Am I the Only One Getting Annoyed by the Jeopardy Challenge?

63 comments on “Am I the Only One Getting Annoyed by the Jeopardy Challenge?”

1. What do you think about the speculation that we are a few short years away from an actual thinking computer–that once they have a memory capacity equal or greater than the human brain they will gain actual cognizance?
   .
   I guess the next question would be, can Watson learn from its mistakes? Not just that Toronto was a stupid answer but WHY it was a stupid answer and how it could avoid making that kind of mistake in the future.

   1. Define “memory capacity” for me first. Data storage? Check. Ability to do calculations? Check.

      Critical thinking and language processing? …yeah, not so much. (That was what Watson was actually designed to test, and frankly, though it is impressive, the Final Jeopardy! question demonstrates the problem.)

      As to learning–try putting “why it was a bad answer” into terms a computer can understand, would you? (For that matter, Watson can’t even “learn” at all. In a question yesterday, it repeated an answer that had just been given, because nobody thought to give it the capacity to rule out answers already guessed. Not that this would have been easy to accomplish…)

      1. Actually, it would’ve been extremely easy to give Watson the capacity to rule out already guessed answers. Note that Watson doesn’t have any visual or auditory input abilities; the clues are sent to it via a text file. Just add an input where someone quickly types the wrong answer, tagged as such rather than a clue, and have Watson remove that answer from consideration. Admittedly, this’d take up enough time that it’d probably only be useful if the third player either doesn’t buzz in or also misses it.

   2. I agree Peter. I think the questions are purposefully phrased so that Watson can recognize them. Also, human reaction speed is a little slower than electromechanical switches.
      Bill, I believe the work they’re doing is outstanding, however, when Deep Blue beat Kasparov (a controversy in itself), it was accepted that IBM had mastered human logic. It was discovered later when they made the software public that the computer was drawing from a “book” of moves, not intuitive thought.
      I hope day three is more challenging and more on the level of Jennings’ and Rutter’s difficulty. We’ll see tomorrow.

      1. Actually, according to a former student of mine who heard one of Watson’s designers speak, the contract w/ Sony expressly stipulated that the writers were not to be told that a particular game was for Watson; questions were as normal and a “neutral third party” (her words) picked which game would feature Watson.
         .
         If people are interested and I get my ex-student’s permission, I’ll post about another half-dozen tidbits she told me. It’s all pretty neat!

      2. Yes, very interested. Post what you can!

      3. Um, just about all grandmasters draw from opening books. Up to a certain point, the first N moves of a high level chess game are usually done from memory with respect to well-established opening lines. Admittedly the grandmaster understands why those are the right moves at a deeper level than Deep Blue, but it’s still the same basic thing.

         Watson is using a mechanical device to push the buzzer button. I’m not sure how they’re notifying it that it can buzz in though; Ken and Brad are working off a combination of lights lighting up on the sides of the board and getting into sync with the style of whoever is hitting the buzzer unlock button behind the scenes.

   3. As Dr. Michio Kaku of PHYSICS OF THE IMPOSSIBLE points out, we’ve been “a few short years away from an actual thinking computer” for decades now. Given we’re still not entirely sure how the human brain works, I’m not seeing us duplicate it in a machine any time soon. Oh, it’ll happen. Of this I have no doubt, but whether any of us will live to see it, that I won’t make book on. And, yes, the extended infomercial format would have had me switch off fairly quickly, assuming I’d been watching in the first place.

      1. Could Deep Blue play anything other than chess without a massive reprogramming? Could Watson play Wheel of Fortune, or Family Feud without massive reprogramming?

         When a machine like that can “switch gears” like a human, then I’ll be a little more impressed…

         Sure the internal search engine and processing all the data and looking through the database fast enough to come up with the answers was cool, but so far from “human” it’s not really a competition…

2. I’m being disappointed by how much of the game is Watson being able to buzz in before anyone else. (I hear from other sources that buzzer control is key, because at the level of play here, you can know IF you know it in plenty of time to decide to buzz or not.)

   That being said, you seem confident that the FJ question was clearly different from the others in a reproducible way, and without having the others in front of me I’m not so sure.

