Richard Posner, a very influential US judge, has written an article about the current patent system’s problems and their causes. The article provides useful support for many software patent abolition arguments. He unfortunately closes with suggestions which are unhelpful or even counter-productive (with the exception of his suggestion of a shorter duration for some domains, which I discuss below). In particular, his suggestion of giving more resources and power to the patent office would be a catastrophe. What we really need is to rein in the patent offices and remove their power to grant software patents. (And this would have the side-effect of reducing their workload and thus easing their resource problems.) Here’s his article: Why There Are Too Many Patents in America.
It’s a pity. The build-up of his argument seemed to be leading straight to a conclusion that patents shouldn’t be granted in domains where research and development are cheap (compared to pharmaceuticals):
the need for patent protection in order to provide incentives for innovation varies greatly across industries. […] Most industries could get along fine without patent protection. […] I would lay particular stress on the cost of invention. […in industries where…] the cost of a specific improvement may be small, and when that is true it is difficult to make a case for granting a patent. The improvement will be made anyway, without patent protection, as part of the normal competitive process in markets where patents are unimportant.
Framing the discussion in terms of "industries" also propagates a bias by ignoring the problems caused to developers who are not part of the relevant industry. While it’s true that large scale production of pharmaceuticals and cars is exclusively done by industries, software is developed by a mix of industry, hobbyists, user communities and people in other domains who develop software incidentally as a means to getting their other work done. And it’s the non-"industry" developers who are most vulnerable when attacked by a team of lawyers. A good use of "industry" could be: patents should only exist in domains where production is exclusively an activity of industry.
The first measure Posner mentions for alleviating the problem is to reduce the patent term for inventors in certain industries. If he means reducing the term to 3 or 5 years for patents on software, then that could indeed be a big win. To do this, a government first has to declare that software ideas are not eligible for normal patents (the kind defined by TRIPS) and then create a new type of short patent. However, the examination procedure would have to be much faster, which brings a risk of much lower standards and thus an explosion in the number of software patents. Part of the benefit would also be undone by software monopolists adapting their tactics and updating their media formats and communication protocols more frequently so that the formats in wide use are always under patent control.
In summary, if this gets handed to us we should be very happy but it’s not a complete solution, it’s not easy, and in terms of effort it’s a detour rather than a stepping stone to where we have to go.
Posner’s other suggestions are less hopeful. He suggests getting rid of jury trials, and better training for judges. That might help invalidate more of the wrongly granted patents, thus reducing the number of patents somewhat, but that won’t change much (until patents on software ideas get classified as wrongly granted). Take the very popular H.264 video format for example. The MPEG LA cartel manages a portfolio of 346 patents in the USA alone which it claims are necessary for an implementation of that video format. Reducing that number by 50% would change nothing for developers of video software. It would still be covered by a big thicket of patents.
Worse, he then suggests giving more resources to the patent office and giving it the additional power to hear patent infringement cases. The patent office has a financial interest in granting as many patents as possible, so increasing their resources would probably result in an increase in patents granted. Giving the patent office the competence to hear infringement cases would mean that disputed patents would be reviewed by the same biased organisation that granted them in the first place. This would be disastrous.
That said, it’s encouraging to see that he has a good understanding of the roots of the problem, and that he agrees that different domains should be treated differently. The US Supreme Court’s Bilski and Mayo rulings also show that certain judges understand that something has to change. We have to keep explaining that abolition is necessary and smaller solutions just don’t work.