News for hackers. Stuff that matters.
Re:Ask IBM's Watson Research Team Anything by amichail on Thursday February 17, 2011 @05:58PM (#2233286)
Did this competition hurt Jeopardy in general because it is now obvious that the winner is mostly decided by buzzer skills among top players?
Re:Ask IBM's Watson Research Team Anything by nostromo on Thursday February 17, 2011 @06:44PM (#2233431)
With three of the best Jeopardy players ever, I thought they might make the questions a bit more challenging. I can see why this might not work in this case because Watson was programmed using regular questions. But I agree, I would have liked to see the skill level rise with the caliber of player.
Re:Ask IBM's Watson Research Team Anything by Herring on Thursday February 17, 2011 @07:39PM (#2233624)
Landmark AI event, and you guys are worrying about buzzer speed. *sigh
Re:Ask IBM's Watson Research Team Anything by RickHull on Thursday February 17, 2011 @08:15PM (#2233724)
That's because the buzzer competition confounds the knowledge competition. It's a shame, really.

But I'm sure IBM is happy.

Re:Ask IBM's Watson Research Team Anything by JunkDNA on Thursday February 17, 2011 @09:36PM (#2233980)
Can you explain why it is you think that the knowledge part is confounded? If Watson is fast but incorrect, it's a failure.
Re:Ask IBM's Watson Research Team Anything by defen on Friday February 18, 2011 @12:06AM (#2234302)
A Watson that knows 70% of the answers will beat a human who knows 90% of the answers, purely due to better reflexes. Of course, buzzer skill is partly how KenJen won so many games, so I'm not complaining. Just not AS impressed by Watson as some people.
Re:Ask IBM's Watson Research Team Anything by JunkDNA on Friday February 18, 2011 @07:42AM (#2235187)
Except that it's not just reflexes. Watson has to compute the answer before the humans. That's no small task. This video is a great technical explanation of how it works:

http://www.youtube.com/watch?v=3G2H3DZ8rNc&feature=youtu...

Re:Ask IBM's Watson Research Team Anything by RickHull on Thursday February 17, 2011 @10:26PM (#2234100)
The knowledge part is confounded because buzzer performance preempts knowledge performance.
Re:Ask IBM's Watson Research Team Anything by shadowfox on Friday February 18, 2011 @12:51AM (#2234411)
Can you explain? (I didn't quite get the preempt part). Watson seems to be getting answers correct as well
Re:Ask IBM's Watson Research Team Anything by Pyrodogg on Thursday February 17, 2011 @06:18PM (#2233353)
I'm sure the same thing comes out of Jeopardy "Tournament of Champions" games where a number of very good human players come together. It's just that only regular Jeopardy watchers every noticed those games were taking place.

It just seems more pronounced with Watson because it drew in a large crowd of new (but perhaps one-time) viewers.

Re:Ask IBM's Watson Research Team Anything by Kaizyn on Thursday February 17, 2011 @06:15PM (#2233340)
No, I don't think so. It was always known that buzzer speed was always an important factor.
Re:Ask IBM's Watson Research Team Anything by ghaff on Thursday February 17, 2011 @06:30PM (#2233388)
I don't think I realized the degree and I've been an off-and-on regular Jeopardy watcher over the years. I don't see this subtlety as hurting Jeopardy's brand any though. It does make me wish that there had been a better platform for competition of this sort that would have had a more level playing field.

(The humans actually had some advantage of anticipating the buzz-in signal so I'm not saying that Watson had all the cards. I'm just saying that a lot of the contest hinged on a factor that inherently could not be made apples-to-apples.)

Re:Ask IBM's Watson Research Team Anything by tastybites on Thursday February 17, 2011 @07:22PM (#2233567)
I doubt it. Jeopardy's a game, not an exam.