Patenting Software in Canada
While the world waits for the US Supreme Court decision in the Bilski case, the Canadian Patent Office is going forward with establishing its policies as to when patents can be granted for software inventions in Canada. They have done so by publishing a draft revision to the Patent Office manual that addresses this issue.
The Canadian Patent Office publishes the Manual of Patent Office Practice – MOPOP – to serve as a guide for both examiners and the public regarding patenting in Canada. On June 16 CIPO began its consultation on a draft revision to Chapter 16 of the MOPOP which addresses the subject of “Computer-Implemented Inventions”. CIPO invites comments in response up until August 19, 2010. Thereafter, the response will be evaluated in the development of a final policy to be adopted regarding computer-implemented inventions. Note: Comments submitted will be posted on CIPO’s website for public viewing.
This draft chapter is 41 pages long and is available at this link:
The issue addressed goes to the very heart of the patent system: What sort of things should be patentable. Policy wise, patents should probably be granted for things that need to be encouraged to come into existence and would be less likely to arise without the incentive of a patent right to motivate human initiative. In the past, stemming from the Statute of Monopolies in 1624, patents were intended to introduce “new manners of manufacture” into the land. In the last hundred years, the premise for the granting of patents has shifted from establishing new enterprises to encouraging the disclosure of inventive concepts. But we can still ask, inventive concepts regarding what types of things?
The Canadian statute, borrowing from the US Patent Act of 1836, recites that patents are to be granted for “any new and useful art, process, machine, manufacture or composition of matter, or any new and useful improvement in any art, process, machine, manufacture or composition of matter”. This is the definition of an “invention” under the Act. This statutory language was last addressed by Parliament in 1936. There’s no doubt that the issue raised by Chapter 16 of the MOPOP was far from the minds of any Member of Parliament at the time.
The issue is: to what extent can computer-implemented inventions be said to qualify under the definition of “invention” and therefore be qualified for the granting of a patent in Canada.
As long as hardware is involved, there is virtually no doubt that a patent can be granted if the other requirements for patentability, namely novelty and unobviousness, are met. The more difficult issue arises when the novelty resides in the manner of organizing or treating information, or surrogates for information, in the form of electric signals. This is the strength of computers. The information that they sort and analyze is valuable because it represents an analog to the real world. The mere fact that a computer process is used to carry out some other tangible activity, such as adjusting the color of paint in a paint production line, should not disqualify the overall process from being patented. The difficulty arises when the invention, the purported invention as presented by the patent applicant, resides in the processing of the information itself. An example may prove helpful.
One of the most powerful innovations in communication and computer technology was the conception of the checksum digit as part of computer words. Originally, when computers used eight bit words based on a string of eight binary values of zero or one, one digit was reserved to be used to indicate the “parity” of the other seven digits. Thus if a computer word consisted of: 11101010, the parity of the last seven digits is established by adding the digits and checking whether the result is an even or odd number: 1 + 1 + 0 + 1 + 0 + 1 +0 = 4. As 4 is an even number, by convention its parity is considered to be 1. Accordingly, the first or checksum digit is given the value: 1.
There are actually numerous mathematical algorithms that can be implemented to carry out this type of function: http://en.wikipedia.org/wiki/Check_digit
They all serve to catch errors arising from the corruption of information that is being transmitted. Within a computer, this transmission is occurring millions of times a second and it’s important to catch errors when they occur. This checksum parity check doesn’t catch all errors; two errors in one word may cancel out. But when a single error has occurred, and the checksum digit doesn’t match, a computer circuit can be programmed to ask for the retransmission of the corrupted data until apparently valid information is received.
While this example has been given in respect of the transmission of an electronic message in binary form, it would be equally applicable to the semaphore signalling systems of earlier times. At regular intervals, a flag signal could be raised that reflects on the nature of the earlier signals. Thus the concept which constitutes a valuable advance in signalling can be divorced from the mechanical way in which it is implemented.
Should such a process be entitled to the grant of a patent?
There are two approaches. One approach is to determine whether the new concept can be fitted within the words of the statute. On its face, almost anything could qualify as a new and useful “art”. But the meanings of words in the statute are constrained by the traditions of judicial interpretation. Another approach is to go to a higher level and ask the policy question: “Is this a case where patents should be granted to encourage innovation to occur that would be less likely to arise without the incentive of a patent right to motivate human initiative?” Some would quickly answer: “of course”, relying on the presumption that anyone who discloses a new and useful gift to humanity deserves a reward. On the other hand, a sophisticated review of the modern economy may show that there is sufficient incentive already for people to conceive of more efficient ways to communicate, analyze and manipulate information.
Allowing that innovations that include a mechanical component are accepted as being patentable, creative patent agents can, and have in the past, conceived of very cunning ways of defining an inventive concept so that it appears to apply to something tangible. Even the semaphore system operates on the basis of flags.
A further policy question therefore arises as to whether patents should be granted for essentially software innovations when they are presented in combination with hardware; or should a restriction be applied based on where the essential inventive feature resides. If the inventive feature resides in the conceptual manipulation of information, should this be grounds for rejecting a patent application, even if the prospective monopoly is presented in terms of a mechanical hardware environment?
The draft revision to MOPOP guidance on this issue takes the position in section 16.08.04 that a computer program per se is considered to be an abstract scheme or set of rules that is not of itself patentable. This policy would appear to bar patenting of the higher principle arising out of the checksum concept. But it would still leave open the prospect of granting patents for such an invention when incorporated into an arrangement involving tangible components.
This presentation of a draft document to define how the Patent Office expects patent applicants to behave is a compendious analysis of the issue. It is a valuable contribution to understanding how a patent office should manage applications that address computer-implemented inventions. While everyone is awaiting the long-delayed US Supreme Court Bilski decision, this document prepared by the examiners and staff at the Canadian Intellectual Property Office, is well worth reading.