Ars Book Review: “Patent Failure” (original) (raw)
A pair of Boston University academics have just penned "Patent Failure," a …
How bad is it?
Patent Failure: How Judges, Bureaucrats, and Lawyers Put Innovators at Risk (buy)
James Bessen and Michael Meurer
Princeton University Press
352 pages
Anyone who has paid the slightest bit of attention to the technology press in recent years knows that the patent system is out of control. Barely a week goes by here at Ars that we don't report on an absurd patent application or a frivolous patent lawsuit. And we're no longer surprised when an innovative technology company is forced to write an 8- or 9-figure check as a result of a frivolous patent lawsuit.
But although there's now plenty of anecdotal evidence that patents are harming innovation, especially in software, there has been little hard data on the subject. Most people in the software industry suspect something is wrong, but we haven't been able to prove it, much less quantify the extent of the problem.
A new book by James Bessen and Michael J. Meurer, both of Boston University, aims to fill this gap with one of the most comprehensive empirical analyses of the patent system that has been performed in decades. Rather than piling up anecdotes of beleaguered innovators and rapacious patent trolls, Bessen and Meurer have done the hard work of collecting detailed data about the patent system. And the findings documented in Patent Failure: How Judges, Bureaucrats, and Lawyers Put Innovators at Risk are sobering.
Measuring patent performance
Patents are supposed to create a positive incentive for innovation by enhancing the profits of companies that develop new technologies. If the patent system is working properly, the average firm's patent portfolio should generate more profits than the total cost of defending against patent infringement lawsuits from other firms. If, in contrast, litigation costs exceed patent profits, that suggests the patent system is actually creating a net _dis_-incentive to innovation. In that case, innovative companies would be better off with no patent system at all.
Focusing on public companies, Bessen and Meurer use several sources—including the results of patent auctions, the rate of patent renewals, and the effect of patent portfolios on market capitalizations—to determine the value of public firms' patent portfolios. They estimate the costs of patent litigation (which can include factors such as business uncertainty, executive distraction, and bad publicity in addition to attorneys' fees) by observing how defendant firms' stock prices changed on the day the patent lawsuits were announced. They also divide their study into two categories of patents: chemical and pharmaceutical patents in one category, and all other patents in the other.
Their data suggests three broad conclusions. First, patent litigation began rising sharply in the early 1990s. The costs of defending against patent litigation for non-chemical firms held steady at about 2billionperyearduringthe1980s,butby1999(thelastyearforwhichtheyhavereliabledata),patentlitigationwascostingdefendantsmorethan2 billion per year during the 1980s, but by 1999 (the last year for which they have reliable data), patent litigation was costing defendants more than 2billionperyearduringthe1980s,butby1999(thelastyearforwhichtheyhavereliabledata),patentlitigationwascostingdefendantsmorethan10 billion annually. Second, while the chemical and pharmaceutical industries also experienced increased litigation during the 1990s, the problem was much less dire; the patent system still appeared to offer positive innovation incentives for drug and chemical firms.
Patent profits and litigation costs for publicly-traded US companies
Most shockingly, Bessen and Meurer's data suggest that outside of the chemical and pharmaceutical industries, litigation costs for the average public firm actually exceed profits from their patent portfolio by a wide margin. By 1999, the last year in their sample, defendending against patent lawsuits cost non-chemical public firms about 12billion,whiletheirpatentportfoliosgeneratedonlyabout12 billion, while their patent portfolios generated only about 12billion,whiletheirpatentportfoliosgeneratedonlyabout3 billion in profits. This data suggests that outside the chemical and pharmaceutical industries, the patent system actually reduces the net returns to innovation; firms don't earn enough from their patents to offset the costs of defending themselves against patent infringement lawsuits brought by other firms.
Patent profits and litigation costs for publicly-traded US companies
These results come with an important caveat: Bessen and Meurer focus was on publicly traded companies, not privately-held firms or individual inventors. The data does show that the patent system works relatively better for smaller firms, so including private firms in the calculation might mitigate the patent system's negative effect somewhat. But even if the patent system works relatively better for such firms, the net effect is still likely to be negative because public companies perform the lion's share of research and development. Moreover, the evidence suggests that the patents held by small firms are actually less valuable than patents held by large firms. The patent system may be a net positive for smaller firms, but the overall impact outside the chemical and pharmaceutical industries is still likely to be negative.
Patents and property
What explains the dramatic deterioration in the patent system's performance during the 1990s? Bessen and Meurer note that patents are often called "intellectual property," but patents lack one of the most important attributes of traditional property rights: clear boundaries. Property law has precise rules that allow anyone to determine the owner and exact boundaries of any given plot of land. That makes the rules predictable and makes it easy to avoid accidentally building on other peoples' land.
