The Bilski patent claims were directed to hedging risk in the field of commodities trading.  The claims did not include any combination of machine or apparatus.  Bilski relied on the 1998 Federal Circuit decision in State Street Bank which set out patentability for claims that produced a "useful, concrete and tangible result." The Federal Circuit discarded this test in favor of the machine-or-transformation test. Thus, a process claim must either recite a particular machine or apparatus or transform an article into a different state or thing. The Federal Circuit also warned against “mere insignificant post solution activity.” For example, claiming Pythagorean Theorem could be usefully applied to existing surveying techniques would not be patentable. 

 

The Bilski decision uses fairly generic language but its target is clearly business methods and software patents. Luckily, it provided some concrete examples of “do’s and “don’ts.”

 

DO: specify the particular type or nature of data claimed (i.e., bit, string, integer, float);

DO: specify how or from where the data was obtained (i.e., received from database SQL query, intercepted from HTTP stream, read from text file);

DO: indicate what the data represents (i.e., “a numerical value less than 7 indicating an acidic solution,” “a latency value of greater than 100 millisecond indicating a packet error”);

DO: claim the transformation of raw data into a new visual depiction of a physical object on a display;

DON’T: claim undefined “complex systems”;

DON’T: claim indeterminate “factors” drawn from unspecified “testing”;

DON’T: assume that adding a “data gathering” step alone will make an algorithm claim patentable;

DO: claim unique “data gathering” steps applied to an algorithm;

DON’T: assume that adding a “data recording” step alone will make an algorithm claim patentable (In re. Schrader, 22 F.3d 290).

 

Unfortunately, the Bilski decision focused primary on “transformation” and not how a machine should be recited to meet the machine-or-transformation test. It is likely that the Federal Circuit and perhaps the Supreme Court will expressly prohibit patent attorneys from “gaming” the test by throwing in a CPU to an algorithm claim. 

 

I anticipate we’ll see Benson (409 U.S. 62 (1972)) come up again in the context of “special purpose” computing devices. Setting meaningful limitations on patentability when reciting a machine might be cumbersome for the courts to enumerate. I propose the following test: Whether the machine collects data from at least one extrinsic source and applies intelligence dynamically without human-derived mental steps to achieve its claimed utility. 

 

The whole purpose of the computer is to automate calculations that would be slow for a person to do manually. However, it must apply some novel intelligence based on outside data to be patentable. Otherwise, the machine is merely a tool facilitating a mental step which is prohibited under 35 U.S.C. 101