Watching the three-day Jeopardy match between Watson the IBM supercomputer, and former Jeopardy champions Ken Jennings and Brad Rutter, I couldn't help but feel bad for humankind. As a former Jeopardy contestant myself, I knew Watson was beating them where it hurt, not with superior general knowledge of trivia, but by avoiding typical human logical weaknesses.
If you've ever watched Jeopardy at home yelling out answers, you've noticed that contestants can't buzz in until after Alex Trebek finishes reading a question, but in the studio, the procedure is a little more complicated. The buzzing system is turned on by a Jeopardy employee who presses a button in another room once Alex finishes reading. Once the button is pressed, lights come up on the sides of the screen and the buzzers are activated. If you try to buzz in before the activation button is pressed and the lights go on, you're locked out of trying again for a quarter of a second, pretty much guaranteeing you won't get another shot.
After all, whether you're in the College Championship like me, or on the standard show, when you're on Jeopardy, you'll almost never be the only contestant who knows an answer. Once more than one person knows the answer, it all comes down to buzzer timing. Really excellent players like Brad and Ken buzz even if they haven't yet figured out the answer, since they're confident they will have figured it out by the time they're called on. To stand a chance, you have to sync up your reactions to the reflexes of the persons turning on the board, and if you lose your rhythm or the button-pusher varies his timing, you quickly get left behind.
Watson does better at buzzing, not because his electrical relays are so much faster than our nervous system (also running on electrical signaling), but because he has less data about when to buzz in. Watson doesn't receive any audio input -- he doesn't hear Alex read the clues or other players try to answer them. His buzzer timing is based solely on the electrical signal he receives. Players are much slower to react to the cue, since the signal lights may not come on when we expect them to, based on the pacing and tone of Alex's voice and the previous timing of the board-unlocking button.
More data can be a detriment to accurate decision making. In his 2004 book The Paradox of Choice: Why More Is Less, psychologist Barry Schwartz made a case that when people are confronted with too many options, they cannot efficiently choose between them. The strain of choosing and filtering causes emotional exhaustion and pain. And even when we escape emotional consequences of data overload, our reasoning can still be perverted by too much information.
Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations.
Which is more probable?
1. Linda is a bank teller.
2. Linda is a bank teller and is active in the feminist movement.
Eighty-five percent of those surveyed picked option two. Despite their conviction, option two is definitionally less likely. Both options assume Linda is a bank teller, but option two assumes that Linda is a bank teller and also is an active feminist. If proposition two is true, proposition one must be true, but if proposition one is true, proposition two could still be false. Option one must be more likely, but the additional data given in option two cause more respondents to gravitate to it. This mistake is very common and is called the conjunction fallacy. People feel like more details point to plausibility, even though the opposite is true.
You can take advantage of this flaw in human reasoning if you're ever stuck in a long line. In a 1978 study, researchers Langer, Blank, and Chanowitz found that people in line at a Xerox machine were more willing to let someone cut in line if s/he said, "Excuse me, I have five pages. May I use the Xerox machine because I'm in a rush?" than if the person said, "Excuse me, I have five pages. May I use the Xerox machine?" Being in a rush wasn't new data -- why else would they ask to jump the line -- but the perception of more data, explanatory data, influenced the people in line to accede to the line-skipper.
Human reasoning is littered with similar flaws -- the detritus of evolutionary pressures. Overcoming our biases is difficult, but learning where we're weakest is a good first step. The IBM team did an excellent job programming Watson. Maybe their next task should be reprogramming us.
Follow Leah Anthony Libresco on Twitter: www.twitter.com/leahlibresco