If Wishes Were Horses - Or Patents - Or Something Like That

Oops. I was so busy Friday litigating two different ERISA cases that I plum forgot to post the latest episode of our semi-serialization of Robert Plotkin’s book on automated inventing and its impact on patent law, The Genie in the Machine: How Computer-Automated Inventing is Revolutionizing Law and Business. Anyway, better late than never, so here is the next installment.

Automated Inventing: Should Wishes be Patentable?

In my previous entry, I explained that you could view a human engineer who uses "artificial invention" technology to create a design for a new product as being analogous to an Aladdin who creates a new product by making a wish to a genie, as follows:

Human Inventor --> Wish --> Artificial Invention Technology --> Wish Come True (Product Design)

Based on the history of the patent system, we should expect that people will start filing patent applications for everything in this diagram that comes after the human inventor, namely the wish, the artificial invention technology, and the resulting product design. (In The Genie in the Machine I provide examples of patent applications which have already been filed -- and granted -- on the latter two of these three.) If we want patent law to grant such patents only when doing so will promote innovation, we need to make sure that the legal rules for patentability produce the right outcomes when applied to such patent applications.

First, consider wishes. What exactly do I mean by a "wish"? It is any description of a problem, written in a language that artificial invention software can understand, that the software can use to create a design for a product that solves the problem. For example, if you want to design a frame for an automobile that is as aerodynamic as an existing frame, but weighs 10% less, your wish might describe the aerodynamic requirements of the frame and its maximum weight. Some kinds of artificial invention software can use such a description to generate potential new car frame designs and -- using a simulator -- evaluate how well they match up to the requirements specified in your wish. The software then eliminates designs which fared poorly and modifies designs which performed well, in an attempt to produce even better designs. It then evaluates the new designs based on the criteria specified by your wish. It might repeat this process hundreds of times in an attempt to produce a product design -- what I have been calling the "wish come true" -- which satisfies your wish as closely as possible.

It may seem strange even to ask whether wishes should be patentable. Yet the kinds of wishes I am talking about are similar in some ways to traditional computer programs. Artificial wishes are a combination of instructions and data, written and stored in a physical form that can be processed automatically by a computer to perform a useful function. Therefore many, if not all, of the reasons that traditional software has been subject to patent protection also apply to artificial wishes.

Yet patent protection for software has remained highly controversial for almost 50 years. One reason is that computer programs are more "abstract" in some sense than cotton gins and other traditional nuts-and-bolts machines. Artificial wishes can be even more abstract than traditional programs. Therefore, to the extent that granting overly abstract patent claims can impede innovation, we need to be even more concerned about granting patents for artificial wishes.

We shouldn't, however, throw the baby out with the bathwater. Just as significant advances in software should be entitled to patent protection, even if the patents covering such software are relatively abstract, so too should new and useful artificial wishes be entitled to patent protection, so long as the legal requirements for patentability are applied with particular care to such patents. We should only grant a patent on an artificial wish if such a wish is truly new, useful, and nonobvious (the three fundamental requirements for patentability), if the specification of the patent describes how to use the wish in combination with artificial invention technology to produce new products, and if the patent's claims are written clearly.

I explain why this is the right solution in much more detail in The Genie in the Machine, where I also explain why artificial invention technology itself should be patentable. In my next, and final, blog entry I will explain why computer-generated inventions -- what I have been calling the "wish come true" -- should be patentable, and why it will be particularly tricky to apply patent law's "nonobviousness" standard to such inventions.