Legal analysis of software patentability

Why the question of software patents is badly put

The proponents of software patentability justify it by the need to
protect the software components which are increasingly present within
innovative tangible inventions, and for which, according to them,
copyright does not give sufficient protection. An example that is
often considered by members of patent offices is the one of a new
washing machine that washes the linen more efficiently, and such
that the innovating washing sequence is implemented by means of a
computer. The software carried out by this computer would be,
within this framework, at the origin of the technical contribution
brought by this new machine, and should thus be covered by patent
claims.

This test case is used as a conceptual basis to postulate the existence
of "technical" software, as opposed to software "as
such" which would not be patentable because they do not
produce any technical contribution.

However, postulating the possible existence of a discrimination
between "technical" and "non-technical"software
based on the examination of their effects, which we call the
"doctrines of software technicality", cannot hold vis-a-vis a
rigorous analysis. In the above-mentioned case of the washing
machine, the technical effect is the improvement of the washing of
linen by the machine. This effect is obtained by application of an
innovating washing process, consisting in the dispensation of
detergent at such or such moment of the washing cycle, by the
realization of better mixing of water and of the linen, etc., all
operations carried out by the machine in the material world. The
software operating the machine does control this innovating
washing process, but is not, in itself, the source of it.
It is the process which is innovative and produces the technical
effect wished for, not the software.

Any software, carried out by a computer, does nothing but handling symbolic quantities according to a
pre-established program, independently of the way in which these
symbolic quantities are materialized within the computer which carries
out the program (electrons, photons, magnetic moments, quantum states,
etc). The computer, in order to interact with the physical world,
needs peripherals, which materialize the symbolic systems quantities
into physical actions in the outside world: activation of a servo unit,
displaying of a particular luminous information on a screen, etc.
The software which drives the washing machine can thus be carried out
on a simulator, and run exactly as it would do within the washing
machine, but without producing the expected technical effect. It is
because the technical effect is independant from the considered
software, but resides in the effective carrying on of the process.

The legislator had already analyzed this situation to draw the
conclusions which it imposes, and to authorize the claiming of
industrial processes. The manufacturer of innovative washing machines
can thus claim his process of innovative washing as well as the
machine which carries it out without needing software patents at all.

Let us notice to conclude this section that a software which would sum
the values of two quantities of reagents that are present in the tanks
of a chemical reactor could be regarded as "technical"
according to the doctrines of software technicality, while it would
not if it were used to compute the sum of the balances of two bank
accounts, while both software perform exactly the same process.

Similarly, one cannot consider that some software produces a technical
effect when it allows to "perform computations more
efficiently" or it "improves the efficiency of a
computer". Let us remark at first that these doctrines are even
more radical than the ones of software technicity, because it would
allow one to consider as "technical" any software that
provides better processing performance than any existing software,
without any reference to any given problem, or rather by considering
that any use of a computer constitutes in itself a technical application.

The actual efficiency of a given algorithm being completely
dependent on the architecture of the computer onto which it has
been implemented and is running, the alledgedly technical
features of an algorithm are not linked to it, but rather are
based on the adequacy between the subjacent hardwere and
implementation considerations that belong to the realm of
copyright, since they are specific to each individual
implementation of software on a given architecture.
The consideration of increased efficiency is therefore not
effective to define a stable frontier between alledgedly
"technical" and "non-technical"
software.

The impossible formalism of software technicity

The conceptual impossibility to discriminate between "technical"
and "non-technical" software condemns in advance to failure any
attempt to embody this discrimination into legislative texts.
An obvious example of this failure is the text proposed by the Irish
presidency which, while defending itself from allowing total
patentability of software, indeed achieves this result.

Recital 7b, supposed to limit software patentability, is contradicted by
Recital 13 and Article 2b which say explicitly that software alone can
carry out a "technical contribution", and by Article 5b which
allows the patenting of software carrying out such "technical
contributions". These articles allow any “inventor” to
patent any type of software, because those are always created to solve a
given problem. In the same way, Mr. David Sant, official representative of
the European Patent Office attached to the European institutions, confirmed
in a conference at the European Parliament the official interpretation of
the European Patent Office, which considers as a technical contribution
any situation within which one can say that one has to consider technical
factors before beginning to program, a situation encountered by all
computer programmers and which neutralizes Article 4a, allegedly
designed so as to limit software patentability only to
"technical" software.

The choice of the legislator thus cannot be a half-measure: either he
fully endorses the enablement of a regime of total software
patentability, or it entirely rejects it, as does the doctrines worked
out by the European Parliament during the vote of the directive in
first reading, which rejects claims on software, while still allowing
the patenting of material inventions and of industrial processes that
make use of software for their control, which was already the case in
the past, since it is the very meaning of Article 52.2 of the
European Patent Convention.

This binary choice, of considerable economic and strategic consequences,
cannot be done lightly, all the more than the extension of patentability
would be practically irreversible.

Are software patents legally mandatory?

Some players refer to Article 27 of the TRIPS Treaty, which requires
the existence of patents for any field of technology, to justify the
legal need for software patents. As underlined by Mr. Paul Hartnack,
General Comptroller of the United Kingdom Patent Office, this
interpretation is not the only possible one.

One can consider, like the Commission and the Council do in their
proposals for a directive, that software is intrinsically technical,
and that Article 27 of TRIPS requires to have software patents. One
can also consider, like the European Parliament, that software does
not reside within a field of technology but with that of intellectual
works, and thus beyond the range of Article 27 of TRIPS.

Here again, policy makers are free to decide, because the choice
between the two doctrines is still possible.

Software Patents and copyright are exclusive

Until now, the copyright and patent regimes had coexisted without
conflict, their purposes being different, since copyright dealt with
the protection of individual works of authors, while the purpose of
patents was to protect inventors of innovative processes implemented
within (non-unique) apparatuses designed to solve a given problem.

The extension of the patent system to software constitutes the first
major attack to copyright law, to which the legislator had, after deep
thinking, incorporated software. This assimilation was altogether
natural because software, just like books, consist in the production
of a textual original work (the source code of the program), resulting
from the combination of elementary ideas. For an adventure novel,
these elementary ideas can be: "love scene on a balcony",
"twins being mistaken one for another", etc. For software, it
can be: "alphabetical sorting in a list", "displaying a
progress bar telling the user to have patience", etc.

Unlike copyright, which protects final works, software patents, which
protect against the imitation of features, allows the protection of
these elementary ideas, and thus prevent whoever to realize a program
implementing a protected idea. This would amount, in the case of
books, to be allowed to claim elementary ideas such as the balcony
love scene, although the same idea can lead to very different
implementations, such as the balcony love scenes of Romeo and
Juliet and of Cyrano de Bergerac, which are not plagiarisms
one of the other.

Software patents, by allowing their holders to claim elemenary ideas,
thus constitutes an extremely powerful monopoly-building tool, because
the holder of a single patent can prevent the selling of all
software implementing this idea, whatever their application domains
can be.

In this extent, software patentability can be regarded as violating
the TRIPS and WIPO treaties. Indeed, Article 10 of TRIPS stipulates
that software belongs to the realm of copyright, and Article 13 of
TRIPS, as Article 10 of WIPO, that one cannot systematically prejudice
the legitimate interests of the rightholders. However, considering
software patentable indeed prejudices the rightholders, since the
author of an original software can be prevented from marketing it
for the reason that this software infringes on one or several
software patents, irrespective of legitimate exceptions such as
harm to public morals. As more and more artistic creations
make use of software (such as video games, for instance), the
occasions for litigation can only increase, at the detriment
of creation in the considered industries.

Although one can argue that these articles in TRIPS do apply only
to the exceptions which the legislator could introduce in the
copyright regime only, it is clear, considering the spirit of these
treaties, that the forging of software patentability, which has
no legal necessity, tends to have a destructuring, rather than
structuring, effect on existing intellectual property regimes that
exist to date, and no one knows what could be the long-term
consequences of the extension of the patent regime to fields for
which it had indeed not be designed for, neither for which it has
economic justification.

The decision to introduce software patents amounts to answering
implicitly the question of knowing which system, between the patent or
the copyright regimes, must have preeminence, without any possibility
of debate nor of being able to check the validity of this approach.

The copyright regime is oriented towards the protection of existing
works, already accessible to the public since already created, the
existence of the protection making it possible to regulate by
subsequent contracts the way the public can access these works.

The software patent regime allows one to claim an idea, irrespective
if there exist or not concrete achievements implementing these ideas
and being able to benefit to the public.

In the case of software, priviledging the patent regime compared to
the copyright regime thus amounts to priviledging the defensive and
anti-competitive behaviors which are inherent to the patent system,
based on exclusion and rent seeking, as opposed to the active behavior
of the creator of original works putting them at the disposition of
the public in a context of free competition.

Economic analysis of software patentability

The legal analysis of software patentability presented above made it
possible to show that there exists no legal obligation by international
treaties, TRIPS for example, to create them. Quite to the contrary
even, the existence of software patents creates incoherence and legal
insecurity for software rightholders. The decision to have or not
software patents within the European Union can thus be freely decided
on the basis of economic and strategic criteria only.

Another consequence of this analysis is that, contrary to what the
writers of the directive have announced, the conditions for software
patentability in Europe will indeed be similar to those into force in
the United States, with the result that the current situation of the
United States on the matter constitutes the exact reflection of what
the situation will be in Europe if the version of the Irish presidency
of the directive were accepted.

Let us also notice that, with respect to this analysis the conceptual
bases of which are simple and accessible to any person who has some
knowledge of the domain, , one can legitimately wonder about the
the apparent ingenuity of the writers of the directive who pretend
that there exist a discrimination between "technical" and
"non-technical" software.
If one gives them a minimum of credit, one will arrive at the
conclusion that they are fully aware of this impossibility, which
leads to wonder about their motivation to legitimate total software
patentability even at the price of a distortion of the truth.

If simple ideological motivation cannot be neglected, it is not
sufficient to account for all of the observed behaviors. One therefore
has to acknowledge the existence of a collusion of economic interests
between the leaders of the patent offices, the members of the various
commissions of experts involved in the drafting of the directives, and
the people in charge of the legal departments of some big companies,
with the deliberate purpose of providing these latter the most
efficient anti-competitive weapons as possible, considering that the
legal services of these companies are the biggest customers of the
patent offices, which themselves live on the fees of granted patents,
and considering that there exists an important professional porosity
between these three communities.

Software claims, interoperability, and monopolies

The analysis of the version of the directive endorsed by the Irish
presidency shows that most of this text is designed so as to favor the
creation and enforcement of monopolies on whole sectors of the
data-processing industry, without any counterpart for the consumer.

Article 5b of this text introduces program claims, which make it possible
to sue any software creator, vendor, or user, provided that a patent has
been granted on any of the elementary ideas that the software implements.
This would allow big companies with large legal departments to scare
potential users of competing software, well before inolving courts, thus
distorting competition in an insidious way.

This text also rejects any right to interoperability. Article 6 does
nothing but maintaining the exceptions of retro-engineering and
decompilation allowed by the directive of 1991 regarding the
protection of software by cupyright. But whereas these exceptions are
sufficient in the context of copyright, they are worthless in the
context of patents, since any manufacturer who reimplements, for the
sake of interoperability, processing methods for a patented file
format, would be infringing of this patent.

The mode of creation of software, based on cumulative innovation, as
well as the very short development cycles of this industry, do not
allow any longer to account for the existence of a patent system which
is expensive, heavy and slow.

Indeed, unlike in the material world, the time necessary for a
competitor to imitate some software is hardly any less than time taken
by the development of the original software, because the development
itself constitutes the most expensive activity in comparison to the
simple fact of having the idea of the software, which yields that, in
this very reactive market, initial innovators have time to profit from
their protucts before any competitor appears.

Moreover, any such competitor can gain market shares only if its
product is of better quality, brings additional innovations, or is
less expensive than the original product, which can only be, as
development costs are almost identical by what precedes, if the market
price of the original product were too high. A proof of this is that,
on any given market segment, a product of good quality and at a
reasonable price has little competitors.

Therefore, without software patents to build monopolies, competition,
innovation, and consumers win. These consumers are both private
individuals and companies of all sizes and all fields of activity,
because the impact of information technologies on productivity affects
every sector of the economy.

To conclude, let us note that patents does not bring any information
which is not already public, since the features of a software that are
protected by some patent naturally appear when using the software,
thus allowing any "person skilled in the art" to understand
its principles of operation. Let us also remark that the jargon used
to write these patents is in general incomprehensible to any computer
professional, and of a technical level which is so low that no
information is usable in practice, since the goal of the drafting of
the patent is to be as generic as possible in order to block as many
competitors as possible.

