer program,
but in your research in how to send and received the signals
that are in the airwaves (i.e., how to modulate and demodulate them, ...), or what exactly your washing machine should do to
get those stains out (i.e., which motions to make, what temperature the water should have, ...).
These innovations may be patentable inventions, but that will be decided
like for any other innovation by traditional patent law and case law. The fact that there's
also a computer (program) involved, is irrelevant.
- Example 2: compare the above with a program that guesses names of contacts in your
phonebook as you type them on the mobile phone, or with a more
configurable setup for the washing machine (so that you can tell it
exactly how much water you want, and the temperature for each washing
- These second examples contain innovative software, but
computer programs as such cannot be patentable according to Art 52 of the EPC.
- The washing machine and the mobile phone are also not patentable inventions,
since they are not new (nor do they contain an invention). This means no invention is present, and no patent must be granted.
- The antennae signal processing or washing process of the first example could be
inventions, and since computers participate in their implementation, you might
even talk of
computer-implemented inventions if that expression were defined as by the European Parliament in its first reading of September 2003.
second examples (washing machine management application and phonebook
user interface) there is no invention, there is only an innovation in
- Conclusion: if you insist on calling all examples computer-implemented
inventions, you force yourself to consider the software in the second
examples as the invention (as it's the only new thing), so you are granting pure software patents.
This is the risk of using
contradictory terminology, without even giving an explicit definition
that solves the contradiction (with some violence to semantics).
Synthesis: how patents are supposed to work.
- Patents are for inventions.
- Mobile phones and washing machines are already invented, so you
can't invent them again. They can all contain inventions though,
and the invention may deserve a patent.
- The patent may claim a mobile phone with a new antennae
(invention = antennae, not mobile phone), or claim a washing machine
with a better motor (invention = motor, not washing machine).
- The objects that are claimed are different from the invention.
The invention must be in the claimed object in order for the applicant to
be allowed to claim this object, but the invention is not the whole claimed
- The invention is what you add to the prior art, so the invention is
the contribution. This contribution must be technical, inventive,
new, and susceptible of industrial application. It is not enough that
the claimed object is technical.
- The reason you have to look at the invention and not at the claimed
object as a whole, is that you want to give patents to promote
innovation. You can only promote innovation if you look at the
contribution, since that's the innovative part.
- Computer programs as such are not patentable according to EPC Art 52:
this means that adding a computer program to a patentable invention
does not render this invention unpatentable, but it doesn't mean
that adding a (new) computer program to an otherwise unpatentable
technical apparatus makes the computer program patentable! (however,
that's the way the EPO/CEC/JURI interpret that statement)
- The CEC an JURI language confuses things terribly, and
pretends the (computer-implemented) invention would be the mobile
phone or the washing machine (e.g., because it contains a known motor
or antennae or even a computer), and says that additionally there must also
be a technical contribution (possibly including non technical features
-art 4.3- such as a new program driving the motor). According to them
however, the contribution is not required to be technical, new or
susceptible of industrial application (so it can be pure software), it
must only be not obvious.
- If you (like CEC and JURI) say that the invention is the claimed object, this implies that the invention and the technical contribution are two different things. You can then apply
some patentability test to the invention and some to the contribution. This leads to doubts such as "Should this invention (a mobile phone) be
patentable?" There is no answer, because a mobile phone is not an invention. The invention is something it may have, but you can never tell whether it should be patentable by looking at the whole mobile phone.
Can you give an example of how the proposed directive allows software patents?
- Simplified, the European Patent Convention states:
- Art 52.2: Programs for computers shall not be regarded as inventions.
- Art 52.3 The provision "programs for computers shall not be regarded as inventions" shall exclude patentability of programs for computers only to the extent to which a patent relates to programs for computers as such.
This is equivalent to the Swedish law (harmonized with the EPC in 1976):
"That which only constitutes a computer program, is never regarded as an invention."
("Såsom uppfinning anses aldrig vad som utgör enbart ett datorprogram.")
This contradicts article 2 of the directive, which simplified says:
a "computer-implemented invention" means an invention the performance of which involves the use of a computer and having features which are realised by means of a computer program.
- Conclusion: article 2 in its present form is a Trojan Horse, not a washing machine.
So how can we fix the directive?
- Do not push for amendments that mix the tests for patentability
or which loosen its conditions. Examples of bad amendments are am4JURI (to recital 11, but
note that the CEC article is no good either), am5JURI (to recital 12,
restatement of scrapped article 3), am7JURI (recital 13(c)), am14JURI (to
article 2 (a)), ITRE-6 (to article 2 (b)), am16JURI (to article 4) and
am18JURI (comp-1 to article 5).
- Do push amendments that clearly restate Art 52 of the EPC and which introduce the
requirement for handling of forces of nature in the definition of "computer-implemented
invention" (and not just in the requirements for making a computer-implemented
invention patentable). Introduce clear and strict definitions for the the terms
"technical" and "industrial application". Hold up the right for publication and conversion
(art 6(a) that was voted in JURI).
- The reason the "forces of nature" are so important as a condition to have an invention, is that a computer program as such can not control the forces of nature, so this condition will properly exclude them from patentability. All innovations that have been traditionally patentable, will remain so in the future this way, but you can stop the EPO's drift to granting patents on pure software and business methods this way.
- FFII will soon publish a
list of amendments that are to be tabled by different political
groups and will explain which ones limit patentability and clarify the
current situation. Of course, all amendments will be accompanied by the
detailed FFII analyses you've all come to love so much, similar to the analysis
of the amendments tabled in JURI! Mail me if you want to be
notified when that page becomes available.
This document was written by Jonas Maebe, but was created with a lot of
insightful input from Xavier Drudis Ferran. Without him, this page would
not exist. Credits go to Erik Josefsson for the Trojan Horse.