Subscribe to:
Post Comments (Atom)
skip to main |
skip to sidebar
Jeffrey Shallit
Recurrent thoughts about mathematics, science, politics, music, religion, and
Recurrent thoughts about mathematics, science, politics, music, religion, and
Recurrent thoughts about mathematics, science, politics, music, religion, and
Recurrent thoughts about ....
-
Jeffrey Shallit
American mathematician, professor of computer science at a major Canadian university, & skeptic.
All comments are moderated to remove spam and duplication. Please be patient; your comment will usually appear within 12 hours.
Recursivity
Blog Archive
-
▼
2011
(96)
-
▼
February
(11)
- Famous Astronomer with Moose
- "Any" Considered Harmful
- More Mary Poplin Nuttiness
- Waterloo's "Watson" Connection
- 'Watson' on Jeopardy
- Mary Poplin Round 2
- Harris v. Wolpe
- Rock Lobster - Performed by Old Computers & Robots
- The Pascal Lecture: Another Year, Another Embarra...
- World's Dumbest Scammer?
- Fibonacci Quarterly Goes Electronic
-
▼
February
(11)
2 comments:
A detail about Jeopardy that I never really appreciated until Watson arrived is that there are two essential games in Jeopardy: 1) knowing the answer to questions, and 2) being able to ring your buzzer as soon as the producers enable them (apparently there is a short time penalty if you ring in too soon).
Top-level Jeopardy players typically all know 1) pretty well, and so, at the top levels (e.g. Tournament of Champions), it is mainly 2) that distinguishes between winners and losers. Ken Jennings and other Jeopardy champions say as much, and the very best players have a near super-human ability to anticipate just when to press the button in 2).
Watson receives an electronic signal at the exact moment it can ring its buzzer. It will never get the time penalty, and, being a computer, it will have near-perfect reflexes compared to any human.
Obviously the most important and interesting aspect of Watson is its ability to do 1). But in 2) it has a clear --- probably unfair --- advantage over humans. When Watson and the two humans all know the answer, then I suspect that Watson almost wins.
So if Watson wins at Jeopardy it is most likely because of 2). Interestingly, IBM does play this point up, although they are certainly open about how Watson plays 2), and have said they think it gives them a definite advantage. The game show Jeopardy itself does not ever dwell on the fact of how important 2) is in the game, and who can blame them: pressing a (non-nuclear missile launching) button at just the right moment doesn't make for compelling TV.
Nevertheless, it still has to be able to come up with the right answer - that impresses me much more than whether it wins or not.
Post a Comment