Stuart Jarvis A troll's eye view

This started as a comment on Ade’s blog post about Hugin, but I don’t think comments should be bigger than the original blog post, so I’ve moved it here instead 🙂

Ade makes some interesting points about the GPL, particularly the opportunity to exclude certain jurisdictions where the use of the software would infringe a patent. He also raises the question of whether distribution of source code – as opposed to binaries – can be counted as infringement and reflects on the trend to explicitly claim for a medium containing code that would cause the invention to be ‘realised’ on a computer.

Background

If you’d like an elegant and (generally) very well argued primer and history of what can be patented in UK law see this judgement (from 2006, things have changed a bit since then). The judge in that case, Peter Prescott, was actually only standing in and therefore wasn’t really expected to take a fresh look at it as he did and go against patent office practice that had been making software patents a bit easier to obtain in the UK up until that point.

He’s something of a personal hero of mine 😉

Why I’m writing about this

I guess the first thing to say is that I’m no patent lawyer, this is not legal advice and my knowledge is limited to UK and (a little) to EU patent law. The knowledge I have was gained through working as a patent examiner (one of the people who decides whether or not a patent is granted on an idea) from 2004-2007 at the UK Intellectual Property Office. During the last year of my time there I was in the department working on ‘unpatentable’ inventions: software patents, business methods and the like.

This area of patent law moves quite fast in varied, contradictory and nonsensical directions and I really have not been paying attention to it since I left almost three years ago.

Who can infringe? (Why claim the software medium?)

Within UK law, at least, there is the idea of primary and secondary infringement. The primary infringer is someone who does something directly reading on to the claims of a patent. So if the patent claims only the method of doing something and your software performs that method when run then the primary infringement would occur only when the method is actually performed (i.e. when the program is run). In that scenario, if you were selling the software then only your customer would be a primary infringer when they used the software. However, as a supplier of the means to infringe, you could be open to a claim of secondary infringement. The original logic of this was – I believe – to allow the owner of a patent to sue, say, the British importer of an infringing device rather than having to pursue a foreign manufacturer abroad.

However, pursuing a case for primary infringement is easier and so, as Ade noted, many patent applications will also include a claim along the lines of:

A storage medium containing code that, when run on a computer, performs the method of claim x

This would allow them to go after anyone supplying the program on a disc as a primary infringer.

There was, briefly (I think it got overturned) a court judgement in the UK that ruled that any claim to a program on some kind of medium was a software patent as such (well, obviously…) and not permissible under UK law. I had great fun in my last few months as a patent examiner rejecting loads of applications simply by saying:

Claim x is a claim to a computer program on a storage medium. The invention is therefore clearly a computer program as such and cannot be patented.

Applicants and their patent attorneys didn’t much like that.

So does source code infringe?

I doubt it. There may be an argument for secondary infringement, but that gets in to all kinds of difficulties as there are various defences such as the research defence (pure academic research and the publication of methods arising from it are immune from patent infringement) or freedom of speech (anyone remember the haikus containing the deCSS code?).

Also few people make money from distributing source code, so where’s the point in suing them? It’s easier to sue the end user who directly infringes (and doable if they’re a big company: SCO tried it with Daimler, albeit in a copyright rather than patent case) or someone distributing the binaries on a commercial basis (if the patent owner remembered to claim for software discs too).

Disclaimer: as far as I’m aware, the above has never been tested and a lot of it might not apply outside the UK anyway.

Why does this matter to KDE?

Most of the software patents that came to me originated in the US and were fairly easy to dismiss under UK law. However, there was a patent application that came to me before I worked in the unpatentable section that could have had a direct impact on Amarok.

I don’t remember the details of the patent in question (and probably couldn’t reveal them anyway as I don’t want to mess with the Official Secrets Act). However, it was for a hardware audio player and one of the features it claimed had, I knew, been in Amarok for a little under a year, but before the filing date of the patent. It was my opinion that Amarok would have infringed had the patent application been granted. It took me a couple of days to find the disclosure of the feature in the Amarok infrastructure (trawling wiki page histories and forum posts), twice as long as the target time allotted for processing that particular patent – and I had to hit that target to get promoted.

The point is that the Amarok wiki and forums are not somewhere we routinely searched for prior art and had that patent application been passed to someone else who didn’t know (or care) that the feature was already in Amarok then the prior art probably would not have been found. I would not have been criticised for not finding it and it was against my interests (other than as a user of Amarok) to spend the time doing so.

The problem is more than software patents

As illustrated by the Amarok case, it’s not only software patents that can be dangerous. The ‘invention’ there was obvious – to you or me – but it probably wouldn’t have been patently obvious, by which I mean the the level of invention required for getting a patent is low.

There is also a problem not only with a lack of time to do things properly, but also a lack of technical knowledge. I examined applications concerning JIT compilation, despite having no knowledge of programming. I examined applications for magnetic hard drive read heads without having the faintest idea of what would be obvious to someone in the industry, or really even really which features were important.

I always had the impression, since we saw the results of prior art searches from other offices, that the European Patent Office were far more thorough than we were. They certainly had more time per patent application and found things in places we would never have looked. We tried not to look at what the US Patent Office was up to.

I’ll finish off by quoting Peter Prescott in that judgment I linked to at the start:

Despite the prohibition on granting patents for computer programs … it is said that the EPO [European Patent Office] has granted more than 40,000 of them… From the point of view of the applicants … if there is any chance of getting such a patent it may be said to be a rational business choice to try it. If not, their competitors might … patents that are wrongly granted can be very expensive to challenge, and perhaps beyond the means or inclination of small and medium enterprises. An accumulation of patents of that sort … may be a serious barrier to entry.

The only safeguard against that wrong – and it is a wrong – is the vigilance of the Patent Office. When I was a Patent Office examiner … we sometimes granted patents that we shouldn’t, but did it anyway because we thought the Patents Appeal Tribunal would not support us.