Digital Law Online: Benson

To test the waters of patenting software-based inventions, a number of companies (for example, Bell Telephone Laboratories) filed patent applications. Some applications claimed the invention in terms of an apparatus built from special-purpose hardware as well as implemented using a conventional digital computer. The Patent Office rejected many of these applications, and a few were appealed to the Court of Customs and Patent Appeals (CCPA) with mixed results.

In 1963, Bell Labs filed an application on behalf of its employees, Benson and Talbot, entitled “Conversion of Numerical Information.” It was for a method for converting from binary-code decimal (BCD) numbers to pure binary numbers. The Patent Office Board of Patent Appeals and Interferences had affirmed the examiner’s rejection of the application, and the company appealed. At the time there were only two claims remaining:

Claim 8. The method of converting signals from binary coded decimal form into binary which comprises the steps of

(1) storing the binary coded decimal signals in a reentrant shiftregister,

(2) shifting the signals to the right by at least three places, until there is a binary ‘1’ in the second position of said register,

(3) masking out said binary ‘1’ in said second position of said register,

(4) adding a binary ‘1’ to the first position of said register,

(5) shifting the signals to the left by two positions,

(6) adding a ‘1’ to said first position, and

(7) shifting the signals to the right by at least three positions in preparation for a succeeding binary ‘1’ in the second position of said register.

Claim 13. A data processing method for converting binary coded decimal number representations into binary number representations comprising the steps of

(1) testing each binary digit position i, beginning with the least significant binary digit position, of the most significant decimal digit representation for a binary ‘0’ or a binary ‘1’;

(2) if a binary ‘0’ is detected, repeating step (1) for the next least significant binary digit position of said most significant decimal digit representation;

(3) if a binary ‘1’ is detected, adding a binary ‘1’ at the (i+1)th and (i+3)th least significant binary digit positions of the next lesser significant decimal digit representation, and repeating step (1) for the next least significant binary digit position of said most significant decimal digit representation;

(4) upon exhausting the binary digit positions of said most significant decimal digit representation, repeating steps (1) through (3) for the next lesser significant decimal digit representation as modified by the previous execution of steps (1) through (3); and

(5) repeating steps (1) through (4) until the second least significant decimal digit representation has been so processed.

The examiner rejected the claims because they did not recite statutory subject matter. They were not rejected because of a lack of novelty or nonobviousness, although those factors may indeed have been absent. Instead, the examiner thought the claims set forth “mental processes” or “mathematical steps,” neither of which he believed were statutory subject matter.

The CCPA reversed the examiner in In re Benson. {FN7: 441 F.2d 682, 169 USPQ 548 (1971)} Considering claim 8, the CCPA first asked “would a reasonable interpretation of the claims include coverage of the process implemented by the human mind?” Because of the reciting of a “reentrant shift register” as an element of the claim, the answer clearly is no. That is a hardware element, so we have a process that can be implemented only on a machine, not by a human mind.

(The lawyer for the Patent Office referred to the inclusion of the hardware device in the claim as a “most embarrassing phrase.” In his decision, Judge Rich correctly pointed out that it is not a mere phrase, but a key limitation of claim 8, being referred to in most of the elements.)

More important, the claimed process could be carried out without any human intervention, unlike processes with mental steps that require a subjective judgment to be made. The Patent Office urged that the method was not a process within the meaning of Section 101, because the method worked on numbers, which are mathematical abstractions. The court did not find that argument persuasive, since cash registers and other patentable machines also work on numbers and that has never been a reason for taking them out of the category of machines.

Claim 13 was more problematical, since it did not recite any hardware, not even the reentrant shift register. It could be done with any hardware, with paper and pencil, or even with “red and blue poker chips and a surface to put them on or slots to put them in so that O’s and 1’s can be represented.”

While the court conceded that the method of claim 13 could be performed by a person, it thought it was improbable that anyone would ever do it by hand in actual use:

Realistically, the process of claim 13 has no practical use other than the more effective operation and utilization of a machine known as a digital computer. It seems beyond question that the machines – the computers – are in the technological field, are a part of one of our best-known technologies, and are in the “useful arts” rather than the “liberal arts,” as are all other types of “business machines,” regardless of the uses to which their users may put them. How can it be said that a process having no practical value other than enhancing the internal operation of those machines is not likewise in the technological or useful arts? {FN8: 441 F.2d at 688, 169 USPQ at 553}

The decision of the Board was reversed by the CCPA, and the Patent Office appealed to the Supreme Court.

