History of US Software Patents

Diamond v. Diehr: software is not patentable subject matter

In 1981, the US legal system would not uphold any software patent. This rule was so strict, that even including computerization as part of an overall process could lead to a patent being rejected by the patent examiner and by the Patent and Trademark Office Board of Appeals. Exactly that kind of rejection led to the US Supreme Court case of Diamond v. Diehr, 450 U. S. 175 (1981), quoted here:

Respondents filed a patent application claiming invention for a process for molding raw, uncured synthetic rubber into cured precision products. While it was possible, by using well-known time, temperature, and cure relationships, to calculate by means of an established mathematical equation when to open the molding press and remove the cured product, according to respondents the industry had not been able to measure precisely the temperature inside the press, thus making it difficult to make the necessary computations to determine the proper cure time. Respondents characterized their contribution to the art to reside in the process of constantly measuring the temperature inside the mold and feeding the temperature measurements into a computer that repeatedly recalculates the cure time by use of the mathematical equation and then signals a device to open the press at the proper time.

I've emphasized the part above about industry not being able to measure temperature precisely inside the molding press, as this patent has been popularly misconstrued as having the computer control as its only novel aspect. In fact, if you have the patience to keep reading the court opinion past its summary and move on to Part I, you'll see that they acknowledge that the computer control merely implements a well-known equation, nothing novel. However, they're careful to note, "The respondents claim to have overcome this problem by continuously measuring the actual temperature in the closed press through the use of a thermocouple." Also, "We note that the petitioner does not seriously contest the respondents' assertions regarding the inability of the industry to obtain accurate cures on a uniform basis."

The idea that the computer control is the only novel aspect of the patent claim appears in the dissenting minority opinion of the four judges who voted to rule the patent invalid for that very reason.

It's clear that, among the individual components of this patent claim, the court majority saw the continuous temperature measurement as most important. The computer-control part was obviously not patentable in and of itself, but a non-patentable part does not make the whole non-patentable. This seems obvious; just because a machine includes a screw or a nut as one of its parts, you wouldn't expect the whole machine to be non-patentable. However, in 1981 computer programs were "patent poison", and the US Supreme Court needed to give this clarification, which they justify at great length in Part II of their opinion.

In Part III, the court reaffirms previous decisions in which computer programs were found not to be patentable. Only as part of a larger process or product that is itself patentable, may they be part of a patent. The court wisely realizes that the computer program in question is in essence an algorithm or mathematical equation. "Their process admittedly employs a well-known mathematical equation, but they do not seek to pre-empt the use of that equation. Rather, they seek only to foreclose from others the use of that equation in conjunction with all of the other steps in their claimed process."

In Part IV, the court clarifies that algorithms, formulas and computer programs do not become patentable by rewording them to sound like a real-world machine, or by adding insignificant (presumably from a patent point of view) activity to the process.

The court tried to make it acutely clear that it was not opening the door to software patents. Nonetheless, it is widely believed that the US Patent and Trademark Office (USPTO) used this case as an excuse to begin granting software patents. I haven't verified this myself, but have no reason to disbelieve it. The USPTO is paid according to the number of patents it grants, not according to the number of applications it processes, so its interest is in granting as many patents as possible.

Further confounding the situation are lower court opinions that contradict the US Supreme Court. For example, in the US Court of Appeals for the Federal Circuit, case 96-1327, Circuit Judge Rich