   1. Let me put it this way: By the time they were into Double Jeopardy, I was usually able to guess as the questions were being read whether the computer would know them or not, based purely on the wording.
      .
      PAD

3. I work in the parallel computing field. And honestly, Watson isn’t that big of a deal/technological advance.

   Yeah, it is fun to see the IBM infomercial/laugh as the jeopardy champions get spanked (and spank). But it isn’t anything that couldn’t have been done five years ago.

   I haven’t read too much on this, but from all the literature I HAVE read, Watson has a very simple algorithm:
   The information is parsed (the question is interpreted), then said information is passed to a large number of computational threads that cross-reference other information with a varying number of algorithms.
   The percentage bars are just an indication of how many of the threads came to a given conclusion.

   To put this into simpler terms (and to reference X-Factor :p): it is doing a “Phone a Friend” to a room full of Madroxes. They get given the question, all of them use their own information/thought processes, then they all report what they think the answer is. And if enough of them think the answer is “spoon”, then the answer is probably spoon.

   The ONLY complex parts of this are:
   Parsing the question (and Wolfram Alpha is evidence that this is already largely feasible)
   Collecting all the responses (and this is “trivial”, for all intents and purposes).

4. Well, I suppose it depends on what you mean by “cognizance.” Even if a computer does have a memory capacity equal to or greater than the human brain, will it be able to write it’s own programming ? Will it be able to do things it wasn’t programmed to do, on its own initiative?
   As Peter points out, so far Watson hasn’t demonstrated an ability to reason, just an ability recognize the human voice and very rapidly cross reference information to come up with a statistically weighted response. It sounds like a very sophisticated talking encyclopedia (the incorrect use of which could still cause you to fail your geography report, given that it doesn’t know that Toronto isn’t in the US).

5. I keep flashing back to TERMINATOR: THE SARAH CONNOR CHRONICLES, where a machine that could learn to play chess was integral to the eventual creation of Skynet. I wonder where Watson figures in to that. And how cool would it be if one of the questions involved the TERMINATOR franchise?

   I don’t mind IBM tooting their own horn — but I hate how long they’re taking to do it. Normally when a person or organization has something to plug on JEOPARDY it’s integrated into the questions (as when Oprah appeared to give the questions about her show, or museum questions are set at the museum). With the length of the IBM plugs, it’s taking three days to play two games. I’d rather have seen three separate challenges, with a final three-day total.

   1. I read an article where they interviewed the two contestants, and one of them actually made a reference to the Terminator movies after the humiliating defeats during rehearsals. He made a joke about how he was worried that one day a machine from the future would travel through time to come after his progeny and claim the ultimate victory.

      I particularly was interested in Watson’s future applications. IBM failed to explain in great detail the potential of the technology. It seems to me the language/communication advancement in machines would clearly lead to the new super salesman, telemarketer: The new and improved Robo-Dialer that could actually respond to the other end. Or perhaps, the new outsource of customer support.

      “Hello, Dave. My name is Hal. How can I help you today?”

6. Peter, did you see the NOVA documentary on Watson? It was about the programmers’ climb to program Watson so that it could hold its own against humans. Throughout most of the hour, its performance was fairly pathetic, literally eliciting laughter with some of its ridiculous answers. It could not, for example (IIRC) take into account the category, something crucial when a human arrives at some of the answers, and it did not (at first) note when one of its two competitors gave a incorrect answer, so it would sometimes give the same incorrect answer itself afterwards, which was funny. The sense of continuity created by seeing the documentary and then the actual Jeopardy! episodes (I only saw tonight’s because I didn’t realize it was on this week) made it a bit more impressive.
   .
   That said, I agree that repeated shout-outs to IBM seemed annoying. Maybe that’s precisely the flipside to having seen the doc, since I’ve had my fill of those IBM interviews, whereas someone else has not, but even so, seeing only one full round of the game is annoying. One of those shout-outs was enough.
   .
   And how the hëll does Watson decide what to wager during Double and Final Jeopardy?

7. PAD: “It’s a pocket calculator with voice recognition.”
   .
   It doesn’t even have voice recognition. All the answers are in a text file that is fed to the computer as Alex reads them.
   .
   I enjoyed the first day of it, though it was obviously an IBM commercial. It is impressive, mistakes and all. I didn’t bother to watch the second day because I figured the padding on the second day would mostly be a repeat of the padding from the first day.

