This year, the Supreme Court will be handing down three decisions on U.S. patent law. This is unprecedented. In the early 80s, the Court of Appeals for the Federal Circuit (CAFC) was founded to hear patent appeals from all districts, and since then the Supreme Court has generally left the CAFC to take care of patent issues. The CAFC has shown itself to be vehemently pro-patent. There is little debate in the academic literature over this point, except to ask why this is the case, a question I will not discuss here. But the pro-patent bias has made it very easy for some parties to harass others using claims of patent infringement. Companies that engage in such harassment are affectionately known as “patent trolls.”
One definition of the patent troll is that it is a company that holds a patent not to produce a product, but to extort royalties from others. However, non-producing patent holders are common. Say an inventor designs a million-dollar machine to bake cakes. Until she finds an investor to provide capital, she remains a non-producing patent holder, and if she shows the blueprint to an investor and he uses it without permission, then she has every reason to sue.
A more useful definition is that a patent troll is one who unfairly takes advantage of informational asymmetries by suing those who are ignorant of the field of patents in which the troll works. We would consider it unfair if a company independently works on its own product, and one day finds that it owes money to a patent holder for using the blueprint that the manufacturer drew up by itself.
A story like this is impossible in most fields of patentable endeavor. A pharmaceutical company that is ignorant of patents is an oxymoron, and the same could be said for virtually every other field of manufactured goods. But in the fields of software and business methods, parties ignorant of patents are rampant. In theory, every company that uses software or has a business method — that is, every last company with operations in the U.S.A. — should be doing a patent search on its web pages, accounting database, payment schemes, investment arrangements, and so on. We would need a thousand times more patent lawyers to make this even vaguely feasible. Thus, one who holds a patent on software or a business method need only wait and search to find someone who is using the described system and has never seen the patent.
The Supreme Court’s trilogy
With the intent of squashing trolls, the Supreme Court has stepped in, agreeing to hear three cases on patents. Our Justices certainly know how to build drama: the decisions will come in order of increasing importance and suspensefulness.
ebay v MercExchange
The simplest case was about the rules for preliminary injunctions. The CAFC had inexplicably stated that courts must grant preliminary injunctions halting the operations of patent holders’ competitors under very weak conditions. In an entirely unsurprising ruling (May 2006), the Supreme Court ruled that the standards for preliminary injunctions for patents should be as stringent as those everywhere else in the U.S. legal system. Perhaps the only surprise was the vehemence of the ruling: it was unanimous, and the main decision by Justice Thomas was accompanied by two concurring opinions providing more reasons for why the CAFC’s rule should be reversed. This ruling clearly slows down the patent troll, because its patents are not as powerful in court, but the troll can roll on with court cases that can tax defendants for years.
KSR v Teleflex
The second case, not yet decided, is about the definition of obviousness. The U.S. Code says that a patent may not be granted for an invention that would be obvious to a person having ordinary skill in the art (a PHOSITA). In the KSR case, the CAFC ruled that an electronic gas pedal may be in the literature, and an adjustable-height gas pedal may be in the literature, but that does not mean that an electronic adjustable-height gas pedal is an obvious invention. More generally, an examiner at the Patent Office may not reject the combination of two existing elements as being obvious unless there is a suggestion in the literature that they should be combined. Again, there is no question that the Supreme Court will overturn the CAFC’s supremely narrow view of what the PHOSITA is capable of without help from the literature. The question is whether it will only strike down the specific “suggestion test” regarding combinations of elements, or will grant broader scope for patent examiners to rule that an invention is obvious.
The patent troll depends on others independently inventing its patented invention, and an obvious patent is likely to be independently invented over and over again, so our patent trolls seek out the most obvious patents available. But if we were to somehow eliminate all obvious patents (an impossible task), even this would not be sufficient to stop the patent troll. In a field such as web design that includes millions of practitioners, very non-obvious inventions are still likely be simultaneously invented by many parties.
LabCorp v Metabolite
The third case, also not yet decided, is the one that has the real potential to kill the patent troll. It is also the most subtle and contentious. The case is about a test for a deficiency in one specific chemical (herein chemical B) via measurement of another (chemical A). Should the following two-step procedure be patentable?
- Measure chemical A in the subject’s blood.
- Correlate a high level of chemical A with a lack of chemical B.
The CAFC upheld this claim as entirely valid, but the Supreme Court is certain to overturn that ruling. The real cliffhanger comes in whether they will choose a narrow path, which halts some of the worst patent trolls, or a broad path, which will completely kill the concept of the patent troll but will also roil companies like IBM, who collect a significant amount of money from other companies via patent claims.
