News

5) What is your impression of Bill Gates?

Bill is a very smart individual who has an amazing capability to focus intently on different topics. It amazed me that when we did a briefing on data mining, he would focus on the details and ask very specific and relevant questions. The reason this is amazing is that he does these briefings with many groups at Microsoft. The capability to move the focus of attention from one area to a totally different one, without being distracted, and with strong analysis, is a great talent.

The other great respect I have for Bill is that he actually takes the time to sit through product and technology reviews. Detailed ones. All this while he was still running the most-valued company on the planet. He also has a great ability to pick very smart and capable executives around him. Steve Ballmer is a great business leader. Not just very smart, but an amazing motivational and inspiring speaker. Rick Rashid (Sr. VP of Research) is also a super-smart and accomplished academician who is in tune with the business issues and strategies. Microsoft had an impressive bench of powerful and smart executives. Many of them wound up early investors in digiMine after they left: Pete Higgins, Sam Jadallah, and many others :-)

What I admire most about Bill Gates is that he managed to build a clean, agile, and effective organization. His recent relinquishing of the President and CEO duties are admirable as they came at the right time and showed that he can make the tough business decisions.  Ballmer was a great choice for succession.

6) How did you decide to start DigiMine?

Late 1999 and early 2000 was an environment where venture capital was very accessible. I would get calls from many VCs saying that if someone like me were to start a new company, funding would be instantaneous. Psychologically, it is always scary to give up a dream-job at Microsoft Research, to go out and start something from new from scratch. The scariest part was sacrificing family life and giving up all the security and the huge amounts still locked up in unvested stock options.

Two major factors had an overwhelming impact on me. One was personal and one was professional. On the personal front, I had achieved success in an academic research institution - Caltech's Jet Propulsion Lab - and had won NASA awards and medals. I also achieved success in an industry research lab: Microsoft Research, having built a basic research group and then a product group.  I had shipped technology in 3 major products ay Microsoft. I felt I also experienced the world of technical publications, having chaired several conferences, including KDD, and being Editor-in-Chief of the primary technical scientific journal in Data Mining.  I had also co-edited the KDD book [and many articles with you Gregory]. What I was missing was the element of risk. The feeling that everything is on the line and that what I do next could make the difference between success or utter and disastrous failure. The "security" of my position was dulling my senses, and I needed a new sense of risk and a true thrill of the hunt. Doing a high-tech start-up was definitely the next thing. And indeed, the capital materialized instantly and digiMine grew very fast from three cofounders to over 100 people in less than a year.

On the professional side, I was watching companies really struggle with data warehousing and data mining. Companies could not effectively run and maintain data warehouses, warehouses were overly complicated and expensive, and the business users were not getting any value from the data. Consequently, running a data mining application was essentially out of the question for an overwhelming majority of all businesses. To run a data mining application, you had to embark on a multi-month data quest. We figured the only way to get data mining to work, was to simultaneously solve the data warehousing problem and integrate business solutions within it. Key to this vision was running the data warehouse as a fully-hosted, fully-managed service for companies. The technology is so complex that a hosted model was the only way to make it work, to make it economical, and to make sure business users get the benefit.

next questions