Web of One

“We believe in a web of one,” Carol Bartz stated at TechCrunch Disrupt. Yahoo produces thousands of custom versions of their home page to reflect users’ interests and preferences. I was struck by her comment because we all waste too much time trying to find what we’re looking for—online or offline. At AOL, our mantra was convenience, which at the time summed up what the newly-created web could do for consumers. But since then the web has grown enormously: Google has 1 trillion url’s in their database. Ideally, personalization services could save time and effort by filtering and recommending. The most reliable model is to make recommendations based on past behavior, which isn’t a completely reliable predictor of future interests or needs. Or to try to group people with similar tastes. Users who bought this also bought this… But these services are not perfect and a single blooper often destroys their credibility. I’ve told my story about Amazon recommending power tools to me (I’m not a handy guy) in a previous post.

I spoke to a former editor at Daily Candy recently and she offered a sobering view of AI-based recommendations. She’s biased toward curation, which, by definition, implies human intervention.

It’s true that no one has shown that a back end can offer relevance to millions of users on a consistent basis. There are very big companies with huge development teams trying to crack the code on this. Our solution? I wouldn’t say we’re pivoting, as much as re-focusing on gigmor, our musician matching site. There we’re solving a real problem with a unique solution. Sounds good to me.