I.B. The Supreme Court’s Benson Decision

The Supreme Court has gone through phases where it is hostile to patents, viewing them as a monopoly that should not be extended any more than necessary. In 1949, Justice Jackson wrote that “the only patent that is valid is one which this Court has not been able to get its hands on.” {FN9: Jungersen v. Ostby, 335 U.S. 560, 572, 80 USPQ 32, 36 (1949)} Justice Douglas, who wrote the opinion in Gottschalk, Commissioner of Patents v. Benson, {FN10: 409 U.S. 63, 175 USPQ 673 (1972)} was certainly not an advocate of expanding the scope of patents. In a unanimous decision (with three justices not taking part in the decision, presumably because of conflicts of interest), the Court reversed the decision of the CCPA.

Citing past Supreme Court decisions, the opinion states “While a scientific truth, or the mathematical expression of it, is not a patentable invention, a novel and useful structure created with the aid of knowledge of scientific truth may be.” That statement followed the long-standing rule that an idea of itself is not patentable.

Here the “process” claim is so abstract and sweeping as to cover both known and unknown uses of the BCD to pure-binary conversion. The end use may (1) vary from the operation of a train to verification of drivers’ licenses to researching the law books for precedents and (2) be performed through any existing machinery or future-devised machinery or without any apparatus. {FN11: 409 U.S. at 68, 175 USPQ at 673}

This is a strange argument. Because the claimed technique for converting from BCD to binary has great utility, in that it can be used in a wide variety of applications, it isn’t entitled to patent protection! Many things have a variety of uses – a novel fastener can be used to hold together parts of airplanes, furniture, or even broken human bones, but it is still a patentable manufacture.

Justice Douglas then revisited Cochrane v. Deener, {FN12: 94 U.S. 780 (1876)} an 1876 Supreme Court patent decision, for its definition of a process as something that transforms its subject matter to a different state or thing, which he notes as a “clue to the patentability of a process claim that does not include particular machines.” But then he states:

It is argued that a process patent must either be tied to a particular machine or apparatus or must operate to change articles or materials to a “different state or thing.”

We do not hold that no process patent could ever qualify if it did not meet the requirements of our prior precedents. It is said that the decision precludes a patent for any program servicing a computer. We do not so hold. {FN13: 409 U.S. at 71, 175 USPQ at 676}

So we now know what Benson does not mean: that computer programs as a broad class are unpatentable. But how do we tell if a software-based invention is patentable or unpatentable? Justice Douglas states “in a nutshell” the reason why the claims are unpatentable:

It is conceded that one may not patent an idea. But in practical effect that would be the result if the formula for converting binary code to pure binary 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. {FN14: 409 U.S. at 71-72, 175 USPQ at 676}

Previously in the decision, an algorithm was described as “a procedure for solving a given type of mathematical problem.” But an algorithm, as the term is normally used, is simply an orderly procedure for solving a problem. While many algorithms solve mathematical problems (such as finding a square root) or involve the manipulation of numbers (such as Benson’s invention), other algorithms solve purely nonmathematical problems (such as finding a character string matching a pattern) or problems where numbers represent properties of items (such as the width of a line and the width of characters when trying to justify lines of type). The decision does not recognize these differences.

Of course, the claimed algorithm isn’t the only way to convert BCD numbers to pure binary. Other, presumably less efficient algorithms had been used before Benson’s invention, and any valid claims by Benson cannot read on that prior art.

The opinion then repeats the comments of the President’s Commission on the Patent System discussed above, and indicates that if patent protection for software-based inventions is desirable, it should be Congress and not the courts that should extend that protection. But since Congress had described statutory subject matter using essentially the same language for almost 180 years at the time of the Benson decision, it apparently did not think that a special category of statutory subject matter was required for each new technology.

