Tech session: Vast and the long tail of classifieds
Niall Kennedy's second Tech Session event is moving upscale. The first Tech Session was at CNET's offices in San Francisco, a ground floor office space. Tonight's event is on the 32nd floor of the Westin St. Francis hotel in downtown San Franciso with spectacular 360-degree views of the city, courtesy of Blish, a digital media ecommerce service. Products presented included Skobee (social networking), Vast (classifieds search), Mozes (mobile services) and Songbird (open source desktop Web media player).
My favorite was Vast, a search service that crawls 5 billion Web pages with a different twist than Yahoo, Google, et al. Launched earlier this month, rather than just indexing Web pages, Vast detects, crawls and structures data to build the long tail of classified services, including cars, jobs and people profiles. "We are assembling the single largest data base of cars--3.7 million cars from over 15,000 sites versus 1.3 for Autotrader, including 794,852 Fords," according to Vast CEO Raval Ravikant Vast will soon be larger and have more personal profiles than match.com or FriendFinder, he said. Vast has about 9.3 million profiles currently from 10,000 sites. "The long tail of profiling is truly a scary thing," Ravikant said, referring to the the porn-related profiles that need to be removed from the index.
"[Vast] is the same idea as GoogleBase, in which user input items, but we are crawling instead and have a much bigger data set currently," Ravikant claimed. "Yahoo and Google hardest sites to crawl, which is ironic because they their businesses built on crawling," he said. "I wish they would be more open." Vast plans to refresh the data every seven days, which means that the data on the site won't be as fresh as what can be found on other classified sites with less of a long tail.
Ravikant, who was a co-founder of Epinions, has more than one startup cooking. He also co-founded Hive7, a 3D browser-based virtual world.