When U.S. Rep. Rush Holt beat IBM’s supercomputer Watson in an untelevised exhibition round of Jeopardy in February, the New Jersey Democrat became a minor celebrity. “I got tweets in languages that I still can’t recognize,” said Holt of the victory, where he amassed 8,600 points to Watson’s 6,200.
Anyone familiar with Holt’s biography shouldn’t be surprised he triumphed. He is a five-time Jeopardy winner and was the assistant director of the Princeton Plasma Physics Laboratory until his 1998 congressional campaign. But he had a greater purpose in going mano-a-machine: bringing attention to scientific research, an area that he believes is underfunded in the U.S. We caught up with the congressman this week to talk Jeopardy strategy, research dollars and the right pronoun for Watson. This interview was edited for length and clarity.
Txchnologist: Did you feel any nerves going into the showdown with Watson?
Rush Holt: When I was on Jeopardy in the mid 70s, I didn’t really have nerves because I didn’t realize how seriously people take Jeopardy. It was sort of a lark for me then. As soon as they told me that this was not being televised and it was not for keeps then I didn’t have any butterflies for this either. I just regarded this as a lot of fun. I was fascinated by the ideas that went into Watson.
I was really quite eager to do it as long as I was able to make the principal point that I wanted to make, which is that research is important. The research that goes into a game could be really useful in medical diagnosis or transportation planning later.
Txch: Were you not afraid of the headlines? “Nuclear physicist thinks he’s so smart, gets crushed.”
RH: I didn’t think I was so smart. I tweeted earlier in the day: “When I was on Jeopardy many years ago, Watson was just a little Atari. Now he’s all grown up and I’ve slowed down.” I also tweeted, “I’m glad this won’t be televised, that will make it fun.” It kind of shows the attitude that I was taking to him.
Txch: What strategies did you use to beat him?
RH: You click before you know the answer when you think you will know the answer. You’ve got to do that whether you’re playing humans or a machine. You’ll just get beat out if you wait until you’re sure of the answer. Watson evidently does wait until he’s sure of the answer… it’s sure of the answer. What’s the pronoun here?
Txch: I’m not sure.
RH: The other strategy is to be bold: nothing ventured, nothing gained. Being bold on the Daily Doubles. There’s a lot of luck involved in Jeopardy: what you know and what associations your brain happens to make that day.
Txch: Did this experience tell you about how your own brain works?
RH: As a scientist, I like to think about things like that. I’m fascinated by the technology of Watson as well as by the economic and social of importance of Watson. This approach toward solving things, which is very different than Deep Blue for example, it’s not quite computing by exhaustion. It’s much more a relationship. It’s building or searching for relationships.
Txch: The explicit purpose of the match you played was to raise awareness for science and math education. Do you think it accomplished that goal?
RH: I spend a lot of my time promoting the idea of research and explaining why we as a nation are underinvesting in research. Also, why it’s important for reasons of the economy and quality of life of Americans that we invest more in research. This was an opportunity to make that argument.
Courtesy IBM
Txch: Did you see Watson as an opponent?
RH: The idea of intimidating or strategizing against the opponent never enters my mind. I didn’t think of Watson as a cyborg, I just was doing my best to answer the questions. It turns out that on laundry detergents, which is one whole category, Watson didn’t know much. I don’t think it called for great intellectual depth to be able to remember All and Cheer.
Txch: What do you think it means for the economy as machines become more and more capable?
RH: Take medical information technology: If it’s done right it will help patients tremendously and will cut costs. Having good techniques for diagnosing and accounting for illnesses will be very important. And you would like to use computer systems to be able to recognize patterns and make connections. In order to play Jeopardy, you have to be able to get past puns and get some meaning out of the wordplay.
With a patient who is afraid to tell the doctor that he fears he has cancer or isn’t very good at describing her aches and pains, the doctor has to be able to find out what’ s behind this vague or misleading description. I think this technology is much more than a curiosity. It is an opportunity to demonstrate to the world at large that they had a computer that could know a lot, but even more, could find answers to questions that are poorly posed.
Txch: Do you have any suggestions for improving that technology?
RH: No. Those guys are much smarter than I am and have put a lot of thought into it. I just see this as an example of the kind of thing we should be doing.
Txch: Are you worried that, with everything you’ve done in your life, this Jeopardy match is how you’ll be remembered?
RH: A reporter said to me, “You understand this will be on your tombstone.” But, you know, this too will pass. Years ago, being on Jeopardy was the biggest thing that had happened to me.
Top image: Holt, Right, with Watson and other competitors. Courtesy IBM