8. There was an hour-long PBS documentary last week about the development of Watson, and how it’s been being tested over the last couple of years by several Jeopardy matches.

   The problem of abstract and complicated questions was a big hurtle, with Watson having more difficulty with questions that had a variety of key words. I knew Watson would struggle with that Final Jeopardy question because in it you had mentions of Cities, Airports, World War II heroes, World War II battles…it would be compiling lists of all these categories and trying to guess what it was being asked to find. Where the heck Toronto came from, I’m not sure (unfortunately I don’t know much about Toronto airports, but they may possibly be WW II related).

   But to answer Bill Mulligan’s question, the documentary showed that Watson was eventually programmed to learn from its mistakes. When it struggled with an abstract category, it would note the correct answers and use the pattern of their common links to figure out what it was supposed to be answering. If they had had a whole category of U.S. Cities clues that were equally as complicated as the Final Jeopardy one, it may have botched the first couple, but would eventually at least learn that the correct response had to be a U.S. City and start guessing correctly.

   The biggest unfair advantage in this Jeopardy challenge is that it receives the clue as a text message at the same time as its opponents are given the clue. Which means while the humans are busy reading and listening, it’s already “read” the clue and is already searching for the correct answer with a couple of seconds head-start (plenty of time for a computer).

   All that being said, I’m enjoying this challenge, maybe partly because I enjoyed the documentary and am seeing these games as a culmination of all the challenges that were outlined in that documentary. I think Watson is an amazing achievement.

   Plus, I love seeing the mistakes it does make, or similar questionable actions it takes (I loved how the audience laughed at that initial bizarre Daily Double wager, a theme that continued right up until Final Jeopardy).

   And seeing as how they’re going to do the last game all in one episode, I trust the “infomercial” stuff is over, and we’ll see a straight-up match like we would in a normal show. I’m looking forward to it.

   1. “already searching for the correct answer with a couple of seconds head-start (plenty of time for a computer).” Not really. As of a couple of years ago, Watson was correctly answering 80% of the clues…it just took it several minutes for each one. Note that what’s difficult for Watson isn’t doing a knowledge base lookup; it’s figuring out what the clue is actually asking for, particularly given J! clues have a lot of oblique references and wordplay.

9. I was very excited to watch these matches. And I too have been somewhat disappointed that it’s taken two days to see one complete game. I can sort of see how that happened – they wanted to have some explanation of Watson, a normal Jeopardy game doesn’t have 5 minutes to spare, so they decide to split the first game over two nights and once you do that you have to fill the time somehow. Personally, I would prefer that they had replaced the normal commercials with the IBM promotional stuff (since they are essentially commercials), but for whatever reason they didn’t go they route.
   .
   My real reason for writing, though, is to point out that Watson is hardly “Google 9000”. The problem they are trying to solve is a much different one than the search problem. For example, take the Beatles category yesterday. Watson did very well with it, and at first glance it seems like all you needed was Google – just take the quoted song lyric, search for the rest of the words, and pick out the name that fills in the blank. But actually getting the correct answer is much harder than that, because one needs to figure out where the blank is. The name could come before the quote, in the middle of it, or after, and there could be one or more unquoted connecting words in between. Getting the correct answer requires understanding a little of the context in which the quote was presented, and presumably some understanding of the category to know that what is required is a person’s name (which, in turn, requires knowing what makes something a person’s name). All of these things sound trivial because we do them all the time without thinking, but it’s actually fairly complex to get a computer program to do it.
   .
   Of course, it still isn’t perfect and the main challenge going forward is probably solving the problem that when it is wrong, it can be very wrong in ways that even children wouldn’t be – such as answering Toronto in the U.S. cities category or failing to come up with answers that were decades for the decades category yesterday. The solution seems easy – program it so that if the category is US cities, make sure the answer is in a list of US cities. But how do you generalize that to all possible Jeopardy categories? Then it’s not so straightforward.
   .
   So, all of that to say that, while it may not be the compelling television event they were hoping for, Watson does represent a significant advance in its field and will likely bear fruit in some of the other domains mentioned during the IBM infomercial segments.
   .
   (Full disclosure – I have nothing to do with Watson or IBM. I did do a postdoc in a university department where other people were working on problems related to what IBM is trying to do with Watson, so I did develop a rudimentary understanding of the challenges and a sense of sympathy for the difficulties they presented.)

