Sir

Software can now be patented in the United States, and earlier this year a court decision lifted the ban on patenting ‘methods of doing business’. These moves could have an important impact on scientists — many programs and protocols routinely used in laboratories may now qualify for patent protection.

Over the past 20 years, the US courts have gradually lifted the constraints against patenting software. The argument against patentability was that software is an arrangement of algorithms, similar to any mathematical formula or abstract idea, that cannot be removed from the public domain. But there existed a recognition that software could embody novel and non-obvious inventions which should be the protectable fruit of an inventor's labour.

To try to clear up the confusion, legal tests were established that allowed one to claim a patent for the application of a mathematical algorithm or abstract idea if it brought ‘a useful, concrete and tangible result’. There was general feeling that the application of an algorithm had to be in the context of a machine or similar transformation of physical quantities. One could claim a machine, or method of operating a machine, that applied a mathematical algorithm, formula, or calculation with the practical result of generating an electrical signal or image on a display. It was also acceptable to claim software when reduced to a fixed medium, such as a computer disk. But these tests were cumbersome and confusing, especially where the application of a program produced a pseudo-abstract result such as a set of numbers.

Any question as to the patentability of software was eliminated by the decision in State Street Bank v. Signature Financial Group, Inc.1. That decision was re-emphasized in AT&T Corporation v. Excel Communications, Inc.2.

At issue in the State Street case was the patentability of a system for pooling mutual fund assets in an investment portfolio. The court confirmed that the system was patentable and held that the “transformation of data by a machine through calculations into a final share price, constitutes a practical application of a mathematical algorithm, formula, or calculation, because it produces ‘a useful, concrete and tangible result’ — a final share price momentarily fixed for recording”. One can surmise from this that the requisite useful result of the application of an algorithm by a machine running software need only be a number in electronic form.

The courts did not stop there. There had been a general understanding that methods of doing business were not patentable. It appears now that these are as patentable as other processes or methods, subject to the requirements of novelty and non-obviousness. Arguably, such methods could include a host of laboratory protocols and data manipulation routines run on computers.

In the Excel case, even the requirement that a machine be implicated in a data processing operation was essentially eliminated. The issue was the patentability of a process using Boolean logic to create data structures for telecommunications switching applications. The court found that the bare process of using Boolean principles to produce useful, concrete and tangible results, without pre-empting other uses of these mathematical principles, was patentable. The need to claim a process involving a mathematical principle within the constraints of the operation of a machine was discarded.

To the scientific researcher, the State Street and Excel decisions present a two-edged sword. On the one hand, novel and non-obvious software test protocols, data reduction programs, calculation schemes, and numerous other computer-based manipulations of formulas, which once were not thought patentable, arguably now qualify for patent protection. On the other hand, one must now be wary of infringement of the patent rights of others in inventions that may apply well-known algorithms or calculations.

Businesses are becoming aware of the possible impact of these decisions. A deluge of patent applications, for such diverse business methods as Internet advertising and invoicing systems, is arriving at the US Patent and Trademark Office. I fear that the patent office will be ill-equipped to examine such applications, especially for inventions in fields where algorithms are widely used and accepted as conventional, and therefore not regularly described in print.