How do you find great stuff on TV? You don’t search. Because TV interfaces, with the screen resolution being meager, the viewing distance being far (it’s called tele-vision, not near-sightee), and the current input devices being somewhat lackluster, typing a search string is not a very promising approach.
So, you browse (uhm, zap). You use an EPG. You get recommendations by kind of smart engines (people who watch The Simpsons might like the Odyssey as well). You get recommendations by friends, be they real, Facebookish, or just somebody you follow on Twitter.
Behind all this lies the power of descriptive metadata. Because, in digital speak, a video is just another BLOB. No, not the young Steve McQueen’s Sci Fi foe. BLOB means Binary Large OBject. Any video BLOB is watchable, if the structural metadata is intact (telling your device: this is a video file of a certain resolution, in this and that encoding …)
You will only find the right BLOB, if the descriptive metadata is in order. Otherwise, you might get this BLOB. Or this. Or this. Or whatever anything bla bla bla. That’s why descriptive metadata is the backbone of most content discovery mechanisms. There are a couple companies out there dealing with this, covering movie data like Amazon’s IMDB, TV schedules and more from rovi, film and music metadata from Gracenote. There are XML schema definitions for practically any media area of interest.
Descriptive metadata is basically static, like:
- Title: The Blob
- Directors: Irvin S. Yeaworth Jr., Russell S. Doughten Jr.
- Year: 1958
- Actors: Steve McQueen
That’s fine for a basic EPG and some information overlays. Maybe you can even do some genre mapping. But how will we get some recommendations out of this? With TV, it’s a bit complicated. Mostly, you have to explicitly ask the viewer some questions (please rate this, or do you like that …). The results are not too convincing.
A more powerful way to get some viewer input is what LastFM did for music with its audio scrobbling technology. The scrobbler just notes what you are listening too and creates a profile out of that data (inferring that, if you’re listening to Rihanna, you most likely do like Rihanna). By matching your profile with other close profiles, the recommendation engine can extract a delta of possible content you might like. In theory, this would be possible with TV as well. You would need a smart STB or a really smart Smart TV, which both might be some kind of an oxymoron.
So let’s get back to something I already mentioned above: recommendations by friends. For some obvious reasons, friends have the most recommendation power. Not just because our peer group might be good for some peer pressure as well. You know your friends, your friends know you, and might have an idea what you like or not, and vice versa (and as you know your friends, you might discard certain recommendations because you know too well what the source of the recommendation usually likes …).
And here’s where the meta data mania starts. With social media, friends (and followers) have entered the meta data universe. You’re not only neighbors in a city, you’re neighboring entries in a graph database. Second screen TV check-in apps like GetGlue, Couchfunk, or the likes tried first to harvest those kinds of data. This never really worked out, because, as it happens, the use case for those apps is rather screwed.
- TV in general lacks the luster of physical check-ins. Becoming the mayor of the Waldorf Astoria beats outing yourself as being one of several million couch potatoes watching simultaneously the semi finals of X Factor.
- If the content is immersive and great, you do not want to fiddle around with a second screen.
- If the content sucks, you switch.
- … and even if you really favor a show, it’s not too helpful for your friends, as TV is still mostly a linear experience. So when you read the recommendation, it’s most likely too late.
And, not to forget: first you have to find and then install an app. And, most likely, your friends aren’t there yet as well.
So, the winner is? Facebook and Twitter. Both operate kind of like textbased TV networks anyway. Timeline and newsfeed are constant linear streams of information, be it text snippets or images or video. And both platforms are already heavily used all day round. And yes, when the TV is running, a certain degree of status updates and tweets is about the current TV programming as well.
Both platforms are already trying to capitalize on that, by offering big data subsets to their TV partners. The value in this is there, but sometimes still a little bit unclear: Marc DeBevoise, executive vice president and general manager of entertainment, news and sports at CBS Interactive, explained it nicely to the WSJ. The hope is that tweets drive viewers to their shows.
Part of the ambivalence stems from the uncertainty about what, exactly, the payoff is for television networks. “We see a connection between increased Twitter activity and increased ratings,” Mr. DeBevoise said. “The problem is, we can’t tell which is doing which.”
This going to be tough, at least in a world of linear scheduled broadcasts. Following a tweeted recommendation means: you missed the beginning of the show. But if the tweet or status update automatically points to a catch-up resource, you’ll get the word of mouth recommendation you have been looking for.