Friday, February 23, 2007

Game Friday: Netflix and Other Games that Care

When we were kids, my dad invented a game he’d play with my sister and I called “yum/yuck.” He served as gamemaster, and would throw out names of foods: “Spinach!” “Cottage cheese!” After a food was named, each of us would make our pronouncement—yum or yuck. His goal: demonstrate how impossible it was to cook a meal we would both like. But it turned into an incredibly fun game for us, because it was about us. We’d already acquired that American (human?) taste for self-obsession, and yum/yuck let us revel in our particular preferences.

I was reminded of this game last week when I logged onto Netflix to update my queue and was prompted to rate the films I’d recently returned. I felt that same self-absorbed pleasure I’d felt during yum/yuck as I blithely doled out stars. What do people love more than being asked their opinion? And Netflix, like the best dad in the world, actually cares about your opinion. It cares equally what you think of Delicatessen and Tank Girl. And it wants to give you tailor-made suggestions based on your opinion.

I wrote a post a couple of months ago about Amy Jo Kim’s excellent presentation, “Putting the Fun in Functional,” in which she examines several “game metrics” and how successful sites like MySpace make use of them to promote stickiness. Among these metrics is personal feedback from the system. There’s something magical about a machine that cares about your inputs and responds accordingly. Most traditional games rely on the human players for this feedback; I move my rook, you move your knight. I say “cilantro,” you say “yuck.” Presumably, one thing that makes games fun is the expectation that the other player—or in this case—the system, will evolve its strategy based on your inputs. If your opponent is arbitrary or uninterested, you stop playing.

Perhaps the best example of a non-human opponent of this kind is 20q, an artificially intelligent game that is remarkably good at 20 Questions. 20q is an interesting application for other reasons, particularly how it learns, but the reason people keep playing is because it’s fun, not because it’s interesting. Because the game, on the surface, doesn’t care about its own growth. It cares about you.

Netflix isn’t selling movies; it’s selling the activity of movie-watching. To do so, Netflix assumes the persona of an interested opponent who has one goal: to find out what you like to watch. Just by providing a platform for you to rate movies, Netflix celebrates and ascribes value to your preferences. Sure, Netflix gives you access to critics’ opinions as well, but it’s your stars that show up under the movie title.

Like Netflix, museums are portals that offer access to a wide range of content. But museums usually work the other way. Rather than selling museum-going, they sell the content. Instead of privileging visitor opinions, the experts are on display. If you like the content, you can play the role of the 20q machine—asking the museum questions, probing for the answers. But that’s work. Most people prefer to be polled.

Providing a forum for the expression of visitor preferences doesn’t require that museum content fall mercy to the whim of the visitor. Netflix doesn’t change its content based on your reviews. They don’t dump movies that get rated poorly. But they do give you a fun way to express your preferences, and they try to reward you for doing so. They make a big, impersonal repository of stuff something personal and fun to interact with. What more could museums ask for?

2 comments, add yours!:

Sibley said...

So how _could_ museums tap into the desire to opine and customize accordingly? I could see particularly large art museums guiding you to paintings you'd like, or maybe museums with frequent temporary exhibits or events could further promote to you those relating to your preferences. I think it's surprising that ticketmaster or fandango don't to nearly as good a job at this as netflix.

More useful might be in selecting the right audio tour to use with a particular exhibit/museum. Movie recommendations via collaborative filtering start being decent after asking users the right 3 multiple choice questions; I could see museum tours being far more to my taste if selected through a similar set of queries about my taste.

Nina Simon said...

Some of this could be manifest playfully, without too much notable output to the user. There's an online yum/yuck like better, in which you just select your preference between two images. It's fun to do, because you feel like the system "cares" about your responses. And if the system can use that to give you a recommendation of art to see, so much the better.

This could be implemented as a PDA app as well where you walk through the museum and tag each artifact with your level of interest, which could then be used to generate custom tours etc.