Future of Television Recommendation Engines to be Built on Socks? Caterina's Hunch

written by: Paul Johnson

Details

Published on Sunday, 27 June 2010 11:23

According to Caterina Fake of Hunch.com, the type of Recommendation Engine used by Netflix and other Connected TV players do not work because they are vertically challenged and they dont know what socks you wear.

Thats right… socks. Eh?!

If you dont know Caterina Fake, she sold Flickr to Yahoo a few years ago, after growing it into one of the most popular photo sharing sites around.

If you don’t know Hunch.com, it is a company we first came across when they launched their beta at the beginning of 2009, and since then, they have been attracting consumers and collecting a huge number of answers to the broadest range of random questions I have come across. They have been building a recommendation tool that gets smarter the more their users use it.

So what relevance does Hunch.com have to the future of Connected TV, could they improve the recommendation engines currently in use by content providers, and does it actually matter?

Hunch.com believes they will be part of the Connected TV future, because of the masses of data they have been capturing on consumers, and what they are now turning into a new set of views on likes and dis likes across a vast range of different products and services...including Film and TV.

To begin to understand whether Hunch.com has a chance to play in this space, we first need to consider what current recommendation engines do - there are fundamtentally 2 different types used today:

1. Recommendations based upon what your friends are watching
2. Recommendations based upon complex algorithms that relate content to other content by giving it context e.g. Jinni creates categories of moods for each piece of content, whereas Netflix creates recommendations based upon what you have watched before, and marries that together with similar types of genre of content.

When you consider the announcements we have seen over the last 6 months - Recommendation Engines, EPG interface, UI, Augmented Content, Search and Discovery have all become hot topics across the connected TV market - CE manufacturers announcing new web enabled TVs; EPG companies announcing big screen versions that combine internet and TV; Google TV with a promise of bringing TV and Internet into a single experience…on a STB; Jinni, TVGenius, FindmeTV, LocateTV, Rovi, TVLinks and a myriad of others announcing deals to launch recommendation engines and social tv interfaces onto Internet enabled TV platforms, it begins to make sense why Hunch.com believe they have an opportunity to play a part in this space.

What Hunch.com believe is wrong with current Content Recommendation Engines is that they are restricted because they are all predominantly delivering a content centred affinity between entertainment and consumption - or in other words, recommendations that do not take into account the personalities and wider preferences of like minded consumers. They are vertically challenged.

Hunch.com has started to deliver recommendations based upon a user centred affinity between products & services and consumption.

For example, did you know that “...people who like diagonally cut sandwiches, prefer Ray Bans...” ? Nope, neither did we, and it will be up to Ray Bans to decide how they can use this snippet of insight...

So, does Hunch.com have an opportunity to shake things up by adding a new dimension to the way we discover entertainment on our big screens…can they succeed?

Recommendation Engines are in need of improvement, and there are many ways being explored in this space across both types of Recommendation. Hunch.com would bring a new dimension to this 'Great Rec Engine Race'. In addition, any of the answers it gives include advertiser sponsored links so the is potential relevance to TV advertising of future, however they have a few barriers to overcome if they stand a chance of playing a part in the future of tv.

1. There is the risk that I have just spent the last 12 months answering random questions with random answers ie. there is nothing to qualify accuracy of the data they are capturing...
2. There is a significant effort for them to convince broadcasters and other content providers to use their profling data - I dont know one broadcaster who would understand how to turn what socks people wear into justification for investment into a new TV show!
3. Most people tend to watch what a content provider chooses to promote at the time, or what their friends are recommending.

So it is early stages for Hunch.com, however I believe if this kind of service continues to build up the datasets and profiling, and they prove there is a real correlation between content affinity and the profiles they have, then it is something for connected TV providers to consider.

The hardest thing they are going to have to overcome, is to convince this emerging market that knowing what socks we wear can increase the eyeballs and profit for content owners.

To get a better idea of how it works and what their plan is, watch the interview over at Wired http://www.wired.com/epicenter/2010/06/flickr-co-founder-has-a-hunch-personal-data-will-drive-the-future/

For an even better idea of what it is, go and start answering some random questions, then start to ask it for some recommendations yourself and see if what they recommend is something that is relevant to you www.hunch.com

What do you think about the current Recommendation Engines? What do you like? Dislike?

What do you think of Hunch.com? Do you agree they could find a place in our emerging landscape, or do you think that its irrelevant?

Are there any broadcasters or content retailers out there willing to give Hunch.com a try?

What type of socks do you wear?!

Paul is a founding Director at Montgomery Aston and has been specialising in making sense and acting upon the convergent areas of Media, Communications and Consumer Experience for the past 15 years. He helps drive innovation across marketing; advertising; digital devices & platforms, user experience, broadcast, on demand, publishing, social media and mobile. Hook up with him at Linkedin and @mediadventurer.