Thursday, May 19, 2011

Is Computer Science Cool Again?

Back in 2005 I worried about loss of excitement about computer science among America's youth.
Today computers have become almost as commonplace as televisions and teens use them for a variety of tasks, including researching on the web, communication via email, instant messaging and blogging, and writing papers, all without an inkling of how to program. Computers have become a commodity and they don't see an additional value in knowing how and why they work any more than they need to know physics to drive their cars.
Yesterday IBM's David Ferrucci gave a talk at Northwestern on the challenges of creating Watson to play Jeopardy. The large lecture room was packed mostly with undergrads. Ferrucci didn't disappoint showing the challenges of Watson with lots of examples keeping the talk not too technical. Ferrucci will give a similar talk as an FCRC plenary speaker or you can watch a short version here.

My 12-year old daughter Molly came and enjoyed the talk and insisted on meeting Ferrucci afterwards just to tell him how cool Watson was.

The number of CS majors has grown dramatically in recent years at Northwestern and other universities. Bank of America runs ads on their ATMs that understands peoples checks ("How do they do that?") and Ford on how its Focus parks itself. And our local high school next year is teaching computer science again.

15 comments:

  1. The job market must be over! Who is going where?

    ReplyDelete
  2. Watson was made clever, but whoever gave Watson a decent voice was a marketing genius.

    ReplyDelete
  3. Will Comp Sci be cool again?
    I think that these things are cyclical.
    within 10 years Watson and other things that are cool now will seem commonplace.
    Hopefully something ELSE cool will come along. but there may be a dead spot.

    Will this affect Comp Sci enrollment?
    YES- but so will the job market and
    MORE IMPORTANT the PERCEPTION of the job market.

    Enrollment is UP at UMCP.

    ReplyDelete
  4. No, computer science isn't cool.

    But *theater* is always cool, and Watson is excellent theater.

    ReplyDelete
  5. But *theater* is always cool, and Watson is excellent theater.

    Watson shows in practice and in layman terms the amazing recent advances in language processing and data mining and you reduce that to "theater"?

    This is an embarrasingly poor vision what outreach is. Attitudes like that create an environment in which NSF budget cuts are easy to pass in the house. Unless we take the time to explain to the people the value of what we do they won't see a reason why to fund us.

    ReplyDelete
  6. You misunderstand me, whoever you are who won't sign your name. Theater is important!

    Watson shows in practice and in layman terms the amazing recent advances in language processing and data mining and you reduce that to "theater"?

    Well...yes. Exactly. What else would you call it?

    ReplyDelete
  7. That's interesting, because the complaints I've heard from the ML/NLP researchers was that Watson did not fundamentally represent any new research, just clever packaging of results that had been around for several years already. To exaggerate slightly, it would be like claiming that TV repair is going to be a hot field now that we have High-def TVs in every home.

    ReplyDelete
  8. Well...yes. Exactly. What else would you call it?

    Outreach? Science dissemination? public service?

    just clever packaging of results that had been around for several years already.

    gosh, you mean it wasn't research that had been discovered yesterday??? what a fraud!!

    ReplyDelete
  9. gosh, you mean it wasn't research that had been discovered yesterday??? what a fraud!!

    No, the claim was stronger, something to the effect that the team that put Watson together did no "original research" but "merely" built a system using existing techniques from other researchers. (I put everything in quotes partly because I can't evaluate these claims; this is just what I have heard from people working in that field. Moreover, it smacks of downplaying the hard practical work of actually building a working system.)

    ReplyDelete
  10. Your daughter has been a CS fan for sometime and has performed in a complexity video so she does not count.

    Face it, we are never going to be as cool as rock or movie starts, and being cool is not that good for science. Being respected as doing something valuable is much more important, too much attention would bring too much intervention from outside (particularly from politically influential idiots whose favorite drink is not coffee but tea).

    ReplyDelete
  11. I suspect that caring about what other people think is cool... isn't cool. I also think that CS rules, and Watson is about as cool as Jeopardy. Not that you should care.

    ReplyDelete
  12. OK, let's start from a basic question: who and what the hell is Watson?
    I'm not from the US and I have no clue what you're talking about. (Incidentally, most humanity nowadays don't live in the US, so I suppose it is wiser to assume that they don't know who Watson is as well.)

    ReplyDelete
  13. I was teaching a class this semester on "Science behind Computing" - we also had some guest lecturers talk about AI, P2P systems and the students enjoyed those. The main focus of the course was to discuss the connection between software tools and algorithms -- so we did talk about Google maps (shortest paths) and Web search (specifically the Kleinberg HITS alg). I was trying to tell them that getting something to work right requires a lot of engineering and clever design, and the main focus in the freshman course was to understand some basic concepts - for example how the web graph structure can help web search.

    Its a bit early to tell if this worked or not, students had a lot of trouble with very elementary programming - for example just scanning an array to find the maximum element seems like an impossible task to them.

    I think the difficulty comes when you make the transition from "high level" ideas to lower level concepts - how is a graph represented. How do you
    write a program to manipulate data. In addition, 25-30% of the students rarely read the posted notes, rarely came to class etc. Maybe that is common in freshmen level courses.....

    ReplyDelete
  14. No, the claim was stronger, something to the effect that the team that put Watson together did no "original research" but "merely" built a system using existing techniques from other researchers.

    Let's assume for the sake of the argument that what you say is true. There is still nothing to your statement.

    Watson was not claiming to be new research or all built in house.

    It was built as a cool demo of what natural language processing can accomplish nowadays. It achieved its aims wonderfully.

    ReplyDelete
  15. Let's assume for the sake of the argument that what you say is true. There is still nothing to your statement.

    Watson was not claiming to be new research or all built in house.

    It was built as a cool demo of what natural language processing can accomplish nowadays. It achieved its aims wonderfully.


    Maybe I should have been more clear: I was not blaming the researchers at IBM, it was more that I was blaming the marketing, commercialization, etc. behind Watson and Jeopardy. And "blaming" is too strong a word. It was just a reaction to what Lance wrote.

    ReplyDelete