They call this the "notice function" of property rights, and argue that this function is lacking in today's patent system. It can be extremely difficult to determine what technologies are covered by any given patent. Bessen and Meurer's poster child is patent 4528643, which covers a "system for reproducing information in material objects at a point of sale location." "Point of sale location" is industry jargon for the cash register at a brick-and-mortar retail store, and the invention described in the patent was a kiosk that would produce on-demand copies of music or other creative works. But the patent's owner, E-Data, convinced the court that "point of sale location" could include an e-commerce website. Suddenly, thousands of e-commerce sites were potentially liable for violating a patent that everyone had previously assumed was a patent on a specific brick-and-mortar retail device.
As a result of recent legal changes, more and more technologies are subject to dozens of overlapping patent claims like those in the E-Data patent. The exact boundaries of any given patent are unclear, and some patents may turn out to be invalid after they are challenged in court. But defendants don't know which patents apply to them until after they've spent millions of dollars in litigation. If ordinary property law worked this way, anyone building a building would face lawsuits from multiple people claiming to own the land underneath it. The resulting flood of litigation would have a chilling effect on real estate development, because the expected costs of fighting off lawsuits might exceed the expected profits from the real estate project.
The surge of litigation in the 1990s can be partly explained by a series of decisions by the United States Court of Appeals for the Federal Circuit in the 1980s and 1990s that allowed such creative re-definitions of patent rights. Over the last quarter-century, the Federal Circuit loosened the "obviousness" requirement, expanded the range of patentable subject matter to include software and business methods, and second-guessed the Patent Office and lower courts on the meaning of patent claims. The result was more low-quality patents being granted and more uncertainty about what any given patent covered.
The property analogy also helps to explain why patents seem to perform better in the chemical and pharmaceutical industries. Chemical formulas provide a precise language for describing patent claims and a straightforward way to search existing patents. As a result, patent law is much more predictable in these industries. It's easier for firms to avoid inadvertent infringement, and patent holders have less difficulty collecting royalties.
Fixing the patent system
Patent Failure is a technical, academic book and its policy recommendations are correspondingly modest and nuanced. Several of their recommendations are designed to improve the predictability of patent boundaries. They urge courts to give greater deference to the Patent Office's interpretation of patent claims, and for appeals courts to give greater deference to district courts' claim interpretations. They suggest that the Patent Office be more aggressive about rejecting vague patent claims. And they propose the Patent Office offer the service, for a fee, of examining a patent and producing an opinion letter on whether a particular technology is covered by that patent. If the Patent Office concluded that the technology was non-infringing, the letter would carry some weight in court if the recipient was subsequently sued for patent infringement. And to curb the flood of low-quality patents, they advocate dramatic increases in patent fees.
Finally, they advocate creating a new safe harbor for good-faith infringers—those who develop similar technology without knowledge of the patentee's own products. They note that copyright offers an independent creation defense, and argue that a similar defense should be available for inadvertent infringers in patent law.
Bessen and Meurer propose requiring that applications for software patents be required to describe their inventions in greater detail (known as "enablement" in patent jargon). They express sympathy for a formal prohibition on software patents, but ultimately do not endorse that reform out of fears that defining software patents would prove too difficult.
Reading Patent Failure makes two things clear. First, the patent system is in desperate need of reform. Second, it is extremely complicated, and it is therefore difficult to predict the results of any given reform. This is one reason to think that, despite its disadvantages, a "no software patents" rule would still be a good idea. Bessen and Meurer are undoubtedly right that such a rule would not fix all of the problems with the patent system. And it's true that clever patent lawyers may find ways to evade the ban through clever drafting. Nevertheless, a ban on software patents would significantly reduce patenting in an industry where patents have clearly done more harm than good. And given that the patent system doesn't appear especially beneficial in related industries, an overly-broad software patents ban is unlikely to harm those industries either.
At the same time, software patent critics should bear in mind that in the long run, protecting the software industry from the scourge of patent litigation will ultimately require broader reform. If the patent system as a whole remains dysfunctional, keeping it from screwing up the software industry will be a never-ending struggle. In contrast, systematic patent reform may invalidate most software patents as a happy side effect. Banning software patents is a good short-term fix, but more fundamental reform is needed in the long run.
Read on for our sitdown with co-author James Bessen.
Interview, part 1
Ars Technica: How did you get interested and involved in this topic?
James Bessen: I was in the software industry in the 80s and early 90s. I studied economics and coming into the industry I came with a classroom orientation expecting very tight control of IP and found that there was a lot of sharing of information going on, which seemed a little odd, and very little in the way of patenting. And then I found in the mid-90s that there was this huge influx of patenting. That seemed odd. Why do we want to introduce patents into an industry already that seems to already be highly innovative? Clearly, a large part of the people in the industry were not entirely happy with that. I started doing some research on it. I first wrote a paper with Eric Maskin (who won the Nobel prize in economics in 2007) just sort of coming up with a theoretical explanation of how could this industry be so dynamic and innovative without patents. That paper created a lot of controversy, and I guess the controversy sort of compelled me to look more and more into patents.
Eventually, I hooked up with my co-author Mike Meurer. I took Mike's patent law course at BU and we set up a good working relationship, and Mike has had a focus on the notice function for a while, and it seemed an important and we started thinking a lot about how property rights can work well, and why they sometimes don't work well.
