Damn clever: Delver makes search social

Results are based on the contributions of your friends (and their friends).

This is one of the most innovative ideas at Demo 2008: Delver, a search engine that displays results for you based on what your friends and contacts are doing online. First, you tell it your name, and it scans the usual social networks to find out who your friends are. At this early stage of development, it scans LinkedIn, MySpace, Hi5, and Flickr, but Facebook and Twitter will be added.

Once Delver discovers who you know (and also who's in your extended circle--your friends of friends), it uses that data to return search results. For example, if you're searching for "Las Vegas," it will return pictures your friends took in Vegas, blog posts they've written about trips there, or even posts where they are just mentioned. It also finds Yelp reviews they've written about restaurants in Vegas, and so on. And you do not have to tell Delver that a MySpace friend also has a Yelp ID; the system's core technology draws the line from your explicit connections to their other contributions on the Web.

Delver will not replace old reliable Google. It delivers different results for each person, not consistent results for everyone. And to some extent, the quality of your results will depend on the quality of your social network. But it may well change the way you look at search. Instead of searching the sum of human knowledge (or close to it), Delver searches through the collective wisdom of your friends and associates. It's a great new idea.

The service will be presented at Demo 2008 on Wednesday. The private beta launch is planned for March.