Specific Description: Defining the patent line with a copy test.

Recently on Groklaw the question was forwarded "Why should software not be patentable?". Gene Quinn was the original asker, and I have to agree with him that the question needs answering. The article got me thinking even more deeply than usual about the issue, and so I'm sharing those thoughts in the hope it will a provide new idea or angle to those with relevant legal expertise.

A disclaimer of my own to start off. I am not a lawyer, and while I do write some code for a living it's not the main part of my job and the code will need to be written regardless of if it's considered to be patentable or not. Given that Gene has intentionally asked the question openly I'm assuming that he's not looking for a legal opinion, but a logical one (not that I'm suggesting the two are mutually exclusive, simply that someone without legal expertise will invariably look at the question differently from some with legal expertise). There are numerous theories out there which are often supplied as arguments on why software shouldn't be patentable, and I do not wish to rehash those.

Instead I'd like to give a single answer, and the best way to answer why software shouldn't be patentable is to draw a line distinguishing at what point something shifts from an asset to an invention. I'd like to propose a concept I've not heard before when deciding whether something is considered patentable, that of Specific Description or a copy test.

The idea is simple. If a proposed patent of an asset could be given a description specific enough that the description, a translation of that description, or the output of that translation became the asset then the item should not be considered patentable, but instead be considered within the domain of copyright.

This should also apply to components of something for which a patent is being applied. So that any element which meets this test cannot have its details patented, but is covered with respect to its use within the asset for which a patent is being applied (assuming that the asset otherwise meets the requirements for patentability).

Gene has expressed the (presumably legal) opinion that copyright is insufficient for computer software citing the example of rewriting code. I find this amazing. I couldn't take a novel change the characters names and genders then publish the work as my own, so I'm surprized to hear that the equivalent in computer code is ok. If this is the case though, I'd suggest that's a deficiency with copyright law rather a case where a patent is the appropriate resolution.

One final note with regards to Specific Description and that is a look at how it would work when applied to various assets. Take a car engine, no matter how much detail we put into describing it we never end up with a car engine. A text only book is a simple case, clearly you can write word for word the contents of the book and so therefore it fails the Specific Description test. Music, even a recording, can be respresented in a form which can be translated to give the recording itself. The same applies for imagery and therefore as a combination of the previous two movies. These are all things which have long been established as the domain of copyright law, not patent law. Suggesting perhaps the Specific Description test is similar to the thought processes of those who created patent and copyright law. There is something else very common in society which fails the Specific Description test. Software.