Daniel Tunkelang wrote an interesting post about the merits (and lack thereof) of software patents. The basic argument is that software patents are overly broad, hard to defend against, are not central to most software companies’ businesses, and only truly benefit attorneys and patent trolls.
I won’t pass judgment on the industry as a whole, as I am sure there are many legitimate cases, but I will point to some evidence to suggest that the patent mechanism, in its essence, is the wrong mechanism for protecting these ideas.
The evidence in question is the manner in which claims are written:
- A method and system for…
- Apparatus for…
- A computer readable storage medium embodying a set of instructions…
- etc.
And of course there are the ubiquitous descriptions of circuits to implement high-level abstractions such as video segmentation or document clustering algorithms.
The point is that all this language is tied to the hardware rather than the software being patented, because the raw algorithms themselves are too abstract to be patented. That this language is necessary seems like organic evidence for the inappropriateness of software patents.
I’m all against software patents, but when it comes to the patent language I have a doubt. As far as I know, patents play quite a big role (and probably much more positive one than in software) in pharmaceutical industry, but the language you displayed seems equally unsuitable for pharmaceutical patents. The language is just a legacy of the past, I think.
I am not familiar with the language of pharmaceutical patents or how they describe the novelty and non-obviousness of the claimed invention. I looked at a few pharmaceutical patents, and saw terms such as method and compound, which seem closer to the gist of the invention. But I am not an expert on this, whereas I have considerable experience with software patents.
The language of software patents seems to be designed to circumvent the restriction against patenting formulas or pure algorithms by tying them in a meaningless way to some tangible hardware. Yet the hardware has nothing to do with novelty or obviousness. One can argue that hardware of some sort is required for utility, but beyond that the language seems intended to circumvent rather than to clarify.
Like Dimitry, I’m not sure the language argument proves more than that the language of the law moves slowly. In any case, I see the case for the pharmaceutical patents vs. software patents as much easier to argue in terms of the incentives for innovation.
As I understand it, pharmaceutical companies would not invest in research without the benefits of patent protection, since they need the temporary monopoly to recoup the cost of the years-long, high-risk R&D to develop their new products. This simply isn’t true for software companies. If there were no software patents, I’m convinced that software companies would continue innovating as they have been, only with less money going to the lawyers and hence more to spend on R&D.
I agree with the distinction between SW and pharma, but I don’t think the language is just a historical accident. It seems to me that otherwise SW is just not patentable, and thus this represents additional evidence against the premise that software can be patented.