Who owns software written by software (part 3)
March 31, 2010
In the two previous posts, I discussed that
- there is a new development: software that is developed automatically, by other software, based on human instructions, but without humans writing any actual code;
- if we want to know how owns that code, we need to look primarily at copyright.
In the mean time, also, my full legal article on the issue has been published in Computer law review international.
So, what is the copyright status of software written by software?
“Written Code” (code written by “Writing Software”) is, technically, like other code. It is a human readable expression of the instructions that will be compiled into machine language by a compiler.
However, the Written Code is generated by another piece of software (or artificial intelligence, if you will), on the basis of general instructions given by the human user of the Writing Software.
What does that mean? It means that the human user of the Writing Software will formulate general instructions (like defining objects, describing relationships between objects, or drawing an algorithm). Based on those instructions, the Writing Software will generate (“write”) the code.
The distinction between the instructions and the code is a very important distinction – it will be instrumental in determining whether Written Code is protected by copyright, the topic of this blog.
It is instrumental because copyright is actually a rather awkward format of Intellectual Property protection for software. That is because copyright, created way back in the 19th century as a legally imposed monopoly on reproduction and distribution, was designed to provide protection to expressions of creativity. And software code is of course an expression of technical efficiency, rather than creativity.
However, in the seventies, when the IP protection of software was discussed, it was decided to opt for the use of copyright as the IP tool for protecting software, mainly because patents were even less appropriate, rather expensive and mostly inefficient for the purpose of protecting software, and because nobody wanted to devise a new, software specific, IP right.
The creativity was deemed to be in the original creation of the code.
Created, of course, by humans – except that that assumption was never really stated as such. But, it solved the creativity threshold problem, and therefore, copyright could apply to software.
The consequence of that choice, though, is that the distinction between an idea and its expression, which is so important for copyright protection, also applies to software copyright protection.
The best way to explain the distinction between an idea and its expression is as follows: the idea “My Baby Left Me This Morning, I’m Feeling Blue” is not protected; however, the thousands of blues songs based on that idea, to the extent each of these songs is not a copy of another song (there are specific rules for that, ask Madonna) are each individually protected by copyright. So, you can always write a new song on this theme, as long as your song, the specific expression of the general idea, is sufficiently new or original.
The analogy with software goes more or less like this: functionality in software (say, an automated back-up function) is not protected, but the code in which it is expressed is. It’s not a perfect analogy, but it is sufficient for our purpose.
Because, as we have seen, the user of Writing Software only expresses generic instructions “ideas”, and the Writing Software writes the code, the “expression”. It would be as if you instructed a software program to write a song about the theme “My Baby Left Me This Morning, I’m Feeling Blue”, and it came up with a twelve-bar blues in Bb, with a nice melody line that could have been played by Eric Clapton and some awkward, unhappy lyrics…
Note however, that the dividing line between “ideas” and “expressions” is not a clear-cut transition, there’s a lot of uncertainty where one ends and the other begins, even in very specialized legal literature.
Coming back to our software: what that means is that, for Written Code, it is the Writing Software that creates the bit with originality (the “expression”), that is supposed to be covered by copyright. The human originality is limited to the instructions (the “ideas”).
The interesting conclusion is that both elements that are involved in generating Written Code would fall outside the scope of copyright protection.
The first element – the instructions – falls within the category “ideas”. Look at the European Directive on Software Protection: underlying algorithms clearly fall outside the protection for software.
The second element, the code, contains no human creativity, and fails to pass the originality threshold required for copyright protection.
Hence, no copyright protection for software written by software.
Is that a problem?
Not necessarily so – I’ll talk about that in my next blog.
Joren De Wachter
