Tom Costello has a PhD in computer science from Stanford University. His emphasis is artificial intelligence (AI), and aside from teaching he has worked as a researcher at Stanford for DARPA and the U.S. Air Force. He was one of the founders of Xift, and he later worked at IBM. Today his research in the area of search architecture and relevance methods is truly groundbreaking and is at the heart of Cuil, the search engine he cofounded with his wife, Anna, a former Google employee.
SM: Tom, take us back to where your journey starts.
TC: I am from Ireland. I went to college at Trinity College, Dublin. I was in Ireland in 1990, when it was in the midst of a deep recession. I was just back in Ireland and find it funny how people here in America talk. They think back to the Great Depression. When I was in Brussels they were thinking back to WWII. Ireland never had boomed, so when we think back it is all the way back to the Great Famine. It was pretty bad in 1995.
My entire engineering class, all 130 people, left the country. There were no jobs. I remember the foreign minister at the time said we could not all live on a tiny island. I decided to go for a PhD, and I ended up at Stanford.
SM: What was your concentration?
TC: Computer science. I did that PhD and stayed on with research faculty working for DARPA and the Air Force.
SM: What areas of computer science were you working in?
TC: Artificial Intelligence. I did planning for the Air Force and knowledge basis for DARPA. In 1999, the department was completely empty.
SM: Everyone had gone off to Google?
TC: It was not to Google, but they had gone off everywhere. There was hardly a professor who was not doing a startup or two. Everybody was involved, including the students. It was like a ghost town, so I decided to do a startup as well.
We did a startup that was a search-based directory. We picked exactly the wrong time because the crash came. We started in 1999, picked up the angel money in 2000, and ran out of money in August 2000. People continued working. We had about 15 people, and they continued working for six months without being paid.
SM: Were you the CEO?
TC: I was the CTO and my wife, Anna, was the CEO. After that I went home to Ireland with my one-year-old for Christmas. When we came back we had a meeting where we were raising money. We had a term sheet, but the terms changed at the last minute, which I could not agree to.
After that, I had to go and look for a job. It was a good experience because there is a point in time where you have to say no, and I learned what that was. I got a job at IBM. They are a big, slow company and had just gotten into the Internet. They wanted to do text analytics for enterprise. Our first customer was Warner Music. They were interested because their problem was that they could sell any album, but sometimes it cost more money to sell it than it did to make it. They wanted us to predict the charts and tell them what was going to be good. We did analysis where we would crawl the Web and fan sites. We could tell by how active fans were on their fan sites how excited they were about a new release. We could predict the charts with a 0.7 correlation, which is pretty good.
Warner Music was excited, and then 9/11 hit. The guy we were dealing with was out of Atlantic Records, and the building is right beside the towers. Their building was uninhabitable and they had to move. IBM then completely changed and decided not to do anything.
While I was doing this, Anna was staying home with our second child. She volunteered to write a search engine in her spare time for Internet Archive, which was Brewster Kahle’s thing. In 2004 she launched a search engine which indexed the archive. Search had become hot again and Yahoo!, Microsoft and Google were all interested. Anna owned the IP, so she sold it to Google in January 2004. Suddenly, she was sitting there and realized that she had to go work at Google because they had her project.
Anna went off to work at Google, and I was sitting at home with three kids. It is quite hard to mind children, much harder than it looks if you have not done it before. It took me a couple of weeks to find someone who could mind the kids for an hour while I went to IBM to quit. The really sad thing is that when I went in to quit, they had not even noticed I was gone. How can you quit a job where people do not notice you have not shown up?