I.C. Trying to Make Sense of Benson

It is difficult to know where the Supreme Court drew the patentability line for computer algorithms in Benson. We do know that the Court did not think all computer programs were unpatentable because it specifically stated “It is said that the decision precludes a patent for any program servicing a computer. We do not so hold.” {FN15: 409 U.S. at 71, 175 USPQ at 676} But we also know that Benson’s claimed invention was not patentable.

In In re Freeman, {FN16: 573 F.2d 1237, 197 USPQ 464 (1978)} the CCPA was asked to decide whether a technique for computer-based typesetting was statutory subject matter. Freeman had invented a system that made it easier to typeset mathematical formulas. The examiner had rejected the claims as involving mental steps, but the Board reversed the examiner’s rejection and instead rejected the claims as nonstatutory subject matter as described in Benson:

The board construed the claims “to cover a known program loaded display typesetting device wherein the novelty resides in the program.” In applying Benson, the board said the improvement covered by the claims appeared to have no substantial practical application except in connection with a digital computer and that the coverage sought “in practical effect would be a patent on the algorithm itself.” That claims 1-7 were apparatus claims was dismissed on the view that appellant should not be allowed to claim indirectly what he cannot claim directly with method language.

Neither the examiner nor the board questioned that appellant’s invention, as claimed, was new, useful, and unobvious. {FN17: 573 F.2d at 1242-1243, 197 USPQ at 469}

The CCPA pointed out the problem with the Board’s reasoning:

The fundamental flaw in the board’s analysis in this case lies in a superficial treatment of the claims. With no reference to the nature of the algorithm involved, the board merely stated that the coverage sought “in practical effect would be a patent on the algorithm itself.” Though the board gave no clear reasons for so concluding, its approach would appear to be that every implementation with a programmed computer equals “algorithm” in the Benson sense. If that rubric be law, every claimed method that can be so implemented would equal nonstatutory subject matter under 35 USC 101. That reasoning sweeps too wide and is without basis in law. {FN18: 573 F.2d at 1245, 197 USPQ at 470}

The CCPA proposed a two-part test for determining whether a claim recites statutory subject matter:

First, it must be determined whether the claim directly or indirectly recites an “algorithm” in the Benson sense of that term, for a claim which fails even to recite an algorithm clearly cannot wholly preempt an algorithm. Second, the claim must be further analyzed to ascertain whether in its entirety it wholly preempts that algorithm. {FN19: 573 F.2d at 1245, 197 USPQ at 471}

The dictionary definition for an algorithm is “a step-by-step procedure for solving a problem or accomplishing some end.” This broad meaning for an algorithm is essentially the same as for a process, and processes are specifically included in Section 101 as a class of statutory subject matter. Hopefully, the Supreme Court in Benson was not trying to remove processes from Section 101, so it must have been considering algorithms in a more restricted sense. In particular, it defined algorithms as a “procedure for solving a given type of mathematical problem” (emphasis added) and appeared to use the terms “algorithm” and “mathematical algorithm” interchangeably.

The CCPA determined that “an algorithm in the Benson sense of that term” is a mathematical algorithm, and the two-part test can be paraphrased as follows: Determine whether the claim directly or indirectly recites a mathematical algorithm. If it does and the claim wholly preempts that algorithm, the claim is nonstatutory.

The CCPA cautioned that there were many ways a mathematical algorithm could appear in a claim.

In some claims, a formula or equation may be expressed in traditional mathematical symbols so as to be immediately recognizable as a mathematical algorithm. Other claims may use prose to express a mathematical computation or to indirectly recite a mathematical equation or formula by means of a prose equivalent therefor. A claim which substitutes, for a mathematical formula in algebraic form, “words which mean the same thing,” nonetheless recites an algorithm in the Benson sense. Indeed, the claims at issue inBenson did not contain a formula or equation expressed in mathematical symbols. When considered as a whole, each of the claims in Benson did, however, recite in prose a formula for converting binary coded decimal numbers into binary numbers. {FN20: 573 F.2d at 1246, 197 USPQ at 471 (citations omitted)}

Because the claims to Freeman’s computer-based typesetting system did not recite, directly or indirectly, a mathematical algorithm but instead an orderly procedure for specifying where the characters of a formula should be placed on a screen or phototypesetter, the first part of the two-part test was not satisfied and the claims were not drawn to nonstatutory subject matter.

Next section: Flook