All Humans Bow Before the Mighty Watson, Master of "Jeopardy"
That whooshing sound you just heard may have been the passing of an age. To the list of games–including chess, checkers and backgammon–in which a computer has beat the best human players, you can now add “Jeopardy.”
In the third episode of the nationally televised publicity stunt put on by IBM (and as publicity stunts go, I’ll grant this one has been mightily entertaining) the computer, named Watson, ultimately prevailed over two of the game’s greatest human champions. The final score for tonight’s game was $44,131 for Watson, $19,200 for Ken Jennings and $11,200 for Brad Rutter. Total scores over two days of play was $77,147 for Watson, $21,600 for Rutter and $24,000 for Jennings.
The humans, however, didn’t go down without a fight. After Watson scored early on what looked like six question in a row. Jennings, who had lagged in the first game, rallied spectacularly, racking up a decent score on questions that Watson seemed to struggle with in the category of “Actors Who Direct” and on one about what you call a section of a newspaper that’s added within the folds. (Answer: What is an insert?) The score at the first break: Watson $4,200, Jennings $3,400, Rutter, $600.
During the second segment it got really interesting. Jennings managed to pull ahead, and in so doing probably made some IBM executives sweat in the process. That “Actors Who Direct” category continued to trouble Watson, which didn’t hit the buzzer as often. Jennings pulled ahead on another question concerning newspapers and then hit a Daily Double, which he answered correctly with a wager of $3,600 that doubled his score to $7,200, giving him the first human lead of the entire match.
Watson then went on to miss a few. Jennings and Rutter both beat it on questions it didn’t try to answer, and Jennings went on to build a considerable lead. By the time the board was cleared, Jennings’s score was $8,600 to Watson’s $4,800, while Rutter had $2,400.
In the third segment, Jennings went for a run, building up a commanding lead. Watson missed a crucial Daily Double in the nonfiction category, choosing “Dorothy Parker” where the answer was “The Elements of Style.” Jennings started what appeared to be a methodical hunt for a Daily Double, selecting high-value questions. Late in the game, Jennings had a $17,000 score to Watson’s $15,073, when Watson found the next Daily Double. This time it answered correctly, bringing it to $15,440.
Watson then had another run, finally pulling ahead of Jennings going into Final Jeopardy. The score was Watson $23,440, Jennings $18,200 and Rutter $5,600.
Here the game came to an interesting point. The question in the category of “20th century novelists” concerned the author Bram Stoker, who wrote “Dracula.” All three got it right. Before seeing the clue, however, Jennings made a conservative wager: Only $1,000, bringing his final score to $19,200. Had he bet more aggressively, he would have doubled his score to $36,400.
Watson bet $17,973, unusually aggressive, as it has generally wagered carefully in earlier rounds of Final Jeopardy, bringing its total to $41,413. My question for Jennings: Why didn’t be bet all the marbles and go for the win, on the chance that Watson might get it wrong?
Episode two had ended strangely. Having correctly answered 29 of 32 “Jeopardy” clues correctly, including two daily doubles, Watson completely blew Final Jeopardy. Asked to name a U.S. city whose two largest airports were named for a World War II hero and a World War II battle, Watson answered “Toronto?????”–the question marks indicating a low level of confidence in the answer. The correct answer was Chicago, where the airports in question were named O’Hare, after Edward “Butch” O’Hare (a U.S. Navy flying Ace), and Midway, an important battle of the Pacific War.
How could Watson have gotten it so wrong? There was at least some logic to its attempt. Readers have since told me that Toronto’s two biggest airports are named for Lester B. Pearson, Canada’s 14th prime minister, and Billy Bishop, a Canadian flying ace, both Canadian figures from World War I. But how could it choose a Canadian city when the category was “U.S. Cities”? Steve Hamm, a former colleague of mine at BusinessWeek and now an IBM employee explains further here.
To its credit, Watson had wagered only a small amount, not enough to blow the game. As we heard from my other former BusinessWeek colleague Stephen Baker last night, there were human TV producers in the room secretly hoping that Watson would bet all its marbles on Toronto, and that it would lose, after having essentially wiped the floor with Rutter and Jennings. It would have made for the kind of dramatic turnabout that’s great on television but would have embarrassed IBM.
For more on all this, I once again turned to Baker, author of “Final Jeopardy,” a book covering the behind-the-scenes story leading up to this matchup. (The book is out in e-book format now, and in hardcover tomorrow.) We talked via Skype and I recorded our chat, which you can hear below. In it he talks about where Jennings made a crucial tactical error, the controversies that will likely remain now that the game is finished, what Watson will likely do next and the possibility of a rematch.
- IBM “Jeopardy” Challenge Day 2: Very Different From Day One
- IBM “Jeopardy” Challenge Day One Ends in a Tie
- That Human Vs. Machine Practice Round of “Jeopardy” Didn’t End the Way You Heard It Did
- “Final Jeopardy” Question: Would You Buy an E-Book Without an Ending?
- This Supercomputer Defeated Human Champions of a TV Game Show in 2011
- I’ll Take Computer Company PR Stunts for $1,000,000