Software patents from a musician’s perspective

A GNU/Linux patent lawsuit has been filed against Red Hat/Novell, marking the first patent suit against the operating system.

If you dig through the slashdot link, you’ll see that there was a lawsuit filed against Apple a few months ago on the same patent, that the patent is a bit vague and that there is evidence of prior art.

Lots of other people with more knowledge about the legal issues involving have been posting about this. Mark Shuttleworth, of Ubuntu fame, responds to recent Microsoft fear mongering about GNU/Linux patent suits by pointing out that Microsoft itself is also a victim of the current patent system. I don’t feel that I have the knowledge or feel the need to make any specific commentary, but I would like to share my view on software patents in general, for whatever its worth.

I was recently reading a transcript of a speech [here’s a similar transcript] given by Richard Stallman on software patents. To me, the most compelling argument against software patents is based on the fact that software involves combining mathematical objects, rather than physical objects.

“The result is that software is really different from other fields, because when we are working with mathematical stuff, designing something is far, far easier. The result is that we regularly make systems which are much, much larger… we are in a system where one product involves many, many ideas that could be patented already.”

This is contrasted to the “one product, one patent” notion which corresponds to physical objects. Software developers don’t have to push atoms, but chemists, for example, do. So with pharmaceuticals, when a given chemical formula is patented, the patent only ever covers one product. Software is on the opposite end of the spectrum, where a single program can intersect with many patents.

To take the example of music, patenting an instrument would be more like patenting a drug – a physical object. A patent for a violin (if it were a new invention) would cover attributes like the number of strings, the dimensions, etc. However, patenting software is much, much more like patenting chord sequences or melodies. Chord sequences and melodies are immaterial, mathematical (in a sense) objects, like the ideas contained within software patents. Any single song contains many different chord sequences, melodies and harmonies which have probably been used many, many times before. But since songwriting does not involve music atoms, it’s much easier to create something that is very, very complex and as a result, is the combination of tens or hundreds of smaller ideas. Just like writing software.

Chord sequences are the fundamental building blocks of music, just as algorithms are to computer science. You study music theory and learn that certain chord sequences and cadences can be used to evoke certain emotions or have a certain musical effect, in the same way that algorithms that yield a certain result or have a certain effect on the input. To grant patents for chord sequences or algorithms is to deny musicians and programmers the ability to use fundamental theoretical ideas in their respective fields.

That argument resonated the most with me (pun intended). To me, it is patently obvious (sorry, I’ll stop) that things like melodies and chord sequences should not be patented. As a result, other similar mathematical ideas (ie. algorithms, software features), should also not be patented. Patents may be effective when one patent corresponds with one invention, like when you are dealing with material things. Its efficacy is questionable at best when dealing with algorithms or chord sequences.