Thursday, January 14, 2010

Sam's ideas

Earlier today, I was discussing with colleagues a problem in feature selection for logistic regression. I recalled that I had discussed the problem with Sam some months ago, and that he had a clever suggestion on how to approach it. I searched my email, and there it was. I read it, I read it again. I thought I got the main idea, but I was missing something. I wanted to ask Sam. He seemed to have seen farther, as he so often did. I needed his help. But I couldn't call him.

I fear re-reading Sam's messages. They are stolen rays from a receding star that I can never reach.

Wednesday, January 13, 2010

Sam Roweis

As I left the gym this morning, I learned that my friend and colleague Sam Roweis had died. I had a busy day of meetings ahead of me, which thankfully provided a pseudo-normalcy. In the gaps, we tried to console each other, we tried to think of useful things to do. I grieve deeply for him and the family he left behind.

Sam was a wonderfully creative, perceptive, warm, funny scientist. I met Sam at a NIPS conference in Denver in the mid-90s. I don't remember the year, but I remember vividly his enthusiasm and sense of humor as we talked about research and many other things at a busy food court near the NIPS hotel. Sam brightened every encounter with warmth, imagination, and humor.

We kept up with each other through the twists and turns of our research careers. We talked about many interesting ideas over the years. We got lost together in Whistler fog. I learned hugely from him, maybe he learned a teeny bit from me. He never dismissed my crazy ideas.

Sam encouraged me to come to Google, where he was already, and soon after I started there, he joined my group. We made plans to work together on several research problems, but life threw at Sam a much bigger problem than any of our scientific questions. We never wrote those papers, although just yesterday I noticed his writing on a whiteboard from the last time we had a detailed technical discussion.

To be closer to the support of family, Sam, his wife Meredith and their twin daughters moved to New York, where he took a faculty position at NYU while maintaining a part-time position with my group at Google. Just last week, he sent me a long message on possible directions for his work. I was very busy, and hoped to reply today. It will never happen.

Sam gave generously his ideas, his energy, and his spark to many friends, colleagues, and to his field. How could someone who gave so much of himself be running dangerously low on what is needed for living, I ask dumbly.

John Langford on Sam.