"Privacy is a part of trust," he said, adding that users are getting better at identifying which sites are trustworthy. "If they do burn their users on privacy, it's going to hurt them."
Fake agrees, noting that the user-first values she built into Flickr helped the site succeed in becoming a global community of photography lovers.
"The most important thing you can do with users is be honest and trustworthy -- and don't use the data in any way they wouldn't want."
After Hunch has boiled you down into a nicely organized set of preference data, it's your turn to make a specific request. Say you're interested in finding the perfect kind of dog for you: The site then asks you a series of five to 10 specific questions to narrow your hypothetical dog by size, temperament, price and so forth.
The result of all that work is the set of dogs that Hunch thinks fit your preferences, purse and personality. Now you know whether you're destined to own a border collie, a vizsla or a plain old mutt.
But before you run down to the animal shelter, remember that this is just the best guess of a nascent system. Hunch's real power, said Fake, will come after it has aggregated data from a huge number of users, the better to decide which buckets and sub-buckets each user should fit into.
"The measure of a really good piece of social software is whether it gets better or not as people use it," she said. Because Hunch has been in a limited preview to the company's inner circle, its user base is lacking diversity.
"For things like video games and blogs, we're in pretty good shape right now," she said. "We're probably less good at guessing what kinds of handbags women in Illinois like."
Fake said that based on user feedback, Hunch is getting it right about 80% of the time, but that she'd like to pump that number up to 90% or 95%.
The 10-person company, which has raised $2 million in venture capital from Bessemer Venture Partners and General Catalyst Partners, is not yet concentrating on turning a profit. But eventually it will use the rich user profiles it generates to sell highly targeted advertising.
As Fake figures it: What better time to sell a product to a consumer than the moment after they've decided exactly what they want?
--
david.sarno@latimes.com
