Apologies for the length of this post. There's a lot to explain.

I don't mention here who my employer is, but I think it incumbent on me at this point to say that this post (in common with any post on this blog) is my own opinion and does not represent the opinion of any company I work for.

I say this because I want to write about patents, and my employer likes patents. I do not.

I'm going to try and write this carefully, because intellectual property issues make me very angry very fast. At least you won't see the foam hitting the screen...

To be honest, I rather like the original idea of patents: someone invents something, and then can protect that invention for a certain amount of time to give them a window to develop and sell it. The bargain is that the invention is published in the issued patent, so that when the patent expires it is in the public domain. This is why, for example, the recipe (or recipes) for Coca Cola has not been patented: the the Coca Cola Company did not want to publish it. Then again, it's just a soft drink - it's not going to be used as the enabling technology for a whole new social infrastructure.

The problem I have is not with patents per se, but with the way they are granted, used, and abused.

invention

patents (and I am talking specifically about US patents here, although much of this is likely to apply in Europe eventually) are granted by the patent office if they are both novel and non-obvious. The problem is that the standards for either of these seem to be pathetically low - "novelty" seems to take little account of actual prior art, and "non-obvious" is far too subject to being argued. The result is patents like Amazon's "1-Click shopping" patent which may have been novel at the time it was filed (although prior art has been found) but certainly strikes me as being deeply obvious.

There is also the small matter of whether software and business methods should be patentable which, in the US, they are. I can't speak to business methods, but I have come to realise that most of my objections to patents on software (with the exception of term - see below) are that the standards for novelty and non-obviousness are too low. I would still rather return to a situation where patents require a physical artifact, though.

development

I think this is the part about patents which makes me most sad... there is no requirement upon being granted a patent that the technology described in it be developed by the patent holder. Thus there are three rather disgusting practices which arise:

1. predatory patenting - this is where someone will think up something patentable, patent it, and then wait for someone else to invent the same idea. When that happens, they slap the new inventor with demands for licensing fees. Of course, this demand will not be issued when the new product first comes to market - it is usually issued once a fair number of units have been sold, in order to recover retroactive licence fees.

   In fairness to the US PTO, the patent procedures used to be even more favourable to this practice. It used to be that a patent's term ran from the point where it was issued, but it now runs from the time when it is first filed. This is because there was a particularly aggressive predatory patenter who would file a patent, but then keep changing it during the filing process in order to postpone the issue date. He would only stop making changes when someone else started a business using the idea in his (as yet unpublished) patent so as to have the maximum term available to extract licence fees.

   There are variations too, such as a company getting a standards body to mandate a standard which relies on its patent, but the "Surprise!" patent is probably the one that bothers me most because it is so hard to defend against even for large companies.

2. defensive patenting - the point of a patent is that a business can use it to prevent other companies competing in the space covered by that patent. The problem is that with patents being handed out without, apparently, being tested for validiity then practically anything you can develop stands a fair chance of having been patented already.

   The standard method among corporations of managing this risk is defensive patenting: a company will maintain its own patent portfolio, so that if they are sued by a competitor for infringement then they can counter-sue and say "Ah yes, but you're infringing one of our patents." This usually leads to each company licensing the other to use their patent, something which is called cross-licensing. Obviously this method is not available to any business which is just starting up.

3. patent trading - patents are considered an asset, and can be bought and sold. I'm not talking here about the cross-licensing of defensive patenting but the situation where a patent is purchased from a company (maybe because it's gone bust, maybe just as a way of getting some cash) and then it is enforced by the new owner who, of course, never invented it at all. The results can often be very similar to predatory patenting.

term

patents last a long time (not as long as US copyright of course, but that's another rant). As I write this, the patent term is twenty years from date of filing. This is probably still a reasonable term for a patent on a physical technology (although given standard consumer product lifespans nowadays I have my doubts) but is ludicrously long for software.

The reason I get so exercised about patents is that their abuse threatens the little guy - small businesses and free software developers. Consider a free software developer in particular - if you write something which incidentally infringes a patent which you had no idea existed then you still stand a fair chance of being sued for licence fees. And since patents are expected to be invalidated in the courts rather being simply not granted in the first place, then it's easy to see that he with the deepest pockets wins.

My real fear though is that some company which feels threatened by free software might use patents as a weapon to squash it, something along the lines of "Oh, so you can't pay? Well, stop using the patented algorithm then". This has happened already, of course: Unisys used their LZW compression patent to try and extract licence revenues from all and sundry over GIF. SCO's current action (where they claim that some of their code is in the Linux kernel) isn't a patent issue, fortunately - I say "fortunately" because if it had then their case might have some weight, whereas my understanding of the situation is that their arguments appear to be largely frivolous.

Which leads me to the story which led to start this post, the news that Microsoft's patent on the FAT filesystem may be challenged.

I'm just really really glad that their is an organisation which is looking to undermine bad patents.

It would be better if the patent system could be fixed, though. For instance, adding just two provisions would fix most of the bad practices I discuss above:

1. require patents to be developed - if a patent holder cannot demonstrate that they're actually developing and/or selling the patent being infringed, then they can't claim to have been damaged by the patented idea being exploited by someone else.

2. require patents to be defended - if a patent has not been defended previously, then a patent holder cannot claim to have been damaged.

   This might actually be quite a dangerous change, of course, if the standards for patents to be granted are not raised, but at least we wouldn't have a repeat of the Unisys LZW patent fiasco.

... but really the standards for novelty and non-obviousness need to be raised substantially, and that's something that can't be fixed without either funding the patent offices better, or opening up patents being processed to public review.

Posted by Dunx at December 10, 2003 10:44 AM