I received last week an invitation to review a patent application, under the pilot (and strangely-named) peer-to-patent scheme whereby anyone who might be thought to be interested is invited to comment on patent applications during their examination by the patent office. The specific patent in question is at http://www.peertopatent.org/patent/20100250576/overview: it's a Microsoft application for an optimization technique for SPARQL queries.
First question I have to ask myself is, should I take part? Intrinsically, it feels like sleeping with the enemy. It seems a very strange process, exploiting the willingness of the intellectual community to share knowledge in order to improve a process whose very purpose is the antithesis - to assign monopolies of knowledge. Do I want to participate in a process that might encourage the authorities to award fewer bad patents? Or do I think that bad patents are ultimately a good thing, because they will inevitably, in time, bring down the whole system? There are issues of tactics here, and issues of high moral principles. I'm going to compromise: I will respond telling them I refuse to take part in the process, but they can read my blog here for my views on the patent application if they want them.
Another little tactical point: apparently in the US, software engineers are routinely advised not to read patents, on the grounds that the fine for infringing a patent is trebled if you knew about it. I don't know if that's really true - I find it hard to believe that in the land of the free, you can be fined for reading information that's in the public domain. If it is true, it makes me proud to be non-American. Anyway, I can hardly claim ignorance of a patent that I've been invited to review, so I'll take the plunge and click on the link; fortunately it turns out to be of little relevance to anything I'm doing in my products.
First observation on the patent application (and others I have seen recently) is that it's much more readable than the applications of ten years ago. I don't think it uses the giveaway patent-speak word "plurality" even once. It's written by a software engineer, not a patent-lawyer trying to make it unreadable to other software engineers. That can only be a good thing. It also hardly tries to disguise the fact that it's a pure software patent. There's a bit of nonsense in the claims about patenting the media on which the software is written, but no serious pretence that the patent applies to a machine rather than to pure algorithms. It seems that in the US at least, you no longer have to pretend that it's not about software in order to get your patent application through.
What about the substance? I'm told that the two ways a patent claim can be challenged are on the grounds of "obviousness" or on the grounds of "prior art" (i.e., someone else has done substantially the same thing before). I'm afraid I really don't know enough about how these tests are supposed to be applied. Wikipedia tells me that "obvious" means "obvious to one with ordinary skills in the art" and interprets that as someone with "the normal skills and knowledge in a particular technical field, without being a genius." Well, if that means obvious to the average programmer, then the stuff in this patent certainly isn't obvious. If it means obvious to the average person doing research or implementation in logic-based query languages, then it almost certainly is. I'm not working wit logic-based query languages, but it looks to me like the kind of solution I would immediately come up with if I were faced with this problem. The "naive solution" to the problem that they claim to be improving on is a strawman; you would have to be very ordinary to think that was the best you could do.
As to prior art, again, I don't know enough about how this is evaluated. Does someone have to be tackling exactly the same problem and coming up with exactly the same solution to qualify as prior art? Or can one simply point out that this is just a specific example of a general approach that has been used by competent programmers for decades?
This patent application points to all that is wrong with the patent system. It's a simple solution to a simple problem. It's impossible to say whether it's original, and it's impossible to say whether it's obvious. But what one can say with certainty is that there is no way it is in the public interest to give Microsoft a 25-year monolopy on solving this particular little problem in this particular way. Anyone who thinks doing that is going to encourage innovation needs their head examining.