So were you still wearing an industry hat when you started getting interested in the software patent issues, or did you only start thinking about them after you entered academia?
Basically, I was a poor CEO because all the time I was in the industry I was sort of observing things and trying to think about them. We sold the company and so I decided I wanted to pursue other interests. My main focus wasn't about patents—most of my research has been focused on the nature and causes of innovation more generally—but patents sort of hijacked my research agenda for a few years.
So, were you involved in patent litigation or did you obtain any patents when you were in industry?
No, our company had only one case where somebody threatened to assert a patent against us and my reaction was that I didn't even read it. I just shoved it away and ignored it, which apparently turned out to be the right thing, although it was probably a pretty stupid thing to do. There was very little litigation around software patents. I got out around 1994. There wasn't a whole lot of it then. The advice we got in the early 1980s was "don't even think about trying to patent software."
There's this really striking graph in your book in chapter 6 where you compare the pharmaceutical and chemical industries to other industries, and you find dramatica differences in litigation rates relative to the value of associated patents. What do you think is different about the pharmaceutical and chemical industries that bring about this striking difference?
I think the biggest difference is that a chemical entity can be pretty well defined by the words in a patent. They may be very complicated words, and when you start getting into biotech, the statement breaks down a little bit. But certainly for the small molecules that constitutes most of what goes on with patenting in those industries it's pretty clear. And those firm definitions mean that there isn't that much need for litigation on the one hand because it's predictable that people are going to lose if they copy and get caught. On the other hand, that means that the patent owner gets much stronger protection, and therefore it's much more valuable.
The graph seems to suggest that the patent system isn't creating the same kind of positive incentives for other categories of invention that can be patented. I was interested that you didn't talk at all about what seems to me the obvious (if maybe naive) solution that those others things maybe shouldn't be patentable at all.
I think one possible solution is: "you can patent chemical entities and nothing but." I'm not entirely convinced that's not what the world's going to come to. That seems to be an extreme recommendation now, especially considering that 20 years or so ago, at least in some of the other technologies the patent system didn't seem to inflict harm and may have provided some benefit. I think if you look at something like instruments, it may be an industry where it's provided some benefit. It's not clear if that's true today.
Interview, part 2
So if the basic thesis of this book is right, it suggests that at least for large companies, the patent system is a net loser, especially in the software industry. Yet Microsoft, one of the largest companies on earth, has been relatively vocal in support of software patents. Why do you think it is that more companies haven't—if it's true that it's a net loss for large companies, why do you think most large companies are not pushing for more radical changes to the patent system?
Microsoft sees where patents are going to give them a leg up in dealing with things like Linux and possibly other sources of competition. For years, Microsoft got beat down by companies like IBM and had to pay royalties and that sort of thing. So it's in Microsoft's private interest to have these patents even though the figures they spend each year on litigation are pretty huge.
I also think this is still not a big enough problem to be at the CEO level in these companies. What happens is, you know, Larry Ellison can be against software patents, but he has been forced to hire patent attorneys for the general counsel's staff, and when they talk about patent reforms, it's these guys who are dealing with it, not Larry Ellison. If we're right that the problems are getting worse, this is going to move up and become something much more prominent, and you're going to see CEOs get much more involved, and I think that's going to make what they're asking for a little bit more radical.
There's a famous Bill Gates quote that I've used in some of my writings where in 1991, Bill Gates basically wrote a memo to his senior executives saying what a disaster patents would be for the software industry. As far as I know, he hasn't repeated that statement, or made that statement in public since then, but I've always wondered if he's changed his mind, or if he's just kept his mouth shut because it's no longer in his interest to say so. Or maybe the patent lawyers at Microsoft have a different perspective than he has, and he's sort of delegated patent policy to them. What do you think is going on there?
I suspect he knows his interest in this case and I'm assuming some of the other top executives have been very vocal about patents and tried to spread patent FUD against Linux and that sort of thing. I don't think it's so much a situation as I described with Oracle. But at the same time I think they recognize the downsides of patents, and it should be interesting.
I heard a talk by Brad Smith last fall, and his initial take was, "Well, the Supreme Court's given us almost everything we wanted." He said that Microsoft had gotten the outcome they were looking for in three out of four recent patent cases. I was struck that at the same time he said that, he admitted that the number of suits per year had gone up, to where before they were doing 30 at the time, now they're doing 50 or something like that.
So what do you think the prospects for reforming the problems you've identified in the short term and in the long term?
It sounds like the immediate patent reform effort in Congress is pretty much dead. It's been removed from the schedule, so it doesn't sound too optimistic. I don't think the political will is anywhere near where it needs to be to enact any kind of serious reform. Our prediction is that things are going to continue to get worse. How long that takes I have no idea. As they get worse, I think you're going to see some stronger political momentum for reform.
Timothy is a senior reporter covering tech policy and the future of transportation. He lives in Washington DC.
Comments are closed.
