Section 101 of the U.S. patent statute (Title 35) states that patentable subject matter is limited to four broad categories: "process, machine, manufacture, or composition of matter..." This implies, presumably, that there other categories that cannot be patented. For example, "algorithms" and "laws of nature" are two areas the Supreme Court has previously stated could not be patented. (Why algorithms are nevertheless now routinely patented is the subject of a paper I wrote last year -- I will attempt to summarize below). The definition of what can and cannot be patented has changed over time. The Supreme Court typically took a narrow view, for example in the Funk Brothers and Flook cases. However, the most recent Supreme Court 101 subject matter cases, Chakrabarty and Diehr, took a broader view. Diehr led to the modern view that software can be patented. The Federal Circuit has tended to ignore the restrictive cases, allowing software to be patented and, in the 1998 State Street case, allowing business methods to be patented.
Recently the United States Patent Office decided to align itself more closely with the Federal Circuit and loosen up its restrictions on patenting business methods. But today, the United States Supreme Court surprised the patent world by saying that it will hear the appeal in Metabolite. In this case, the "inventor" discovered a law of nature, namely that elevated levels of homocysteine in the blood tend to indicate a deficiency in the B vitamins cobalamine or folate. To take advantage of this discovery, the "inventor" claimed (in claim #13):
13. A method for detecting a deficiency of cobalamin or folate in warm-blooded animals comprising the steps of:
assaying a body fluid for an elevated level of total homocysteine; and
correlating an elevated level of total homocysteine in said body fluid with a deficiency of cobalamin or folate.
There is nothing novel about testing the blood for levels of homocysteine. The only novel and non-obvious part of the claim is the scientific discovery, the "correlating" step. If the claim is taken "as a whole," it (and any properly written claim covering any subject matter) passes muster under Section 101. If, however, focus is put on the subset of the claim that contributes to its novelty and non-obviousness, that portion of the claim will fail as a forbideen "law of nature."
Here is how the Supreme Court in its opinion granting the appeal expressed the issue:
Whether a method patent setting forth an indefinite, undescribed, and non-enabling step directing a party simply to “correlat[e]” test results can validly claim a monopoly over a basic scientific relationship used in medical treatment such that any doctor necessarily infringes the patent merely by thinking about the relationship after looking at a test result.
This will be the first time in over a decade that the Supreme Court has heard a case on this subject -- a decade that saw the end of restrictions on software patents (despite "algorithms" being unpatentable) and the rise of business method patents.
So why can software be patented despite "algorithms" being a forbidden subject matter under Flook and Diehr? Because clever drafters have added gratuitous steps or elements so that a patent claim, interpreted "as a whole," involves more than just an algorithm. For example, if you discover a new algorithm, you can patent that algorithm running as a computer program on a "device with memory." The "device with memory" is a gratuitous element, in that it does nothing to add to the novelty or the non-obviousness of the claim. It is only there to turn unpatentable subject matter into patentable subject matter.
There is an old methodology that would have prevented such clever drafting to get around the 101 subject matter restrictions. It comes from the old British Nielson case. This methodology essentially says that, for purposes of determining whether a claim covers patentable subject matter, you don't look at the "claim as a whole," (the Diehr test), but instead just look at that subset of the claim that is the inventive contribution. In other words, look at only those elements or steps that contribute to the novelty or non-obviousness of a claim.
The Nielson test was followed by the Supreme Court in expressly in cases from O'Reilly v. Morse (yes, that Morse, the telegraph case) in the 19th century to Flook in 1978. Had Flook been followed, most software and business methods would be unpatentable in the U.S. today.
The most rational, but also most radical, way for the Court to resolve Metabolite would be a return to the Neilson/Flook test. Rational, because it's the only proven test we have for enforcing forbidden categories of patentable subject matter, whether they be "laws of nature," "algorithms," or business methods. Radical, because it would invalidate at a stroke not only the Metabolite-style discovery-based claims and business method patents, but almost all U.S. software patents as well. Justice John Paul Stevens wrote the Flook opinion in 1978 and is the only justice from that era still on the Court. So there is a good chance that we will at least get a strong dissent along these lines. However, the more likely majority outcome will, if it overturns the patent, do so on the more narrow grounds that it involves a mental step ("any doctor necessarily infringes the patent merely by thinking about the relationship after looking at a test result") or that it fails the 112 written description standard.