June 28, 2006
Google Research Making Waves with Social Intereactive Television Proposal

By this point, some of you have probably heard about the new technology proposed by Google researchers Michele Covell and Shumeet Baluja, in association with Michael Fink of the Center for Neural Computation at Hebrew University of Jerusalem.

The study, which won the "Best Paper" award at the Euro Interactive Television Conference in Athens, Greece, about a month ago, was announced by Google Research on their blog on June 6 and made available for the public to read.

Siddiq Bello from Turner Broadcasting, one of our partners, made me aware of the study's availability, and I've been reading and thinking about it since he first contacted me about it a couple of weeks ago. In short, the proposed system would create mass personalization, meaning that the traditional mass media would become personalized to a degree never accomplished before.

I'm not talking about having a fireside chat where you make the American people feel like you are coming in your home. This is more than changing one's tone-of-voice. The researchers write that "mass-media channels typically provide limited content to many people; the Web provides vast amounts of information, most of interest to few." They propose to use their technology to make television and radio as personalized as Web content, while still providing the ability to be a somewhat passive consumer.

How to do this? The system would use a computer's microphone to pick up on what programs a person is watching on television and would then provide relevant data online to help enrich the viewer's experience. The report breaks this down into four categories:

1.) personalized information, selected by the viewer, that would give further information or related products while watching programming. The example they use is a fashion layer which pops up information about where to purchase clothes that actors are wearing on a particular show.

2.) "ad-hoc peer communities," formed around what a person is watching at a particular time. The socail networking software would be able to put people into chatting situations with others who happen to be watching the same program at the same time. This can be further limited to only within someone's personal social network, so they can be notified if friends happen to be watching the same program at the same time, with new chatting options offered when the viewer changes channels.

3.) real-time popularity ratings that would measure how many people are tuned into a particular programming using this same technology. They emphasize the value that real-time ratings could provide for programmers and viewers alike and how much this would overturn the current system in place measured by Nielsen. Also, for programs like the 24-hour news channels, pro wrestling, and other television shows that are aired live, this could actually give writers a chance to monitor viewer resopnse as they go along and change programs accordingly. Such a response system could help spurn interest in live television once again, due to the ability it gives producers to shift based on audience reaction.

4.) Finally, the program would allow for video bookmarks to help viewers create their own personal video libraries.

This report describes the technology behind this system and how it is possible. It's obvious that this is a while away from execution but presents one potential course television viewing may be going to make it more interactive. This type of programming would drastically alter the structure of advertising as well and may help further the transition from the traditional advertising spot.

However, there are also obviously a lot of privacy concerns involved here as well. The report details all the ways in which conversations are protected by the very way the system gathers data--the way the audio is broken down into coding before it ever leaves the computer to be streamed through the Internet eliminates any chance that a conversation could be picked up or monitored, according to the paper. Not surprising, though, that the privacy fears of many are not assuaged by these assurances at this point.

And I understand. I am both intrigued by the possibilties of this technology to transform the viewing experience and further make watching television a potentially active experience and bothered by the potential privacy expectations. I guess we'll need more than their assurance that "the highly compressive many-to-one mapping from audio to statistics is not invertible" before most Americans welcome a machine that streams data about what they are watching on television out.

The important thing to keep in mind, though? This is not some mandatory program. The only people who would use it are people who choose to. And I don't think we should let our privacy fears not permit us to at least think about the great implications if the security of people's lives can be insured.

What about the rest of you? Do you find this exhilerating...disturbing...or perhaps both?