Goodbye Search … It’s About Finding Answers … Enter Watson vs. Jeopardy!

Does anyone really like searching for stuff?  It conjures up images of looking through old boxes in the attic to find that one thing you can never seem to lay your hands on.  Recently, I went looking for my junior high school yearbook when someone “friended” me on FaceBook and I couldn’t remember them.  The experience was exasperating. I looked through at least 20 boxes of stuff, started sneezing from the dust, and never found the darn yearbook.  As a result, I am still not sure I was actually in the same science class as this person.  The experience reminded me of today’s enterprise search limitations.  I blogged about this recently as part of my Top 10 Pet Peeves for 2010

If you think about it … no one actually likes the searching part.  It’s no fun nor is it intuitive.  You have figure out a “query” or “search string” and hope for the best.  Maybe you’ll get lucky and maybe not.  It’s what I call the “search and hope” model and it can be even more frustrating then my attic experience (I feel a sneeze coming on).

In an AIIM Industry Watch Survey earlier this year, one of the key findings was 72% of the people surveyed say it’s harder, or much harder, to find information and documents held on their own internal systems compared to the Web.  That makes you scratch your head for sure.

In the end, no one “wants” to search anyway … it’s the thing we seek that we care about, and not the searching process.  All I wanted was an answer to my question, which was to see if I could remember this former classmate.

IBM has been working at systems to find answers since the 1950s when the first steps were taken with research on machine based learning.  Over 50+ years (and many millions later), we have history being made.  An IBM computing system (Watson) will play Jeopardy! live on television against Ken Jennings and Brad Rutter, the two all-time most successful contestants, in a series of battles to be aired February 14-16. The series will feature two matches to see if a machine can compete by interpreting real-language questions, in the Jeopardy! format, by using text analysis (natural language processing), automated classification and other technologies to find the correct answers.  Here is a brief overview to Watson.

Watson must find the answers in the same timeframe as the two former champs by processing and understanding the question, researching the possible answers, determining the response and answering quicker than the two former champs … plus it has to be right. WOW!

Jeopardy! is the No. 1-rated quiz show in syndication, with more than 9 million daily viewers. Watson has already passed the test that Jeopardy! contestants take to make it on the show and been has warming up by competing against other former Jeopardy! players.  The top prize for the contest is $1 million, $300,000 for second and $200,000 for third. Jennings and Rutter plan to donate half their winnings to charity.  IBM will donate all winnings to charity.

I can’t wait to see this. I suspect my fascination has to do with my being involved with content analytics as part of my job at IBM.  Or maybe it’s just about the coolest thing ever.

Either way, finding answers sure beats searching and hoping … and this ought to be very very interesting.

Here is a deeper explanation of the DeepQA techology behind Watson for those who are as fascinated by this as I am.

24 thoughts on “Goodbye Search … It’s About Finding Answers … Enter Watson vs. Jeopardy!

  1. I had been hearing snippets of this upcoming event, but it is not getting a lot of (understandable) press here. (Does Watson speak German?) Speak for yourself, though, I love the search. LOL. Let me know if you ever find that yearbook – mine is in storage in the US and there are a few “friends” I’m not sure I remember.

  2. Earlier this year the N.Y. Times did a piece on Watson and had a link where you could go to play a game. Fascinating to see how the decisions were made. I do wonder if Watson will be forced to know the answer before ringing in, or if Watson will be allowed to ring in without 100% knowing the answer at the time of the ring, similarly to how the human players do it sometimes.

  3. Craig,
    I concur with the comment that it is about questions though questions trigger conditioned response processes whether they are related to a ‘friend’ request or E-Discovery project.

    In E-Discovery starting from the data rather than relationships increases cost, time and risk.

    E-Discovery vendors, project managers and services providers offered conditioned responses like going to a surgeon provides a recommendation for surgery.

    My previous post was a gentle way to startle readers out of conditioned responses and consider the end result of an E-Discovery.

  4. There’s a recent article on that explores analytics vs. process oriented traditional search. It seems like sooner or later analytics will be more widely adopted, but there’s alot to be said for a well thought out process both at the search stage and the a little further upstream at the collection stage. Here’s the link to the article for the group –

    1. Pawel – The technology under the covers (for Watson) is detailed on the IBM website and on the YouTube videos … it leverages (in various ways) NLP, text analysis, classification, semantics and decision tools to highlight some. While I agree with your post, it assumes we desire to search for things.

      My question and posting is a little more fundamental. We desire answers, don’t we? We really don’t desire to search. It reminds me of the stereotype that men are afraid to stop and ask for directions. Saying we enjoy searching, is like saying we enjoy being lost. I presonally would much rather get to where I am going in the shortest time possible.

      1. Yeah, you’re right. But I rather see it from technological point of view. So even if it’s all about answering the questions there is still a piece of software that searches for the best answer in some sort of a knowledge base.
        Finally Watson includes modules for data acquisition, text processing and indexing, query processing, result processing … just like advanced search software but in a quite different approach.

  5. I’ll certainly be watching this Jeopardy match!

    I wonder how Watson would stack up against Douglas Adams’ Deep Thought for the purpose of computing The Ultimate Answer to the Ultimate Question of Life, The Universe, and Everything. I’m sure Watson would divine “42” as expected, but could Watson go the next step and yield the Ultimate Question? That would be some serious cosmic omphaloskepsis!

  6. Craig said “Saying we enjoy searching, is like saying we enjoy being lost”. No, I disagree. It’s like finding your way, or like making yourself understood in a foreign language. It’s an interesting challenge which I frequenty, though not always, enjoy. Sometimes I long for GPS or a fluent multilingual colleague.

  7. Searching is one of the most frustrating (and time consuming) activities I perform on a daily basis. I am looking forward to the day when asking a question on any system (whether it is the internet or my own system) will produce useful information.

  8. Craig, regarding “Does anyone really like searching for stuff?” —

    Check out an article I wrote in 2004. “If we view computing as the discipline of automating information management, then search is the computing world’s second oldest profession… Search is a necessary evil, providing shortcuts to documents we know exist but don’t know how to reach. Search is also a means of discovering new sources of information on subjects of interest.”

  9. Craig,

    Like many of your other blog posts, you have yet again initiated a thought provoking discussion. Over a year ago I had several discussions with IBMers and others who helped frame for me the distinction between “search” and “find”. As we all know search is a blunt instrument, not very precise.

    Meanwhile, I would also have to count myself among the 72% who experience web searching as easier than searching my own archives. But web searching also has its limitations and can be time consuming. Hopefully Watson, DeepQA or its equivalents can be commercialized so that finding stuff is not an exercise in frustration and may even lead users to uncover useful information and relationships between pieces of information they didn’t even expect to find.

    Though you didn’t find your junior high year book, I’m guessing you may have found some other things – not the least of which was the inspiration for this post.


Leave a Reply

%d bloggers like this: