Copyright protection of algorithms – impossible or just a question of definition?

Legal protection of algorithms has been subject of much debate over the years, and with the huge technological development, the topic is constantly gaining new relevance. Access to good and effective algorithms has become vital for companies that want to gain strong market positions, as giants like Google and Facebook use algorithms to generate customized content. The prevailing view among copyright lawyers has long been that algorithms are not protected by copyright because they are unprotected ideas behind computer programs. However, I think this is misleading and oversimplified.

The fact that algorithms are just unprotected ideas stems from the basic copyright principle that ideas are not protected. It is only when the author's ideas "have been realized in such a way that through his works something original is created" that it can be protected as a work of intellectual property (Rt. 1962 p. 964). This requirement for originality is now statutory in section 2 of the Norwegian Copyright Act.

One of the starting points for the prevailing view that algorithms are unprotected ideas is perhaps the EU Directive on legal protection of computer programs, which states in recital 11 of the preamble, that algorithms are not protected. However, this only applies to the extent that "algorithms (…) comprise ideas and principles". In other words, the preamble takes into account what is a fundamental problem, namely that it is difficult to establish a general understanding of what algorithms are and how they are defined. For example, in legal literature, algorithms are defined as "the general principle or structure of the program, understood as the coherent, but incremental, principled approach to solving a problem" (Koktvedgaard 2005), a definition that also reflects prevailing view. If we turn to more general definitions, we see that algorithms do not necessarily relate to mathematics or technology, as many believe. Algorithms can, in the broad sense, simply be defined as a set of instructions that step by step indicate how to arrive at a result *.

So why is the prevailing view misleading and oversimplified?

The Norwegian Copyright Act protects "computer programs", which in the preparatory work are defined as "a set of instructions written in programming languages", which enable "to control a computer so that it performs specific tasks" ( 33 (1989-1990) ) page 4). If we compare this definition with the wide definition of algorithms, we see that an entire program can just as well be called an algorithm. The codes that the programs consist of are only descriptions of algorithms, which are protected as long as they meet the originality requirement. Thus, it may be too narrow to say that algorithms are not protected as long as computer programs have protection. The problems arise when lawyers have an idea of what algorithms are, which do not always correspond to how the term is used in practice. Constantly, we see examples of the term algorithm being used to describe the entire program or the source code, which under the wide definition is not a misapplication of the term.

It should be admitted that this conclusion is primarily of theoretical interest, since lawyers use terms other than "algorithms" to describe what is protected as "computer programs". The more practical consequence, however, is that lawyers should be cautious about describing algorithms as unprotected, because the use of the term can create confusion, for example, in communicating with clients who understand the concept of an algorithm in a different way.

An intriguing issue, however, is whether algorithms can be protected beyond the specific descriptions protected as computer programs, including as the more abstract compilation of instructions they consist of. Large, complex algorithms are built up by many small algorithms. Although such small algorithms may be considered as unprotected ideas, copyright protection recognizes that compositions of multiple ideas can be protected as long as they are "assembled in such a way that the work as a whole appears original" (Rt. 2007 p. 1329).

If the planning, sorting, structuring etc. of many smaller algorithms can express original, creative creation, there may be nothing that interferes with the protection of algorithms on a more abstract level. In this context, it can be recalled that the European Court of Justice in Case C-393/09 (BSA) has stated that the EU directive on legal protection of computer programs protects "the expression in any form of a computer program which permits reproduction in different computer languages" (my emphasis). If a program is to be translated into a new programming language, it can be argued that it is the "algorithm" that is imitated or copied. Pure translation of the source code will rarely yield successful results, since programming languages often have little in common and cannot be compared to natural language translation. The translator will therefore have to extract the entirety of the program's structure - the compilation of the instructions - and try to imitate the structure of a new programming language.

There are many reasons that argue against a copyright protection of algorithms on a more abstract level. The definition problems themselves are enough to create scepticism for such protection, and the originality assessments are likely to present difficulties. In addition, copyright protection is not very predictable, as copyright protected works are not being registered, and the long protection period may seem inadequate in light of the rapid technological development. However, I believe that the protection of algorithms as compilation works cannot be depreciated, and if the Court of Justice considers the issue of alleged infringement of translating programs in the future, this is something the undersigned will follow with great interest.

The article is a summary of the master thesis "Copyright protection of algorithms" published in spring 2019. The thesis is publicly available.


* The definition is the basis for the thesis. It is based on a comparison of several sources, including mathematical and technical literature, as well as dictionaries in several languages.