   1. Oh, I dunno. I just went to Google, entered “Beatles” and then tried different song lyrics and titles, leaving out key words in different places. Google instantly came up with the entire lyric, filling in the missing word with no problem.
      .
      PAD

      1. Perhaps I didn’t explain the difference clearly, but what you are talking about is a very different task than what Watson has to do.
         .
         In your test, you give Google some of the words to a Beatles song (and the keyword “Beatles”), and it returns the full lyric to the song. That is a basic search problem.
         .
         On Jeopardy, Watson has to take a clue which contains Beatles lyrics as well as extraneous words that are at best about the song rather than from the song, and from that it has to figure out which words are the song lyric, which words are not, find the full lyric, and then (this is a critical difference) figure out which word(s) from the remaining lyrics to return as an answer.
         .
         So, if Google was playing Jeopardy, it would give back the full lyric which would be incorrect because the category specific just the name of a person. Watson, however, was able to single out just the name which was the correct answer.

10. Only the infomercial aspect bothers me. i think, though, given the way things are playing out, it should have been, Ken, Brad,… and Ben Stein. (At least it would be a slightly more level playing field)

11. I agree Peter. I think the questions are purposefully phrased so that Watson can recognize them. Also, human reaction speed is a little slower than electromechanical switches.
    Bill, I believe the work they’re doing is outstanding, however, when Deep Blue beat Kasparov (a controversy in itself), it was accepted that IBM had mastered human logic. It was discovered later when they made the software public that the computer was drawing from a “book” of moves, not intuitive thought.

    I hope day three is more challenging and more on the level of Jennings’ and Rutter’s difficulty. We’ll see tomorrow.

    1. Phrasing the questions to make them easier for Watson would completely defeat the purpose. They chose Jeopardy because it is not easy for a computer. IBM wants the questions to be challenging, because the domains where it wants to apply this technology are at least as difficult as the Jeopardy challenge.
       .
       Also, you criticize Deep Blue for using a “‘book’ of moves”. However, it’s not at all clear that humans aren’t doing exactly the same thing when they play chess. There is evidence that grandmasters are successful at least in part because they have played so many games that they can recognize the patterns in a wide variety of board positions and know from experience which moves are favorable or unfavorable from those positions. That is very similar to how Deep Blue plays chess.

12. If you saw the Nova documentary on Watson, and how pathetic his performance was in preliminary mock Jeopardy! games, then seeing his performance this week on the show proper was impressive. In the doc, he couldn’t take the category into consideration, he would give an incorrect answer even when the same answer had already been given by one of his two co-contestants because he had not been programmed to take that into account, etc. Some of his answers were so ridiculous they literally elicited laughter. I though it was a bit of an overstatement for the programmers to go on and on about trying to make him think like a human, because everything I see him makes him look like just a glorified search engine. The fact that he couldn’t take the category into account on tonight’s Final Jeopardy would indicate that he still hasn’t incorporated that ability completely, if at all.
    .
    I can understand why they’d want to include some expositional background material about him, but c’mon, two different pieces about him? Being able to play just one Double Jeopardy round and the Final round in one episode? That’s just ridiculous. They should’ve devoted a full episode to each game as usual, and provided a link for interested viewers to learn more about Watson.
    .
    My father was shocked that Ken Jennings “didn’t know” all those answers. I had to explain to him that of course he knew them (he won 74 games in a row, for crying out loud), he just wasn’t buzzing in in time. Timing with the buzzer has to be perfect: You can only buzz in when Alex is done reading the clue, and if you buzz in too quickly, your ability to do so again is delayed by one quarter of a second. This is one reason why Watson dominates the game. It’s largely luck that Jennings managed to ring in about twice, while Brad Rutter didn’t do so even once.

13. i dunno i’ve been enjoying it.

