Day 1: IBM Watson Ties for first on Jeopardy • Hacker News Board

Re: Day 1: IBM Watson Ties for first on Jeopardy

#41

Post by derwiki » Tue, Feb 15, 2011, 4:36 PM UTC

lylejohnson wrote: post #38

toddh wrote: post #25

The longer the question the longer Watson has to generate an answer. Isn't that an important advantage?

I'm not sure I understand your point. All of the contestants "see" the answer at the same time, and none of them are allowed to buzz in until the magic light goes on.

But if the question takes 20 seconds to read, Watson has 20 of his seconds to narrow down his confidence to a particular answer. If he only had 10 seconds, his calculations might not have gone far enough to bubble the correct answer on top. It would be interesting to see a chart of confidence of answers vs time for Watson..

Re: Day 1: IBM Watson Ties for first on Jeopardy

#42

Post by spicyj » Tue, Feb 15, 2011, 4:39 PM UTC

tocomment wrote: post #37

Has anyone considered building an open source Watson?

Not very many people have a state-of-the-art supercomputer in their basement, so it wouldn't be that useful.

Re: Day 1: IBM Watson Ties for first on Jeopardy

#43

Post by kgermino » Tue, Feb 15, 2011, 4:41 PM UTC

e40 wrote: post #39

Spoiler not appreciated by many, I'm guessing. Next time, please don't do that.

I'm confused. Where's the spoiler? I don't remember seeing anything in the article (and I'm glad I didn't I wouldn't appreciated the spoiler).

Re: Day 1: IBM Watson Ties for first on Jeopardy

#44

Post by ubercore » Tue, Feb 15, 2011, 4:41 PM UTC

I'd like to see the next version take the avatar concept further. They went half way by making it physically buzz in. Go the rest of the way, and make it only receive input from the avatar. Add a computer vision and speech recognition component, and make it read the board and listen to Trebek like the human players. Then, make it mobile, and put the avatar in the actual Jeopardy! studio.

EDIT: Clarified "human" players. Welcome to the future!

Re: Day 1: IBM Watson Ties for first on Jeopardy

#45

Post by dfan » Tue, Feb 15, 2011, 4:46 PM UTC

derwiki wrote: post #41

lylejohnson wrote: post #38

Earlier quoted context omitted.

I'm not sure I understand your point. All of the contestants "see" the answer at the same time, and none of them are allowed to buzz in until the magic light goes on.

But if the question takes 20 seconds to read, Watson has 20 of his seconds to narrow down his confidence to a particular answer. If he only had 10 seconds, his calculations might not have gone far enough to bubble the correct answer on top. It would be interesting to see a chart of confidence of answers vs time for Watson..

But the same is true for the human contestants; at this level especially, I'm sure that they scan the whole question in a second and then have the rest of the spoken time (which, even for long questions, is nowhere near 20 seconds - probably more like 4) to think about it.

Re: Day 1: IBM Watson Ties for first on Jeopardy

#46

Post by sharmajai » Tue, Feb 15, 2011, 4:47 PM UTC

asuth wrote: post #17

powrtoch wrote: post #8

"Watson can’t adjust its answers to what the other players say and so it simply answers with whatever comes up as its top answer." This is false, in fact they specifically mentioned Watson learning from the answers of the other players. In the game, Watson answers "The 1920s" after Jennings answers incorrectly with "the 20s". Basically, it didn't write off its own answer because it thought Jennings' was different eno…

I was at an event at MIT with one of Watson's designers as I watched the show. I asked him about the 20s thing, and he said they talked about it when they were designing Watson but they figured the scenario wouldn't happen where another contestant gave the wrong answer and Watson gave the same-but-different wrong answer. Edge case!

Capt. Ed Murphy was a smart man.

Re: Day 1: IBM Watson Ties for first on Jeopardy

#47

Post by umjames » Tue, Feb 15, 2011, 5:14 PM UTC

If Watson wins, it should have to keep coming back as defending champion until someone else beats it. It should also qualify to return in the Tournament of Champions if it gets that far.

Now that would be a true test.

Re: Day 1: IBM Watson Ties for first on Jeopardy

#48

Post by jluxenberg » Tue, Feb 15, 2011, 5:18 PM UTC

Does anyone know why Watson was unable to buzz in immediately in some cases when he had 97% confidence in his answer? Seems odd that a computer would not be able to beat the human opponents at hitting the buzzer.

(It's possible that he didn't have the answer until after the human opponents, but that seems unlikely.)

Re: Day 1: IBM Watson Ties for first on Jeopardy

#49

Post by nyellin » Tue, Feb 15, 2011, 5:27 PM UTC

trafficlight wrote: post #29

nyellin wrote: post #13

Does anyone want to create a community site for the contest? I am looking for another Django developer who can participate in an all-night code sprint. The site would: 1. Show an online score board. 2. Track Watson's performance (like shkb is doing on Google Docs) 3. Aggregate pictures, videos, and written analysis of the event

I think you should've been doing this on Sunday...

I didn't think of it until now. I assumed that the official IBM site would have a way of tracking Watson's progress.

Re: Day 1: IBM Watson Ties for first on Jeopardy

#50

Post by gojomo » Tue, Feb 15, 2011, 5:31 PM UTC

That Trebek gave Watsom credit for "Maxwell's silver hammer" when the clue was looking for the person – Maxwell – gave me a moment's pause.

Jeopardy does tend to be a little more forgiving of contestants during the first round – such as giving them a reminder to phrase-as-a-question, and letting them correct an initially wrong or incomplete utterance, if done almost instantly. Still, I'm not sure if such a interpretive error by a person would have been overlooked.