Kent Pitman

Kent Pitman
Location
New England, USA
Title
Philosopher, Technologist, Writer
Bio
I've been using the net in various roles—technical, social, and political—for the last 30 years. I'm disappointed that most forums don't pay for good writing and I'm ever in search of forums that do. (I've not seen any Tippem money, that's for sure.) And I worry some that our posting here for free could one day put paid writers in Closed Salon out of work. See my personal home page for more about me.

MY RECENT POSTS

FEBRUARY 16, 2011 5:08AM

Just-In-Time Jeopardy

Rate: 13 Flag

“Toronto”? Really, Watson? That's your “Final Jeopardy” answer for Day 2 in the category “U.S. Cities”? You are a farm of super-computers with instant access to multiple terabytes of trivia, totally whomping the humans competing with you, and suddenly you cannot name a single U.S. city? And yet they’re still going to let you win?

No, let me save you buzzing in to protest: The fact that you put a lot of question marks after your answer does not make it better.

Clearly it’s time to suggest a Jeopardy! rules change...

Proposed New Jeopardy! Rule

If the audience moans really loudly at a response by some player, that audience must be polled and if the majority of its members believe the response was Completely Stupid, the other players each get a one-third share of the money that had been held until that moment by the player with the Completely Stupid answer. (This should have the effect of moving the player who made the Completely Stupid move into last place.)

Example

Let’s say, just hypothetically, that a Player A is in “Final Jeopardy” and the category is US Cities. Let’s say that Player A has $600 and Player B has $100 and Player C has $50. If A’s answer is found to be Completely Stupid, then the money A had ($600) is divided in three parts; Player A keeps one of those parts ($200) and each of Player B and Player C receives one part ($200). So after the stupid move, Player A would have $200, Player B would have $300, and Player C would have $250. In effect, this would mean that the player making the Completely Stupid move goes from first to last place.

Benefits

This would mean there was no possibility of a game that was a guaranteed “runaway.” Even the player in the otherwise-unchallenged lead could still make a foolish move at the last minute. It would mean there was always suspense until the end—not to mention justice in the case of someone, or some farm of super-computers, doing something Completely Stupid.

Personally, I think this change would be a perfectly good new rule for the game in general, not just for this exhibition game.


If you got value from this post, please "rate" it.

Click here if you missed the discussion of Day 1
of this unusual exhibition competition.

Your tags:

TIP:

Enter the amount, and click "Tip" to submit!
Recipient's email address:
Personal message (optional):

Your email address:

Comments

Type your comment below:
Kent, I just read your post from yesterday and found it to be very informative. I did not watch Monday's show but did watch yesterdays. I bothered me on a really visceral level. I couldn't help but watch the body language of the two contestants as they got thoroughly trounced. What's interesting to me is that I'm pretty sure they knew all the answers, they just weren't able to click in faster than the machine. Its seems that it would have been a fairer fight if the machine had human 'response time' entered in for the game. That would have made it much more enjoyable. R
I was blown away by the computer not being able to come up with something even resembling a correct (as in, a US city) answer. To me, this created a situation in which the computer, no matter how fast it is, is not a success.

That being said, I was impressed by the speed of the computer in coming up with the answers and I liked the fact that the search possibilities were shown, with their percentages. I do believe I would have been more impressed if it had been using the aforementioned voice to text features widely available. Yes, Watson is a great example of computing and programming, but there is a lot of room for improvement.
Rita, I agree. Paul J. O'Rourke said yesteday, “To me, this is far less about AI and more about a quick solenoid, as far as the competition goes.” Very disappointing in that regard.

If one goes to the New York Times “demonstration” of it (which appears to just be a Javascript app, not really Watson—don't get your hopes up, though it's still fun), you see they understood that a rules change to let the human answer first was needed. But perhaps it would have been most fair because of the difference in capability for humans and machine to simply go round-robin, eliminating any sense that one player had that edge over another. I'm surprised this didn't come up in the copious practice sessions they allege to have had. Maybe they practiced on people who didn't feel “entitled” to win.
CZ, I agree the answering speed was good, but without being able to tell it's answering faster than the other players rather than just outdoing them in timing, it's hard to feel good. By that I mean it may be locked out until the end, but if it can buzz in a microsecond after the end reliably and they have to just guess at the end and buzz in when their brain synapses can do it, then their knowledge is worth nothing. I have no idea if that was the problem but no way to tell it wasn't.
It seems that the two shows aired so far have been thirty minutes of shameless advertising for IBM...and not very entertaining minutes.
Bring back the humans!
R
Steve, I'd lay the blame on the show. I'm sure they could have adjusted the rules as they liked. And I think they probably tested it in advance. I don't know why they allowed the thing click-in speed issue. It certainly must have been conspicuous from the outset.
I agree with Rita and Steve. I'm also bemused by the odd amounts Watson bet on daily doubles -$1,246 on one, if I remember correctly. That seemed odd for an entity and made me wonder if it weren't some of the little 'compromises' worked into it to make it sound less like a machine (more quirky like some of the human contestants).

The far reaching implications of a Watson is what I find terrifying. Imagine a world whose power is held by such a leader who seems infallible, yet at a most crucial instant can make a fatal choice between life and demise of humanity? Yesterday, Toronto - in 2023 ???
I knew that computer had Canadian blood in it..:)
rated with hugs
What would happen if Watson failed to precede the question with "What/Who is/are . . ."? Is that even possible?
It would sure be a lot more interesting.
rated with love
Much ado was made about Big Blue winning at chess, but heretical as it may sound, Jeopardy is in some ways a much more difficult game than chess in that there exists a virtually endless number of "moves (trivia)" in Jeopardy, and those "moves" don' t lend themselves to the sort of rote memorization that makes chess champions and computers a breed apart.

In short, I'm not so sure computers will ever be capable of replacing humans for some kinds of thinking. Computers are useful, and they have their place -- but so do my teeth -- in a jar by my bed.
Tom, you've given me some ideas about something I might put together as a final comment on this competition. I'll mull on that and maybe write more either here or in a real post. Thanks.
Fusun, I think the odd bets are likely based on the probability of winning. If it thinks it has a 93% chance of not getting an answer, its bet is probably correspondingly different than if it thinks it has only a 2% chance of messing up. People have neither an accurate estimate of their chances of being wrong nor a strong command of the math needed to incorporate such information into a bet in most cases, so they just do something simpler involving round numbers and blind hope.

I agree with you that there's an issue of how much to trust all those numbers, though. One really wants to see the rationales in most cases. Though there are exceptions where I think one does not, where things are happening so rapidly that one has to just make a best guess and hope because the odds against succeeding if you have delayed a guess are worse. For example, when landing on an alien planet, one can imagine decisions being made all the time about terrain, fuel use, being batted about by atmospheric effects. And yet it might be there aren't a lot of options. Making a best guess may be the only choice, and taking the time to consider the options may mean a certain crash. Actually, driving a car is a similar thing sometimes. But such situations are unusual. A great many things allow time for contemplation.
Linda, yep, it was just trying to impress Trebek, who is a Canadian by birth and a naturalized US citizen. Now Toronto has followed suit, apparently.

Pedant, I think that not using the form of a question is an automatic disqualification. But I'm sure the program was tuned not to make that mistake. And my vague recollection from when I watched years ago is that Jeopardy is very tolerant of stupid forms like “What is I think there are seven cities?” where literally someone tacks “What is...” onto a statement, so it would be hard to get wrong. If they required you to say “Who is Eleanor Rigby?” instead of “What is...” then it would have been a different matter.

Poetess, thanks for the support. :)
Wow new life for an old game. Thanks for the jeopardy update.