    Darwin vs Watson…yea i’d watch that

14. True, I haven’t seen this much padding in a game show since the episode of Greed where Dan Avila took his team all the way to the $2million question (something that personally affected me, as I was in the green room waiting for that team and the next one to finish so I’d get to play. : -)) Still, I did like seeing a former grad school classmate and co-worker (Ed Hovy) make a brief appearance as an expert commentator.

    Also, Peter, as both a former Artificial Intelligence researcher with experience in natural language understanding research and a former Googler, what Watson’s doing has very little to do with what Google’s doing (currently; if they haven’t started a Watson-equivalent project, I’ve got words for Larry and Sergey the next time I run into ’em). What’s really interesting and amazing about Watson is that it’s able to “understand” the clues as well as it’s done. Few J! clues are as straightforward as “What was Abraham Lincoln’s birthdate?”. Instead, they have oblique references, wordplay, and require combining multiple bits of knowledge from very different areas.

    This is a major leap from what was previously considered possible in terms of NLP (natural language processing). Admittedly, it’s not so advanced that Watson could have a conversation with someone (natural language generation is actually a very different, equally hard, problem), as it’s limited to trying to figure out what piece of information is being requested, rather than fundamentally understanding the words and putting them in a timeline session context, as would be needed for having a real conversation. I’m impressed as hëll by this. The aforementioned Ed Hovy is head of the natural language research group at USC-Information Sciences Institute and a very sharp guy, and he’s impressed with it.

    As a side note, Ken’s a comics fan, and read and posted a few times to the rec.arts.comics newsgroups. As it happens, I met him several years before he was on Jeopardy! when I helped out with a Bay Area quiz bowl tournament his BYU team played in. During his 74-game run, I asked him via the J! message boards if he might have played in any of those tournaments. He replied that he had, and recalled meeting me since I’d worn a Suicide Squid t-shirt to it. I’ve since met him again in person a few times, including when he spoke at Google (http://www.youtube.com/watch?v=6YMRVwdeHB4 he’s interviewed by then-Googler Fritz Holznagel, who himself won a J! Tournament of Champions; the back of my head is in the lower left hand corner at the start, and I’m the one who asks the first question during Q&A about the giant styrofoam head), and he’s a nice, sharp, guy. Never met Brad, but mutual friends say the same about him.

15. You know why those screwy letters that we are asked to type when logging in to sites work.
    Because humans can easily see an “a” or “b” in a vast variety of styles and contortions. Computers find it close to impossible.
    Just saying.

    1. Those used to be difficult for computers. Most of those “captcha” systems have since been solved so that a computer can identify the letters correctly also.

    2. Well, they’ve gotten better at that, as people that have had to administrate guestbooks, blogs or message boards will attest… I am in the process of setting up a forum, and have a guestbook on my site, both use a CAPTCHA to filter out spam bots. Both have been cracked by said bots. I had to set up a robots.txt file to exclude my guestbook from search results (they search for known guestbook/forum/blog software which they have the ability to fill in the blanks for. If they can’t find it, they can’t hit it.) and every once in a while, one still gets through, somehow. With the forum, I had to install a picture based CAPTCHA after several text based ones failed miserably at stopping the flood of spam bots. The forum isn’t even officially open yet, and I was getting several new signups a day that were nothing but spam bots. Added the picture based CAPTCHA called Peoplesign, stopped them cold.

16. I get the feeling that, somewhere, an IBM programmer had to be talked off a ledge when he heard that Watson gave the answer “What is Toronto?”.

17. Anyone else thinking of the late Tracy/Hepburn film, “Desk Set?”

    1. Oh, yeah. “That’s a wrong answer. Peg, King Solomon’s Mines, Sylvia, Corfu. Let’s show ’em what humans can do…” And now I think on it, WATSON is really a tenth-generation EMARAC. And about as smart.

18. I’m picturing Watson in a bar, hunched over a dozen empty shot glasses, with “I coulda been a contener, not a bum, which is what I am” scrolling across it’s screen…

19. Personally, I wouldn’t have minded the “infomercial” aspect of the first two installments as much if they’d just explained the technology more clearly for those of us who are only semi-literate in computer theory. From the descriptions we’ve been given, it’s very difficult to gauge how much of “Watson” is hardware and how much is software — and that’s a key issue in economic terms as well as in technological context.

