Early on in the NoTube project we asked how we address the challenge that people want personalisation without any compromise to their privacy. We posed this question in the context of the NoTube Beancounter, a service for aggregating and analysing activity data around programme watching from various sources around the Web as an input to recommender services. I wondered how willing people might be to share this kind of data, and it seems to me now, fifteen months later, that our initial hunch that TV-related activity data is quite private for most people was probably correct.
Despite the fact that, in theory, more data about you means better personalisation (both for you and for the people you are connected to) because recommendations can be based on what your friends are watching, Beancounter-like data aggregation services don’t seem to be taking off in a big way in relation to Social TV. [However, Dale Lane has been doing some very interesting work sharing and visualising his television-watching habits, which is very similar to our initial ideas for the NoTube Beancounter.]
The findings of a user survey released last month by Sidereel.com (a service that “helps users find, track and watch shows online”) support the notion that, when it comes to TV, the promise of better personalisation in return for a potential compromise to one’s privacy is not enough of an incentive for most people. The survey found that whilst 25% of participants want to see their friends’ TV watching history, only 10% are actually prepared to share their own data:
However, this doesn’t mean that people don’t want to talk about TV online, and we are observing that people are continuing to choose to discuss specific programmes they’re watching using existing tools such as Facebook and Twitter. In fact, a high proportion of the conversations in social media are around what people are watching. For example: a YouGov/Deloitte report published in August 2010 found that 42% of those UK adults who use the Internet while watching television do so to discuss or comment on the programmes they are watching at the time. Such explicit, conscious sharing allows you to take part in Social TV whilst avoiding the potential pitfalls of unwittingly disclosing (possibly incorrect) information that you would prefer to keep private.
One of the reasons why people might be put off using automatic activity data aggregation systems might be the fact that designing a good user experience to control the sharing of personal data at a very granular level (e.g. “share this type of data with this group of people, but not that type of data with that group of people”) is just so difficult to do well. For example, witness how users of Facebook are still struggling with its privacy settings. Further, perhaps this is especially the case where people’s basic need is to watch TV; they really don’t want to bother with the mental effort of thinking about all the possible related privacy implications of sharing what they choose to watch by default.
Anecdotal evidence also suggests that in general people do have reservations about automatically sharing their TV watching activities Beancounter-style, at least at this point in time. Therefore, in exploring our use case for integrating the Social Web with TV to help people decide what to watch, rather than generating recommendations based on what your friends have watched, we’ve been thinking about using publicly available Twitter trending information to generate recommendations for live TV instead.
This approach is based on the observation that during prime-time scheduling Twitter trending topics are often TV-related (at least in the UK), and that this Twitter activity can influence what people decide to watch: for example, people reported watching The Eurovision Song Contest on the basis of what was being said about it on Twitter, even though they wouldn’t normally have watched it.
In situations where people are led by the fundamental desire to be part of something that everyone else is watching, choosing to watch a programme based on Twitter trending topics might be the sort of social recommendations that people are looking for, especially since there are no uncomfortable associated risks to anyone’s privacy in this scenario.
Hi Vicky
Nice post. But made me wonder if the real issue here isn’t about whether punters want or need beancounter type applications for personal / social use, but rather how people should respond to the fact that those applications are already being built.
At some point this year Sky are planning to launch directed advertising (Smart TV) initially based on fairly general data points available via customer sign-up information but in the longer term based on back channel data (what you’ve watched / recorded etc).
One of the deliverables for YouView is targetted advertising .
The standard line on web connected telly boxes is all about content delivery and the data route in. But as soon as the box is connected there’s a data route out and a backchannel.
People are already building the back channels and the bean counters. Being true to broadcasting type they’re just doing it to target advertising rather than target programmes. But if people’s viewing data gets phoned home and aggregated and the beans get counted and the data mined, there’s lots of obvious questions: Do I own that data? Can I see what’s collected? Can I “correct” it? Can I control how it’s used? By whom? For what purpose? If I swap providers can I take my data with me? And get money off? If the aggregator owns the data at the point of aggregation where does my ownership start / stop? Can I hack the STB (hardware or software) to tap into the back channel and pipe it elsewhere (like Audio Scrobbler as a side pipe to iTunes)?
Making what’s happening explicit and giving the user control over how much “spying” their tv does feels like a nicer model. Maybe we need common protocols for “scrobbling” TV (and radio and music and etc) and data portability?
I guess the wider point around privacy is that the focus is always on what gets reflected from applications to the wider web when the real problem is how much data is absorbed by the application and passed onto 3rd parties?
Hi Michael
Many thanks for raising these interesting questions. I think you’re right and these are definitely issues that we also need to be thinking about.
Pingback: The Future is Not Quite Now. Or Something. « Rise of the Inforgs!
Almost on cue, a Wall Street Journal article on targetted telly ads:
http://online.wsj.com/article/SB10001424052748704288304576171251689944350.html
“Bank of America estimates the market for ‘addressable ads’—those targeted to specific household segments—could reach $11.6 billion by 2015.”
“The law doesn’t address activities like combining TV-viewing data with mobile or Web browsing, practices barely imaginable when the Cable Act took effect a quarter-century ago.”
“TiVo says it also licenses anonymous viewing data to TV-targeting upstarts like New York-based TRA, which matches second-by-second data from 1.7 million TiVo set-top boxes and a cable operator with other data types—including 57 million frequent-shopper cards. The matching is done through Experian PLC, a major data company that knows which set-top box and which frequent-shopper card belong to a particular street address.”
The beancounters will be built