My Murdock project in summer 2004 involved studying the evolution of technology as represented in US patent records. Our main goal was to get the data on the front page of all US patents (at least, those available in electronic form) and do a preliminary analysis of their evolutionary activity. This project follows up on an earlier Murdock pilot project, which resulted in the publication: A. Skusa and M.A. Bedau. 2002. Towards a comparison of evolutionary creativity in biological and cultural evolution. In R. Standish, M. A. Bedau, and H. A. Abbass, eds., Artificial Life VIII, pp. 233-242. Cambridge: MIT Press.
The pilot project involved about 800,000 patents, starting in 1996. The present project involved about 4,000,000 patents, starting in 1976 and continuing to the present. There are four phases of the present project, not all of which are completed yet. Phase one involved getting the data from the US Patent Office (for about $2000) and transferring it to a machine at Reed--a non-trivial task since the data format was arcane. Phase two involved constructing a rational database containing this data, including a large number of query scripts. Phase three involved analyzing the evolutionary activity of each individual patent as evident in the data (sample graph below). Phase four involved building a neutral model to normalize all the data. The support of my Murdock project last summer (in combination with some additional funds from the Dean's office) enabled me to complete the first 3 phases of this project and to start phase 4. When phase 4 is completed, I expect that this work will form the basis of further publications on cultural evolution and the evolution of technology. It will also be the foundation of significant new pedagogical materials concerning cultural evolution, the evolution of technology, and the relationship between cultural and biological evolution. For example, in my philosophy of biology course, I have a unit on cultural evolution and the concept of “memes” and the patent project both makes the issues vivid and also helps us see the way to their resolution. Cultural evolution in general, and memes in particular, are controversial for a variety of reasons. For example, unlike in biological evolution, the entities that “evolve” in cultural evolution seem never to go extinct; their selection is Lamarkian (acquired characteristics can be inherited); and the course of evolution is influenced by conscious goals and intentions. And unlike organisms and genes, memes are very difficult to individuate (where does not end and the next one start?), and their genealogy is uncertain at best. Now, all of these issues can be settled clearly and precisely with respect to that form of cultural evolution that involves the evolution of technology as reflected in patent records. For example, it is trivial to individuate a given invention (by its unique patent number), and each patented invention is certified by the USPTO to be novel and useful. Furthermore, an invention’s ancestry can be precisely reconstructed from the “prior art” that each patent is required to cite, so we can reconstruct the entire genealogical tree for all patented inventions. In these ways, the technological evolution project brings the otherwise dry issues of cultural evolution alive in the classroom. And the fact that the data were collected and analyzed (and published) with a Reed undergraduate definitely captures the class’s attention.
The following graph depicts the evolutionary activity of the 100 most active patents over the past 30 years. Each line indicated the life of one patent, and the y-axis (“number of hits”) indicates the cumulative number of times that patent has been cited by some subsequent patent. The more a given patent is cited, the more affect it is having on the further evolution of technology. Especially seminal inventions show up as the highest lines, and the slope of a line indicates the magnitude of a patent’s instantaneous impact on new technology. The data were normalized to remove the bias caused by changing average historical citations patterns. The graph vividly shows that in the early 1990s a newly invented technology (as it happens, ink jet printing, represented by the red lines that rise to the top of the graph) had an especially dramatic impact on the evolution of further technology.