Netflix Puts Bounty on Viewer Habits

  • Playlist
  • Download
  • Embed
    Embed <iframe src="http://www.npr.org/player/embed/6183015/6183016" width="100%" height="290" frameborder="0" scrolling="no">
  • Transcript

Netflix, an online movie rental company, offers a $1 million reward for new software that improves recommendations on what its users should watch next. James Bennett, vice president for recommendation systems at Netflix, explains.

MELISSA BLOCK, host:

If you've done any online shopping, you've probably noticed that the cyber world is big on recommendations. You know, the pop up that says if you like this, then you'll love this.

MICHELE NORRIS, host:

Now the online movie rental site Netflix is trying to improve on that technology to better gauge their customer's taste in films, documentaries, and television shows. The company is launching a $1 million contest today, and Jim Bennett serves as Netflix's vice president for the recommendation system and he joins us now to tell us about that contest.

So glad you're with us.

Mr. JAMES BENNETT (Netflix): My pleasure.

NORRIS: So you're offering $1 million to anyone who can do what exactly?

Mr. BENNETT: Anyone in the world that can build a system that makes better movie recommendations than our current system.

NORRIS: Maybe you could explain this by giving us sort of an ideal scenario of the kind of customer that might sit down and log on and if this system was working exactly as you wanted it to, what would it do?

Mr. BENNETT: Well, for example, it might start asking you about your favorite movie and then it would ask you about types of movies that you really enjoyed, whether you liked happy endings or sad endings. Right now you have to rate several films and then we try to infer whether that's the case. We could just ask you and that would help us then really guide our inferences later and that would be a major step forward for us.

NORRIS: Now that seems pretty easy, that you could just, you know, put the questionnaire up on your Web site right now.

Mr. BENNETT: Yes, but the hard part is actually taking those answers and turning it around to actually finding which of 65,000 movies have all of the answers properly. Imagine that I asked you the question, do you like movies with happy endings? If you said, yes -

NORRIS: And the answer was yes.

Mr. BENNETT: Right. Exactly. If the answer was yes, then we'd have to know which movies had happy endings, which means classifying 65,000 movies about happy and sad endings, for example.

Getting those kinds of things, which are very, very easy for humans to do, it takes a lot of work online to be able to put that kind of information and be able to organize it and then be able to do it for five million people simultaneously.

NORRIS: You know, what you're doing here makes a lot of sense to improve your recommendations. But I'm wondering if there's an even more revolutionary way to look at this, to find another way to deliver content to your customers now that anyone who uses iTunes, as long as they have a specific kind of iPod, can actually download a film onto their hard drive and just flip it on whenever they want - on the bus, on the plane, on the beach.

Mr. BENNETT: Well, we agree with that and we've been working very hard to develop electronic download solutions. But you know, there are a lot of movies out there. You have to make sure to know which movie you want to put on your iPod, for example. That takes a bunch of work. That's my job.

NORRIS: I imagine the conversations that must take place in the cafeteria there at Netflix, the sort of strange version of six degrees of Kevin Bacon. Like, if you like this film, then you must like that film. Is that the kind of things that you talk about there?

Mr. BENNETT: All the time. In fact, it's a company policy actually to start meetings off with what's called movie talk where everybody goes around the room and shares what movies they've watched and whether they've liked them or not. You get a lot of great recommendations that way.

NORRIS: Jim Bennett, it's been good to talk to you. Thanks so much.

Mr. BENNETT: Thanks, Michele.

NORRIS: Jim Bennett is the Netflix vice president for the company's recommendation system.

Copyright © 2006 NPR. All rights reserved. Visit our website terms of use and permissions pages at www.npr.org for further information.

NPR transcripts are created on a rush deadline by a contractor for NPR, and accuracy and availability may vary. This text may not be in its final form and may be updated or revised in the future. Please be aware that the authoritative record of NPR’s programming is the audio.

Comments

 

Please keep your community civil. All comments must follow the NPR.org Community rules and terms of use, and will be moderated prior to posting. NPR reserves the right to use the comments we receive, in whole or in part, and to use the commenter's name and location, in any medium. See also the Terms of Use, Privacy Policy and Community FAQ.