    The impression I had from Alex’s comments at the end of the Tuesday installment is that we’ll get a second complete game in the third and final show. That, one hopes, will improve the pacing. As for the gameplay, it’s worth noting that Watson had the supreme good luck to hit both Daily Doubles in the Double Jeopardy round, which is the main reason the scoring is as lopsided as it is at this point.

    1. What’s interesting about Watson is the software. But the amount of hardware it needs to run on in order to parse the clue and come up with an answer in a few seconds is significant. Right now, Watson isn’t commercially feasible for mass use (which is why IBM isn’t putting up a Google killer website as I write). But with Moore’s Law, sometime in the next decade max it will be, as computing power goes up and its cost goes down.

       1. I rather thought the software had to be the primary breakthrough. At the same time, Alex’s “backstage tour” gave a really mixed impression of the hardware — on one hand, the sheer size of the setup recalls old-fashioned traditional supercomputers (a la Cray and suchlike), but the handful of tech specs Alex rattled off made it sound more like a cross between a large-ish office network and a small-ish server farm, both built with more or less off-the-shelf servers.

          Either way, of course, you’re right that Watson-clones aren’t yet ready for retail….

20. For me, it’s been interesting to see where Watson’s programming succeeds, and where it fails. It cleared the board on Double Jeopardy yesterday, but did poorly on several categories on Monday (most notable being the “Name the Decade” category). It also recognized that Mad-Eye Moody, Severus Snape, and someone else were characters from Harry Potter, but not that Voldemort was the one who killed them.
    .
    And, of course, there was that Final Jeopardy flub.
    .
    As I commented elsewhere, Watson clearly has information and statistical analysis. What it doesn’t have are a grasp of context, intuition, real understanding of what it “knows”, or the ability for fluid learning.
    .
    Today is a full game, no documentary stuff. Should be fun.

21. Can’t watch it because of work, but have a question. How does Watson press the buzzer? And does it answer in form of question?

    Thanks

    1. Watson has a mechanical piston that presses the buzzer.
       .
       Yes, it answers in the form of a question.

22. Here’s a tangentially related question: How soon can a human buzz in in JEOPARDY? Is it the moment Alex Trebek finishes reading the question? Is it anytime after the question is shown — and if not, is there a penalty for buzzing in too soon? I watch the show daily, and I’ve often seen someone frantically clicking on their buzzer while someone who didn’t seem to move gets to answer first.

    And while I’m enjoying the Watson challenge, my favorite was when they used categories from the SATURDAY NIGHT LIVE sketches. Yes, they actually came up with questions for things like “The Color Is Green,” “Write a Number” and “Current Black Presidents.”

    1. It’s the moment Trebek is finished reading the question. And if one buzzes in too soon, then they are delayed from buzzing in again for one quarter of a second. One reason Ken Jennings did so well when he was a contestant is that he practiced with a buzzer (though I’m curious as to how he did this–did he rig up a homemade buzzer?) After he told this to the producers, one of the changes they implemented was to give contestants some backstage practice time to get comfortable with them.
       .
       I imagine that this is a significant reason why Watson is dominating the game: It’s not that Jennings doesn’t know the answers (He didn’t win 74 consecutive games by not knowing the answers to a many of the clues), it’s probably because precision timing is the stock and trade of a machine like Watson. My guess is that those two or three times that Jennings managed to ring in before Watson on Tuesday night’s episode (I missed Monday) was either sheer luck, or that maybe Watson needed a few more fractions of a second to process the answer.

       1. I forget how Jennings practiced specifically, but two common practice techniques are a) hold a ballpoint pen and click it, or b) record the program and use the remote’s pause button as the buzzer. (FWIW, the program calls it a “signalling device” but is aware that no one thinks of it that way.) The pen is closer to the feel of the actual device, while the remote lets you try answering the question before the contestants.
          .
          A friend of mine was on the show and came in second; she agreed that the returning champion’s greater familiarity with the buzzer gave him a slight edge (getting a Daily Double on the topic of a book she hadn’t read and an unusually difficult Final Jeopardy didn’t help, of course).

    2. Mentioned above – you have to wait until Alex finishes reading the clue, and if you buzz in too soon, your buzzer won’t respond for another quarter of a second.

