Over at Forbes, Tim Lee has reminded us that it's the 40th anniversary of the case in which the Supreme Court really banned software patents
, arguing that they were really just math, and you can't patent math. That case, Gottschalk v. Benson
, had been seen to suggest that software programs, by themselves, could not be covered by patents. As the ruling noted:
It is conceded that one may not patent an idea. But in practical effect that would be the result if the formula for converting BCD numerals to pure binary numerals were patented in this case. The mathematical formula involved here has no substantial practical application except in connection with a digital computer, which means that if the judgment below is affirmed, the patent would wholly pre-empt the mathematical formula and in practical effect would be a patent on the algorithm itself.
As Lee notes, the above effectively applies to any software patent that can be reduced to an algorithm:
Of course, a similar argument could be made about any software patent. A computer program is nothing more than a sequence of mathematical operations—a complex mathematical formula. Therefore, any patent that claims a method of solving a problem by programming a general-purpose computer is, like the patent the high court struck down 40 years ago, effectively a patent on a mathematical algorithm.
So why do we have so many software patents today? Well, as Lee notes, it's basically because the appeals court, CAFC, that was set up to handle all patent appeals (among other things) effectively overruled the Supreme Court
on this issue (to be fair: the Supreme Court's ruling was not 100% clear):
Beginning in 1989, the Federal Circuit began handing down a series of decisions that made it easier to get software patents. By the end of the 1990s, all practical limits to patents on software had been dismantled, sparking the software patent arms race that continues to this day.
Yet theoretically, the Supreme Court’s 1972 ruling is still a binding precedent. The Supreme Court re-iterated its rule against patenting software in 1978. The Supreme Court did uphold a patent on a software-controlled rubber-curing machine in 1981, but its ruling emphasized that this was because the patent covered a physical machine that happened to have a software component, rather than claiming a software technique by itself.
I have argued that it's a mistake to specifically try to "carve out" software patents through some sort of regulatory measure, but I have no problem with the court finally recognizing that algorithms alone are math and shouldn't be patentable. I still think that won't fully solve the problems of the patent system (and that we'd be well-served by some other fixes
), but it would be a good place to start. Unfortunately, the Supreme Court has avoided addressing the question:
Unfortunately, the Supreme Court hasn't made any effort to rein in the Federal Circuit on the software patent issue. While the Supreme Court saved us from patents on medical diagnostic techniques this year, it hasn't examined the validity of a software patent since 1981. It's past time for the Supreme Court to insist that lower courts respect its precedents, which, after all, are still the law of the land.
What Lee leaves out is that it's not just that the Supreme Court hasn't taken any such cases, but that when it has taken cases where it could comment on this, it has actively avoided
the subject, and basically done everything to avoid having to make a direct ruling on this issue. That's unfortunate.