Last week we had a great BayCHI panel on Social and Personalized Search. Participants were from Netflix, Live365, del.icio.us, Pandora and digg. I moderated it. The podcast for that panel is now online. In inviting this particular group of panelists, I was hoping to highlight the changing trends in Social and Personalized Search. The companies were founded between 1997 – 2005 and their different approaches tells its own story. More about that inside.
For a more detailed desciption of the panel, go to Rick Boardman’s blog.
Netflix (founded in 1997) uses personalized recommendations (based on collaborative filtering algorithms) as the main way of letting users find movies they might want to watch. As Neil Hunt explained, there is a strong business reason for doing so – instead of everyone looking for the latest hit, Netflix helps users explore the long tail of movies (these movies are generally less expensive than the latest hits from Netflix’s perspective).
Live365 (founded in 1999) takes a different approach to recommending music. It takes the concept of a radio station and makes it the expression of one person’s musical tastes. In doing that, unlike other recommender systems, it retains the concept of an expert who helps shape musical tastes, and introduces new music into the system. As a listerner, you can go in and listen to the stations of particular DJ’s. Live365 also uses collaborative filtering to recommend music to its listereners.
Pandora (started in 2000 as Savage Beast) uses human expert ratings to classify music on multiple dimensions. This forms the basis of its sophisticated recommendation algorithms that suggest music based on inherent musical qualities. Its a recommender system, but not a social one. I think the big problem for Pandora is scalability or how to get new music into the system (since expert involvement is quite expensive).
Next is del.icio.us (started in 2003) – a social bookmarking system that uses tags as the main system for information discovery. On del.icio.us you can save your own bookmarks and find out what others thought of them. Additionally, you can also browse other users, their tags and users directly (rather than the interaction being mediated by a collboartive filtering system).
Finally came digg (started in 2005) that encourages user participation to mine the latest news on the web. Unlike the other systems, Digg uses the time as a design element to let stories that are interesting right now bubble to the top. Kevin Rose mentioned that digg will be adding some more personalization to the site, but at its core its really about social trendspotting.