Current: I am President of Transpose, LLC. The tagline for the company is: "Collaborative systems for knowledge and the arts on the Internet." Working with me at Transpose are Bob Swerdlow, who has won industry awards for previous software projects projects, and Matt Goyer.  Matt previously was co-founder of FairTunes. Transpose's first creation, emergentmusic.com, will be released soon. It will be a very unique service for enabling the best new artists to find their audiences, and vice-versa. It is based on principles of evolution, information theory, and Bayesian statistics; there is a lot of mathematics involved.

Recent: As VP/Advanced Technology of Athenium, L.L.C., I managed the development of TeamThink, a collaborative educational application. My tenure ran from the earliest stages until the product was at use at universities including Stanford, Yale, and Tufts, and at major corporations. (At that point, I felt that the product had enough momentum that my departure to pursue other interests would not harm the company's prospects.) I am one of the inventors named in Athenium's pending patent.

Earlier: In the mid-1980's I founded Microvox Systems, Inc., where we created a voice mail-based dating service in New York City called 212-ROMANCE. As far as I know, this was the earliest such service, and was one of the earliest voice mail services of any type. As a way of determine whose recorded "personals ad" should be played for who, I implemented an idea now referred to as collaborative filtering. As far as I have been able to determine from speaking to many people in the industry, this was the very first time collaborative filtering was implemented in a commerical system, and may have been the first collaborative filtering implementation in any context. Collaborative filtering is now used for such applications as recommending movies, music, and even Web sites.

Also, during this period, I had a long-term consulting contract with New York Telephone (now part of Verizon) where I designed and/or implemented a number of database systems involving many millions of customer records.

I have a patent, 05884282, which concerns a mathematical basis for collaborative filtering. The main algorithm presented there was tested on a database of movie ratings against the competitive algorithms published by the founders of Firefly (eventually purchased by Microsoft) and Net Perceptions prior to the creation of those companies. This technique was shown to have superior accuracy in finding people with similar tastes for recommending movies. That is, recommended movies more frequently turned out to be given high ratings by the users to whom each movie was suggested. (Note: if you are interested in this technology, I suggest that you download the PDF version of the patent and begin reading at column 20, line 59. That's where the description of the high-performing algorithm begins--the earlier material was included in the patent for completeness only.)