I’m not a scholar…

But I play one on the web. With my search engine roots, I follow Google’s activities with great interest, given Excite’s prominence as an early leading search engine in the mid to late 90’s and subsequent demise in the bursting of the bubble. Perhaps I will opine at some later point on the “late-mover advantage” Google enjoyed that led to their current domination of search while early pioneers like Excite, Inktomi, Lycos, Infoseek and Alta Vista vanished or are shadows of their former selves. But I digress…

When Google announced the beta of Google Scholar, I used my usual technique to estimate the depth and breadth of a search engine: the vanity search. In this case, searching for “Ryan McIntyre” was a real test of their index, since I have only one published academic paper to my name. As usual, Google did not disappoint.

The paper, Bach in a Box: The Evolution of Four Part Baroque Harmony Using the Genetic Algorithm, appeared in the Proceedings of the First IEEE Conference on Evolutionary Computing in 1994 and was the culmination of a two quarters’ worth of classes I took during my senior year at Stanford in 1993 from Professor John Koza, who teaches an excellent course on genetic algorithms and GA’s cousin, genetic programming.

This paper was a satisying way to fuse my passions for music and computation, and it was thrilling to listen to the harmonies that my algorithm produced that sounded at least as good as anything any first year student of classical music theory might construct for class excercises. Ultimately, I think the genetic algorithm was not well suited to computer music composition because simpler randomized rule-based techniques could produce similar results while consuming far fewer computing cycles, but it was a fun project nonetheless, and according to Google, it has been cited seventeen times since it was published.

At the time I ran the evolutions for the project back in 1993, it took up to three hours to create a four-part harmony for a given melody on Sun Sparcstations and DEC 5000 workstations in the computer science lab at Stanford. Of couse, on today’s hardware, running the same program would take mere seconds, thanks to Mr. Moore.