       1. Ninja’d by Luigi – he must have buzzed in first…

    3. Alex reads the clue. A light goes on. Players are allowed to ring in. Press your button too soon and you’re locked out for a fraction of a second.
       .
       That’s why you sometimes see a player repeatedly hitting the button to no avail. They initially hit it too soon and are repeatedly locking themselves out. (Hard to calmly wait a fraction when you really WANT to ring in.) Meanwhile, the slower person gets to answer because they didn’t lock themselves out.
       .
       Another thing to not on this arrangement: It’s not just Trebek’s timing/cadence on reading the clue that players have to get the rhythm of. Because there’s the manually controlled switch that turns on the light that indicates players can now ring in. That person’s reaction time figures into the timing, too.

    4. You can’t buzz in until the question is finished. I’m not sure if there’s any kind of penalty. I want to say that there’s a 1 second delay before you can buzz in again, but I might be pulling that out of nowhere.

23. I keep thinking about an episode of The Paper Chase called ‘The War of the Wonks,’ in which Professor Kingsfield challenged the Harvard mainframe to a battle of wits. Not surpringly, the computer did exceedingly well when it came to analyzing masses of data but failed miserably as far as answering a hypothetical that required an additional level of thought and consideration. Sort of like what’s happening now I guess.

    1. In the movie THE PAPER CHASE one student had a photographic memory but failed out of law school because while he could recall everything, he couldn’t necessarily understand or apply it. (I think his character stayed in law school on the TV series.) This was also true on this week’s HOUSE: A patient with total recall worked as a waitress, and when one doctor asked why she wasn’t working for NASA, the patient said she remembered everything but that didn’t mean she could understand it.

       Watson may have had almost instant access to a massive database, but it sometimes clearly had problems understanding what the question was looking for (esp. in last night’s Final Jeopardy). Still, it was the clear winner.

24. Pingback: Tweets that mention Am I the Only One Getting Annoyed by the Jeopardy Challenge? | PeterDavid.net -- Topsy.com

25. I didn’t get to watch the episodes because I worked late the last three nights and didn’t know that the Watson games were on this week until I read this post. I do think it’s interesting that they named the computer Watson while they’re working the bugs out of it. When they bring out the Holmes computer, that should be when things really get interesting.

    1. No, the HOLMES IV won’t be installed until we get the lunar penal colonies up and running – then again, it wants to throw chunks of lunar regolith at Earth. I think I’d rather watch it play Jeopardy.

    2. It actually was named after IBM’s founder, Thomas Watson, rather than the Holmesian character.
       .
       Which is a shame; I guess they just didn’t have the nerve to name it HAL.
       .
       PAD

       1. Perhaps they didn’t want to inspire Ken Jennings to get into the server room and start pulling out circuit boards.

       2. I’ve heard there were some IBMers who wanted to give Watson HAL’s voice. And others who wanted to give it the voice of the Sean Connery impersonation from SNL’s Celebrity Jeopardy!.

       3. Faux-Connery would have been amusing for a minute, but HAL’s voice would have been awesome. Is Douglas Rain still alive?

26. Too bad they can’t use the supercomputer from Harlan Ellison’s “I Have No Mouth And I Cannot Scream,” of course no one would be left to watch such a thing. 🙂

27. Correction: “I Have No Mouth And I Must Scream” and we see how well I’d do on Jeopardy. Sighs. 🙁

28. They did mention on Nova that the name Watson is from the founder of IBM, not Holmes’ doctor buddy.
    Although I was also sorry they split the first game between two shows I just reminded myself there were still more questions asked than in any episode of Who Wants to be a Millionaire or Are You Smarter than a Fifth-Grader?
    Jennings played quite a good second game considering by that point the two humans were playing for second place and his “computer masters” line in FJ was fun.

29. Pingback: Watson Beats Ken Jennings on Jeopardy

30. I wonder if this series of games must have provided a certain amount of temptation for the writers:

    “In 1960s science fiction, robot speak for ‘I don’t understand’.”

    “What is, ‘Does not Compute’?”

31. A cartoon showing a Jeopardy! board full of topics Watson wouldn’t like: http://hijinksensue.com/2011/02/15/culturally-biased/

Comments are closed.