We know that patents are bad for the software business. The whole reason patents exist is because it is thought that giving a time-limited monopoly to people who invent things promotes progress by encouraging, well, people to invent things. This doesn't work in software because:
  • patented software ideas are generally obvious to practitioners of the area the patent is about;
  • the length of a patent monopoly (17 years) is way out of proportion with the speed things move in software;
  • barriers to innovation in software are intrinsically low to begin with - anyone can innovate and no incentive in the form of a time-limited monopoly is necessary to encourage people to do so;
  • arguably most potential innovators in software are individuals and small companies, who can hardly use the patent system because they can afford neither the time to file the patents, nor the money to fight infringements;
  • progress in software is intrinsically cultural and incremental, not epochal - people gradually improve on other people's ideas, nobody does (or has to do) research and testing for 10 years, like in the pharmaceutical business, before bringing an idea to the market.
All of these reasons combine to make the idea of patenting in software very, very harmful indeed. Instead of being used as an incentive, patents are used as:
  • weapons for big companies to fight each other with;
  • weapons for big companies to destroy potentially threatening small competitors before they grow;
  • leverage for patent sharks (like this bastard) who use sleeper patents that no one ever knew of to extort enormous sums of money that they did not even earn and which no one owes them.
So, at least as far as the software business is concerned, the patent system is more so a vehicle for power play and extortion than for anything good, and should be banned.

A valid question remains, though, whether the patent system is still good at solving problems in other businesses, such as pharmaceuticals. In medicine, it can take an outlay of a billion USD or more to bring to market a useful drug. Reasoning goes that no sensible pharmaceutical company would ever engage in this endeavor if it weren't for the patent system, which helps companies recoup the cost by keeping competition at bay for a few years. (The patent lasts longer than that, but it only prevents competitors from using the very same compound, not from using a similar one; and competitors generally need a few years to bring to market an alternative compound with similar effects.)

Joseph E. Stiglitz recently wrote this article listing some of the demerits of this use of the patent system in medicine, and proposing that the system be replaced with large (state-sponsored?) rewards for desirable medicines. I think this is a good proposal, its only fault being that the reward system would need to be competently administered, and I guess we know how competent administrations tend to be. Nevertheless, if the state was bold enough to offer really good rewards for developing valuable cures - e.g. on the order of $2-10 billion USD - I think the idea would work.

An optimal solution would be one that neither grants a monopoly, like patents, nor requires a central administrator, like the reward concept would. Something that's self-managing and yet not prone to abuse. That would be a trillion dollar idea. If someone has it.

Care to patent it?