from Hacker News

Patents on Software: A Nobel Laureate’s View

by ckuehne on 10/16/12, 2:04 PM with 52 comments

• by jandrewrogers on 10/16/12, 3:14 PM

  The argument made is not unique to software or computer algorithms. Chemical process patents are identical to computer algorithm patents in this respect, just replace bits with molecules, and are among the oldest patentable subject matters. Long, complex sequential algorithms are a feature of many engineering disciplines.

  Chemical engineering, to use that example, designs elaborate and complex dynamic systems by chaining together abstract chemical algorithms. Each one of those little algorithms is subject to both patent and copyright. Like with software most of the commonly used algorithms and clever hacks were either never patented or the patents have long expired. It is only on the bleeding edge that some chemical algorithms are under patent; as with computer algorithms there are an unbounded number of potential algorithms but some are more efficient than others. Specific implementations are still covered by copyright and are widely licensed (as libraries).

  Most of the nominal specialness attributed to software as a domain for intellectual property does not really exist. Yet the rarely questioned assertion that computer software is special in some way has created a dearth of comparative studies that would likely be valuable from both a theoretical standpoint as well as a practical policy standpoint. Either these other areas, like chemical processes, are equally broken at a fundamental level and the scope should be extended beyond software, or there are differences in implementation across otherwise equivalent domains and we should borrowing from the better implementation. It seems like an oversight that no one is attempting to do either.

• by PaulHoule on 10/16/12, 2:56 PM

  Note there's a flip side to this -- the timescale of software patents is completely wrong from the viewpoint of a company that wants to use patents honestly.

  I was looking at a face recognition patent that was filed by the US by a Japanese company in 1998 that was finally issued in 2006. Eight years is a very long time in the fast moving software industry -- even if you get lucky and your patent granted in two years, it's quite possible that your invention is obsolete by the time you get your patent.

  Given that software is so fast paced, most organizations that expect to be "practicing entities" find the patent application to be a distraction from the task of getting a competitive product in the marketplace. This is a very different situation from other fields where you really can get a patent for a mechanical or electronic thing and then have the patent as a tool for negotiation w/ manufacturers.

• by ecolak on 10/16/12, 6:54 PM

  The whole "protecting intellectual property" thing could work if it was possible to issue patents in an unbiased and consistent way but patent examiners are human beings and they do some research on the issue, read the arguments from the patent lawyers and come to a (usually biased) decision. You even see cases where a patent is rejected 9 times and get accepted the 10th time. It is extremely difficult to come up with an unbiased, consistent and all-knowing (past work) way to issue patents, I'm not even sure if it's possible at all. And that's another reason why there should be no patents issued at least in the software industry.

• by hermanhermitage on 10/16/12, 9:14 PM

  Doesn't the problem with imaginary things like gods, beliefs or intellectual property begin when someone tries to claim they are real and you must obey them?