The narrow path would be to resort to the “law of nature exception”. A law of nature such as “A is correlated to B” can not be patented, but the CAFC has ruled that the physical act of correlating A to B is not a law of nature, but a physical step on equal footing with other physical steps like taking blood samples. But clearly, it is impossible to use the law of nature (the correlation) without engaging in an action covered by the patent (correlating). The Supreme Court has ruled in the past that if a patent entirely precludes something which is unpatentable, then one can take the two to be equivalent, and the patent should be rejected. A narrow ruling would remind the world of this rule.
A broad ruling would question the combination of steps, in the style of a review frequently attributed to Samuel Johnson:
Your manuscript is both good and original, but the part that is good is not original and the part that is original is not good.
Step one of the patent above is physical but not original and step two is not physical (or trivially so) but is original. So is the whole an original physical process?
To give another example to clarify what can happen, here is another potential patent:
- Specify a new equation correlating variable A to variable B.
- Write the equation down on a piece of paper.
Is the piece of paper patentable? This may seem like a silly example, since printed matter has a long tradition of being protected by copyrights rather than patents, but one nonprofit publisher filed a brief in the LabCorp case contending that if the patent is upheld, their printed books, consisting of novel information with a simple physical extension, would infringe the patent on devices correlating the two chemicals. Relying on the same logic, patents are even pending for storylines and new words.
In its Diamond v Diehr ruling (1981), the Supreme Court foresaw that tacking a standard fluid sample or piece of paper to an equation should not make the equation patentable. The ruling stated that: “… insignificant postsolution activity will not transform an unpatentable principle into a patentable process. To hold otherwise would be to allow a competent draftsman to evade the recognized limitations on the type of subject matter eligible for patent protection.”
“Insignificant postsolution activity” is the cornerstone for software patents and business method patents. If we were to write an equation not to a piece of paper but to a computer hard drive, then we would have a software patent. Given an equation about share prices, writing those share prices to paper comprises a business method.
By no coincidence, 100% of patent trolls work in software and business methods. Along with NTP (the company that sued the makers of the Blackberry), Acacia, BTG, DataTreasury, Eolas, Fogent, Intellectual Ventures, et al., are all eagerly searching for new people to sue over their software and business method patents. As discussed above, when physical innovation is required, the industry tends to be small; when there is no physical innovation required, the industry balloons to such groups as “all software users” or “all businesses.” Until all software users do regular patent clearances (have you done yours today?), patent trolls will have a pool of potential victims, regardless of the fundamental quality of the patent.
Returning to the Supreme Court’s choice, the broad option would be to remind the patent world of its ruling in Diehr, that indicated that a patent must have some physical step that is itself innovative (or is innovative in combination with other physical steps). That is, the combination of an innovative algorithm with an obvious or trivial physical step should not be patentable as a whole.
The broad ruling would return patent law to where it was back in the early 90s, when patents applied to high tech like semiconductors and pharmaceuticals, but not to payment schemes, mathematical algorithms, or storylines. It would thus eliminate the patent troll, because everyone is patent-aware in high-tech fields, so there are no informational asymmetries to be used for unfair profit.
However, the ruling would also annoy the more legitimate businesses that have also embraced software patents. Practitioners in many walks of computer science have gone on record stating that it is impossible to write software of any kind without infringing an IBM patent [Mann, 2004]. IBM collects about a billion dollars per year in patent royalties, much of it for software patents. It is therefore no surprise that IBM has filed an amicus brief in the LabCorp case suggesting that the Court take the narrow path in deciding the case.
Unfortunately, academics have been able to compile zero empirical evidence that the additional revenue IBM and other software practitioners have extracted from their competitors has fostered new innovation. One team of academics even found a weak negative relationship between software patent revenue and R&D; another found that parties in the computer industry are much more likely to patent for the sake of playing licensing games than those in other high-tech fields, whose patenting focuses more on actually protecting the invention [Bessen and Hunt, 2004; Cohen, Nelson, and Walsh, 2000]. That is, what academic evidence there is about patents on software indicates that even established software companies are realizing that trolling is easier than innovating. Not everything should be patentable, and the economic evidence indicates that non-physical goods should not be, even when “insignificant postsolution activity” is tacked on to make them look physical.
Some interested parties have stated that the problems with patents are a complex mix of minor flaws regarding injunctions, patent review, fee schedules, and so on, but all of our problems can be traced back to a single rule that allowed designs with a trivial physical step to be patented like bona fide physical innovation. This is why the Supreme Court’s patent trilogy has the makings of a perfect drama: it opened with a case whose outcome was obvious but which has a relatively limited effect on the patent world, and it will conclude with a nail-biter of a case, which could make IBM happy by putting another patch on a broken system, or which could put every last patent troll out of business in one step.
For for information on this topic, please also consult Math You Can’t Use: Patents, Copyright, and Software, also by Ben Klemens.