Discussion and debate on if patents on software lead to innovation
How software patents are delaying the future
This fall, I went to Amsterdam to talk about "How Software Patents Are Delaying The Future", on a discussion panel organised by the European Patent Office. The other people on the panel were patent attorney Simon Davies, and Ioannis Bozas, a patent examiner at the EPO. The panel was moderated by James Nurton of Managing IP. Despite our very different views on the subject, we had very friendly and informative conversations before, during, and after the panel.
For the EPO, organising this debate was something of a gamble. They’re widely criticised for their practice of awarding patents on computer programs, and the debate tends to get rather heated. While I couldn’t disagree more strongly with the way they do things at the EPO when it comes to software, I give them credit for putting this debate together. It was also refreshing to hear Ioannis state clearly that the EPO grants patents on software, as long as the program makes a "technical contribution"—that’s somewhat clearer than the line about "computer-implemented inventions" we’ve mostly seen the EPO employ so far.
There were many interesting points in the debate, though, as you might imagine, consensus was in short supply. The argument that I made was that we need to encourage innovation; that patents are only one of many possible tools to encourage innovation; and that in the case of software, they’re horribly counterproductive. That’s why we need to get rid of software patents, and bring the patent system under the control of political institutions, so that we can design an evidence-based innovation policy as required by the circumstances. (See here for FSFE's work on software patents.)
The argument against software patents
I started out with a look around us: The Internet, the World Wide Web, Free Software were not only not patented; they were only successful precisely because they were not patented.
Then came the basics. A patent is basically a social contract. The inventor gives us a new idea, we as a society give her a limited monopoly to use that idea. As with any contract, we need to check whether we’re actually getting a good deal.
This contract is just a means towards a more important end: promoting innovation and, ultimately, progress.
So we need to ask:
- Do patents on software lead to innovation that wouldn’t have occurred otherwise?
- Is granting a monopoly the only way to obtain this innovation?
- By granting a monopoly, what follow-on innovation are we passing up?
Looking at the classic Bill Gates quote about how "the industry would be at a standstill today" if programmers had started patenting their software earlier, I pointed out that that’s pretty much where the smartphone industry is today. Every day seems to bring a new patent lawsuit in the area. Patents have degraded into just another stick to beat competitors with. The captains of the mobile industry can only keep their ships steaming ahead through the patent tangles by burning piles of cash in the kettles of litigation.
There are two basic disconnects that make patents so horribly unsuitable for software. The first is that patents last many years, while development cycles in software are measured in months. The other is that a patent gives its owner a very strong monopoly power, while at the same time the hurdle for coming up with a patentable idea in software is extremely low. The "one-click" patent is a case in point, as the technology’s original developer explains:
I was asked, "Why didn’t you patent this yourself, if you developed it first?" My reply was, "It only took me an hour to build; if I went down to the patent office after every hour of programming, I wouldn’t get very much done."
The effect is that there are many ideas around that are patentable under current rules—and many of them actually are patented.
Today, it’s impossible to create software without infringing someone’s patent. If you’re running a basic online shop, you’re touching on ca. 11,000 patents (see Bessen/Meurer (2008): Patent Failure, p.9). Complex programs are made up of thousands of ideas, any of which could be patented. Running patent searches is costly, with a price tag of ca. 10,000 EUR per search (though newer companies are trying to bring down the cost). But that still means that if you were to do things properly and investigate before putting your product into production, you’d spend hundreds of thousands, if not millions on patent searches that won’t contribute anything to your product. In addition, it’s practically impossible to tell where the area covered by a software patent ends unless you ask a judge. So in practice, rather than doing extensive patent searches, even the world’s largest software companies mostly close their eyes and hope for the best.
As a result, infringement is rampant. Big companies get rid of disputes among themselves through cross-licensing deals. Essentially, they put their piles of patents next to each other. The company with the smaller pile pays the other one a bit of money, and then they agree to leave each other alone. (The former CEO of Sun Microsystems, Jonathan Schwartz, explains the game rather well.)
Small companies can’t play this game. Once they become valuable enough to present a rewarding target for patent litigation, the industry incumbents can just blow them out of the water. Cost figures given for what it costs to start litigating a patent, but three million Euro per case doesn’t seem uncommon. That’s not the damages, mind you—it’s just the cost of getting the case going.
Yet litigation is only the last of a long chain of insidious tactics, starting with FUD (e.g. Microsoft claiming to hold 200+ patents on the Linux kernel, but never showing them to anybody) and paying dubious "experts" to toot the incumbents' horn, on through pressuring companies into licensing deals in order to profit from a product that the incumbent had absolutely no part in creating (Microsoft’s campaign to force Android device manufacturers into licensing deals, in order to bring up the per-device cost of Android to the same level as that of a license for Windows Phone 8).
Obviously, this has nothing to do with innovation, and everything to do with reducing competition in the market. Unfortunately, competition is a key driver for innovation—and, unlike with patents, nobody is disputing that.
