Last week I argued that the Supreme Court's widely anticipated ruling in the case of CLS v. Alice wasn't the knockout blow software patent opponents had been hoping for. The Supreme Court struck down the specific patent at issue in the case, but it was vague about when, if ever, other software patents were allowed.
Reading commentary on the case has made me more convinced that software patent owners should be worried.
In a nutshell, the Supreme Court said two things: you can't patent abstract ideas, and merely implementing an abstract idea on a generic computer isn't enough to turn it into a patentable invention. The big question is: what's an abstract idea?
The patents the Supreme Court struck down last week and in a 2010 case called Bilski v. Kappos were extremely abstract. In essence, both patents took an abstract business strategy — like holding money in escrow to prevent either party to a deal from backing out — and claimed the concept of implementing it on a computer. In both 2010 and 2014, the Supreme Court said that wasn't enough for a patent.
Some software patent supporters, like former Patent Office director David Kappos, have concluded that the decision leaves most software patents unscathed. But the respected patent scholar Robert Merges, a software patent supporter himself, is not so sure.
The key problem for software patents is the Supreme Court's first software patent decision, Gottschalk v. Benson, involved a patent that was a lot less abstract than the Bilski and Alice patents. The Benson decision, which Merges calls "notoriously ill-considered," held that mathematical algorithms are themselves abstract ideas. The case involved a patent that claimed a specific algorithm, and even included some details about how to implement the algorithm using computing hardware called shift registers.
In other words the Benson patent was one of the least abstract software patents you can imagine. Yet it was still considered too abstract for the Supreme Court. Evidently, even extremely specific sequences of mathematical operations are too abstract to merit patent protection. In Merges's view, that comes uncomfortably close to saying that software in general isn't patentable:
The Court’s insistence on carrying forward cases such as Benson makes it difficult to figure out just what it means for a claim to cover an abstract idea. To see this, consider as an example a celebrated claim issued to Larry Page, co-founder of Google. The Page patent claims what is known as the page rank algorithm: a way of weighting web pages by the density of links to them. The idea is that when a web search turns up various web pages that include the search term, the more important pages will be those that show a density of links to and from them. This idea was key to the early success of Google in gaining a reputation for superior search results, and the rest (as they say) is history. My point is simply that it is conceivable that the claims in this patent could be characterized as too abstract under Alice. If a court were to become convinced that the Page patent claimed the abstract idea of "weighting," it might be invalid. Likewise, if it were convinced that the page rank procedure was simply a claim to a weighting algorithm, it might invalidate the claim by analogy to Benson.
Merges thinks invalidating the PageRank patent would be a disaster. I'm not so sure. While Google presumably likes having the PageRank patent, patents in general have been more a nuisance than an aid to Google. In the software industry, the patent system has effectively worked as a way to transfer wealth from innovative companies that are too young to have many patents (like Google 10 years ago) to older companies that have become better at patenting than at innovating.
But the bottom line is that the Benson decision is more radical than a lot of software patent supporters want to admit. If the Supreme Court takes it seriously, we should expect to see a lot more software patents invalidated in the coming years.