jump to navigation

Are algorithms copyrightable ? December 4, 2006

Posted by Imran Ghory in Copyright, Software development.

(disclaimer: I am not a lawyer, this article is personal opinion)

If I was asked a few days ago “can an algorithm be copyrighted ?”, I would have replied with the canned answer “no, of course not, copyright protects the expression of an idea, not the idea that’s expressed.”

However since reading Hart & Fazzani’s Intellectual Property Law (part of the Palgrave law series). I was no longer convinced so I decided to google to see if anyone else had discussed it. But other than the canned answer I gave above I couldn’t find any firm evidence supporting the uncopyrightable status of algorithms. And during my searches I found the submission policy of ACM Algorithms which clearly indicates that they at least believe some algorithms to be copyrightable.

Generally speaking if you translate a program from one programming language to another then you have to have permission from the original author of the program (in the same way you would need permission to translate a book from say English into Spanish). The Palgrave book specifically gives the example of translating from FORTRAN to COBOL as being a type of “adaptation” that requires permission. In Canada Apple Computer Inc v Mackintosh Computers Ltd. came to a similar conclusions.

So why should translating an algorithm from pseudo-code into an programming language be any different, I’m sure most implementations of binary search are based on text-book pseudo-code rather than derivation from first-principles.

The obvious counter-response for this has been that if an idea can only be expressed in one form then it’s not copyrightable. However in the UK at least this doesn’t appear to be true if the idea is sufficiently complex. In the case Ibcos Computers Ltd v Barclays Mercantile Highland Finance Ltd (1994) which is one of the most important software copyright cases in the UK it was decided that “copyright cannot prevent the copying of a mere general idea but can protect the copying of a detailed idea” and this is a decision which has been referred to and affirmed in a number of court cases since that time.

And “detailed idea” is a concept which will almost certainly apply to most modern algorithms. As far as I know no-one’s ever taken an algorithm copyright case to court (possibly because such court cases over algorithms have traditionally been in the US where patent law can be applied instead of copyright), but it could just be a matter of time before it happens and possibly opens up a huge can of worms.