A few days ago I ranted about thoughtless reportage - basically reproduction of press releases masqueradings as news. Jumping now to the far end of the spectrum I'd like to point to a thoughtful piece written by an intelligent man, even though I happen to disagree with many of his points.

To wit: Paul Graham's essay on "Are Software Patents Evil?" Graham is a hacker, entrepreneur, philosopher, and yes software patent holder. The essay derives from a talk he gave at Google on the topic. Graham's essay draws mostly from his point of view as adviser to several start-ups, all of whom care about patents. He reasons that patents aren't as much of a problem as people think, mostly because of what's in peoples' economic interests.

He asserts that people opposed to software patents must then be opposed to patents in general. I think this is false. Software has a number of properties not shared by physical processes and instantiations that make patents problematic. These include multiplicity of representations (source language, intermediate language, assembly language, etc) being equivalent, an extreme flexibility on the topic of what is process and what is the data operated on by that process, and so on. Software is neat and cool and (unlike some Copyfight readers) I think it is deserving of some intellectual property protection for authors who choose it. But I think patents are a bad tool for it, and they're being badly abused. Thus I oppose software patents but I'm fine with other uses of patents.

Graham attempts to address the issues of abysmal patent quality, noting that the intersection of software and government almost never turns out well. True, but not the whole story. Prior art, for example, is pathetic in software patents compared to other patents, and the USPTO is absolutely criminal in its failure to enforce prior art (and associated nonobviousness) requirements. Given the state of the law and practice, Graham argues that companies wishing to compete must apply for patents. He makes the analogy to a hockey team that doesn't check opponents - to which one might counter that if you skate fast enough you can avoid checks and not have to do them yourself. So much for the power of analogies, but Graham's point of view is held by quite a few folk.

Graham blames companies such as Amazon not for getting patents - they're simply maximizing their take from the current free-for-all system - but for enforcing silly patents. Though he doesn't use these terms, he describes the classic "sword" and "shield" approaches to patents. Unlike a trademark, a patent holder is not required to vigorously police any potentially violating activity. A patent holder can simply sit on the patent, using it as a "shield" to protect his own investment. Amazon used its "1 click" patent as a "sword" to attack Barnes & Noble for having a similar checkout system.

Now the question is: who actually sues whom over patents? Graham points out that much more often it's small companies suing larger ones. Suing startups is often a waste of resources because you can't get money out of them. If they're a threat you buy them, or at least buy the intellectual property you care about. Most startups are eager to strike licensing deals for their IP with big players and patents give them leverage in those negotiations. Or if you're big and mean you lock out the startup and its novel technology. A patent for something on a PC is worthless if you can't get that thing to run on a WIndows OS.

Graham also injects a heavy dose of fatalism (his own word) into the picture. Basically, once you're big enough to be successful and have money, you'll get sued whether or not you have patents. But you shouldn't worry about it, because it shows you're worth noticing. Being noticed means you're a candidate for buyout, which is the exit strategy for most startups. And if you're not going to be bought out then you need to go to the negotiating table with the existing big players, all of whom have big patent portfolios. Your portfolio in this view is table stakes - it gets you respect and a seat. What's actually in the portfolio isn't really that relevant; what's relevant is whether you can strike the kinds of mass cross-licensing deals that allow the big players to continue in business without having to worry about every clause of every patent in every competitor's portfolio.

I view this as a kind of insanity, much like the nuclear Mutually Assured Destruction phase of the late 20th century was a shared insanity. Armed truces aren't inherently stable and are always unfriendly to new entrants who can upset delicate balances of power. That's not healthy for software in general. Graham would probably agree, but he points out that for all the big press over certain cases, patents play a much smaller role in software than they do in other industries. For example, other industries regularly see large-scale lawsuits over patents well before products are released - think of the drug industry for example. That's exceedingly rare in software, and Graham gives some reasons.

First, the complexity of software renders most patents trivial by comparison. I may have a patent on an order checkout method, but that's only a tiny fraction of what's needed to write, run, maintain, enhance, and grow an online store. Second, software has a tradition of the young and fast displacing the old and wise. Patents are a way of respecting "how it was done before." Software startups take pride in DISrespecting that kind of thing. Third, the nature of the software business is that we inherently engineer around. Graham says that startups rarely try to take on big companies head-to-head - instead they try to engineer paradigm shifts in which the old way is irrelevant, and so are the patents that protect it.

Graham gives a lot of weight to "public" opinion, by which he means hacker opinion. Here I think he's being too optimistic. Microsoft has done a ton of evil things and not suffered much from public backlash, even when they were being flayed in the mainstream press. Sure, principled hackers will avoid evil companies and find better jobs elsewhere, but the globalization of software means that for every principled American hacker there are 10 or more overseas developers who may be less principled but can still produce damned fine code. And maybe cheaper, too.
Sure, he's talking to Google so he has to give at least a nod to the "don't be evil" principle, but I think he's more on the mark with earlier points.

Graham spends some time making a point I do agree with, which is that in the absence of patents (which force at least some kind of disclosure) people will attempti to protect things via secrecy, usually enforced with heavy-handed laws. He notes that this is partly what the Cartel are trying to do, and he's right. He also notes that this encourages some of the worst practices in business.

Graham concludes by stating that he (and his co-investors) advise startups to ignore patents. And they put their money where their mouths are.by taking personal financial risks in such startups. I have to respect that, no matter what else I might agree or disagree with him about.