WebProNews

As the search engine world gets ready for its big-deal conference, Search Engine Strategies: San Jose in August, we tapped the deep intellect of Pandora CEO Tim Westergren for his thoughts on the future of search.

Fortunate attendees of this year's SES San Jose starting August 18th have the usual opportunities to hear from some of the best known and insightful minds in the search industry. Among the usual substantial selection of content tracks will be the Orion Keynote Panel: Technical & Information Giants.

Our readers will recognize names like Matt Cutts, Robert Scoble, and one-time SES host Danny Sullivan from the list of panelists. But while they may already enjoy the intelligent music service Pandora on their computers and iPhones, the people behind Pandora may not be as familiar to regular WebProNews visitors.

We've remedied that courtesy of an email interview with Tim Westergren, on the topic of search. Here's our exchange:

WPN: Since you're participating in a panel on building technology, and search's part in that, where do you see a need for something in search that has not been built yet?
---
Tim: I’d say the primary need is for a search methodology that is not captive to the dynamics of popularity. In one way or another, all search algorithms are fundamentally popularity contests. Clearly, popularity is a decent proxy for quality in most cases, but in my opinion, the systems all share one principal weakness which is their inability to effectively surface the long tail.
---
WPN: The long tail has come into question lately, with Anita Elberse saying "the tail is likely to be extremely flat and populated by titles that are mostly a diversion for consumers whose appetite for true blockbusters continues to grow." Which do you see as more important to tapping the long tail, from a music perspective: technology advances to mine it better, or a change in personal tastes that drives more people to delve into it?
---
Tim: Technology advances will stimulate consumption of long tail music... I don’t see any other impetus for people arbitrarily becoming interested in the long tail.
---
WPN: If we see a development in search come along that resists the dynamics of popularity, what will we call the matching of heretofore undiscovered works (let's say music) to one's existing tastes? I've the feeling you're leaning toward Pandora's model, but if you could expand on that please?
---
Tim: The Pandora Effect :)

I would call it something like attribute-based search, or intrinsic-search, or genomic-search. Something that connotates a knowledge of the actual qualities of the product, rather than trying to infer its contents through exogenous statistics. I think eHarmony does this for dating, for example.

I’m of course advocating Pandora’s approach, but I’m hard pressed to think of a system that can operate any other way without falling into the popularity trap.

Pandora serves as the best case study of his position. The site can tell visitors why it matched a particular song to them, citing measurable factors in the music. As music essentially serves as an audible exercise in mathematics, matching an unknown tune to one's known likes works like solving an equation.

The E=MC^2 of search probably hasn't arrived yet, despite the ongoing work by engineers at places like Google, Yahoo, and Microsoft among others. Text content may not break down as neatly as a song's tempo, chord structure, and other aspects, but we're sure the whiz kids of search are looking at Pandora, even while they are listening.