The risk of being exposed to patent litigation makes it risky and costly to innovate. In the US, an increase in software patents actually went along with a reduction in R&D spending (Bessen/Hunt via Jamie Boyle in 2005). Patents also provide terrible ROI. In 1999, US public companies made profits of 3bn USD from patent licensing—but the associated litigation costs were 12bn USD. (Bessen/Meurer, 2008; excluding pharmaceutical companies)
Patent trolls (or "non-practicing entities") are a growing concern. While I didn’t have time to discuss these, I did briefly highlight an even greater risk: Mass aggregators like Intellectual Ventures, which hoover up patents by the thousands. These entities are basically behaving like hedge funds of the patent world. They’re also pushing litigation through thousands of shell companies. They’re not just a huge problem today; as they refine their techniques of monetising patents, they’re likely to impose a significant additional tax on innovation, which will be paid by all of us. (If you want to know what’s headed our way, read this research paper.)
Hearing the other side
Simon Davies went up to speak after me. (I haven’t seen his presentation online yet, so I’m trying to summarise his overall points as accurately as possible. I’ll be happy to add a link if the presentation becomes available.) He argued that the EPO is acting correctly in granting patents on software that makes a "technical contribution." Regarding my point about political control of the patent system, he pointed out that there have been two recent opportunities to change the rules (an update of the European Patent Convention in 2000, and the software patent directive in 2005, and that both these initiatives failed to resolve the lack of clarity in the EPC. As a consequence, the EPO is left to its own devices in interpreting the EPC.
He has a point there, of course—the political process so far has failed to effectively address the problem of software patents. But that doesn’t mean we should give up on politics. Yes, the process of making policy is often as gruesome and messy as that of making sausage. But even if our democratic institutions aren’t perfect, they’re the best system we know of to take decisions as a society. We need more democratic control over innovation policy, not less.
Policy also tends to lag the real world by years, if not decades (hello, copyright!). Since the reform attempts that Simon cited, the negative effects of software patents have become even clearer than they were in the early 2000s.
Simon also mentioned a number of statistics showing that the number of patent filings on software since the 1990s had grown at about the same pace as the number of Free Software projects; and that the US, with its extremely liberal approach to granting patents on software, at the same time was the most dynamic environment for software companies.
As he himself admitted, correlation isn’t causation. Based on the research I’ve seen and the industry experts I’m talking to, the growth in Free Software projects and the US' vibrant software business scene exist not because of software patents, but in spite of them. Software patents are acting as a tax on developers and as a brake on innovation, and removing them should make both Free Software and the technology sector even more dynamic than they already are. That’s a growth opportunity we’re currently wasting.
Interestingly, Simon first highlighted the moral argument that the Free Software community is making against software patents—that it’s wrong to exclude others from using a good idea. He then made a moral case of his own, arguing that inventors are entitled to benefit from their ideas, and citing the late Steve Jobs' moral outrage over supposed patent violations in Android. I always thought that Jobs' position on Android was a rather difficult one to take for someone who lived by (and loved) the phrase "good artists copy; great artists steal."
He admitted that patents can be a problem for competition. I don’t share Simon's optimism that competition authorities will be able to deal effectively with abuses. Participating in the European Commission's lawsuit against Microsoft has given us at FSFE a first-hand experience of what competition policy can do. While the case resulted in a record fine against Microsoft, competition policy is slow and blunt as an instrument. Instead of waiting for years and years for conflicts to be resolved (and putting in tons of work in order to make it happen), it would be much easier to preclude abuses from occurring in the first place, by eliminating patents on software.
Answering to my point that the Internet and the WWW had grown so rapidly because they were not patented, Simon brought up RSA, saying that this patented encryption algorithm had enabled e-commerce to really take off. On closer inspection, it turns out that RSA was released into the public domain in 2000, two weeks before the patent expired. The bulk of growth in e-commerce came after 2000, when the technology was no longer patented.
Where is Europe headed?
The US approach of generously granting software patents has led to the current litigation crisis in the mobile industry, the rise of trolls and many other problems. Will Europe go down the same road? Currently, the EU is trying to finalise a single patent system for Europe, the "unitary patent." On November 19, the European Parliament's legal affairs committee agreed to a compromise text on the issue. While the text has not been officially published at the time of writing, it seems that the Parliament has given up political control over the patent system to the technocrats at the European Patent Office.
The European Court of Justice will have no role in Europe's unitary patent system. The reach and limitations of patents will apparently be determined by laws in each member state, so it's likely that there will be substantial divergence across the EU in terms of how patents are enforced.
The right thing to do would have been to bring the patent system firmly under political control. This system needs to be understood as one tool in a larger innovation policy toolkit. A final recourse to the European Court of Justice would have been a necessary (if not a sufficient) step to making sure that broader societal interests are considered in important patent disputes. As things stand, Europe's patent system is bound to fall out of step with the realities of innovation sooner rather than later.
Currently, a lot of policy on patents (as well as copyright) is made on the basis of faith and rather dubious argument. We urgently need to move on towards evidence-based policy making. Before a major policy decision is taken, several independent experts need to analyse the options. Results need to be reviewed periodically, and if they’re not satisfactory, the policy has to be adjusted.
This will also make the patent system more flexible, and more responsive to society’s needs. The world is changing, as businesses in countries like China, India, and South Korea are increasingly giving western firms a run for their money. Historically, countries have adjusted their patent system as suited their economic needs. Will European companies still be so keen on a strong patent system when their Chinese competitors start filing and enforcing more patents?