The economic threat of the software patents for the European data-processing industry

According to a report by the French Commissariat au Plan issued
in 2002, the French software sector is essentially made of small
structures. Indeed, 58 % of the 28,500 companies of the sector
have at most two employees. These figures are almost equivalent in
Germany and in the other countries of the Union. Small and
medium-sized enterprises of the software sector represent more than
70 % of jobs and wealth produced by this sector in Europe,
not counsidering the software departments of large industrial and
service companies (such as banks, insurance companies, etc) which
contribute to a great extent to the development of "in house"
software which will never be marketed but could nevertheless be
subject to software patent infringement lawsuits.

Small companies are on average more innovative, more reactive, and
more inclined to base their development on the creation of new
products than on an established system of monopolistic revenues. In
fact, since SMEs of the software sector represent its economic
majority, as well as a pool of creativity that is necessary to the
development of new technologies, all new regulations should be
developed for their benefit.

Regarding the introduction of software patentability in Europe, this
same report of the Commissariat Général au Plan sees in the
"abrupt decision to extend patentability to software[...]
real dangers for the European industry, because of the considerable
unbalance between the United States and Europe on the matter",
and insists that "Only the armed peace prevailing to date,
because of the legal uncertainty that remains around the notion
of software patent, indeed explains that existing patents [the
about 30,000 doubtful software patents already granted by the EPO]
are not used more often.".

A system of incentives for innovation cannot be effective unless it
encourages the production of final goods that benefit to citizens and
thus to the society at large. On the opposite, the monopolization
of upstream concepts disadvantages the production of such goods,
because it is then more profitable for an economic actor which has
patented a concept to seek rents from the entities taking the
commercial risk to carry out the final products, rather than to make
such products by itself and be under the risk of patent infringement
lawsuits from its likes. Consequently, the more the extent of the
patent is broad and lasting, the more behaviors of predation and risk
avoidance are favored, to the detriment of innovation. In the case of software,
the minimum duration of 20 years, which would be imposed by TRIPS if
one would consider software as patentable, represents more that 10 times
the average duration of the commercial life of any software, which is
an economic nonsense, as this blocks further developments by other
innovators who must, in order to enter the market, pay fees for basic
ideas the development cost of which, if any, has been paid back many
times already.

At best do large players agree on patent portfolio exchanges, which
allows them to pretend they live in a patentless world, but this has
several perverse effects. First, what economic interest is there for
these companies, and therefore for their customers and by extension
for the society at large, to finance the filing of patent being used
to counter the existence of other patents, whereas the absence of
patents would amount to exactly the same? Then, after such exchanges,
the only entities having to support the cost of the licenses are the
innovative small and medium-sized enterprises, which cannot enter the
markets which would reimburse their investments. The system thus
functions exactly the opposite as it should, wrongly penalizing the SMEs.

Software patents are moreover a danger for investors. Without patents,
any investor is certain to be able to market the software he financed,
and thus to have a return on investment if these products fulfill a
need. With software patents, any economically worthwile software may
be attacked, and the return on investment is more dubious. The very
recent exhaustive study carried out in the United States on more than
130.000 software patents and 1600 Usonian companies between 1976 and
to date showed that the expenditure in software patents replaced more
than 10 % of the expenditure of R&D of the companies, at the
national level.
During Federal Trade Commission hearings on the subject, some company
CEOs reported that they were going to devote up to 35 % of their
Research & Development expenditure to the building of software patent
portfolios, in order to try to protect themselves from infringement
lawsuits, although these costly portfolios could not be of any help to
protect them from the very high legal expenses that a competitor would
like them to spend, at the detriment of their middle and long-term
competitiveness.

In this context, it is legitimate to think that, because the market
allowing the best return on investment will be a Europe without
software patents, the companies that will succeed better will be
European companies, that will be more close to their customers, and more
able to provide the necessary R&D effort. Any company basing its
software development in the United States can be subject to
infringement lawsuits on the basis of US software patents, since
software developers may implement, knowingky or not, algorithms that
are patented in the United States. On the opposite, European developers
would be free to develop their software without constraints. Thus, an
Europe without software patents is likely to succeed in attracting
Usonian companies willing to secure their developments, reversing the
direction of the flow of emigration of computer professionals.

Let us notice by the way that nothing prevents a European company from
filing patents in the United States to try to improve its chances to
penetrate this market. This investment is small in comparison of the
one to obtain European patents, and without common measure with the
tribute which European SMEs should pay to large US companies if software
patents were legalized in Europe.

Last update : 16 May 2004
Author : François PELLEGRINI