X
Business

Mahalo: A human-powered search directory

Jason Calacanis of Weblogs Inc. (now part of AOL) fame launched Mahalo at the D conference this afternoon, with the idea that human-generated search results for the most common queries can trump machine-generated results and not Mahalo will leave the vast majority, the long tail of queries to Google.
Written by Dan Farber, Inactive

Jason Calacanis of Weblogs Inc. (now part of AOL) fame launched Mahalo at the D conference this afternoon, with the idea that human-generated search results for the most common queries can trump machine-generated results and not Mahalo will leave the vast majority, the long tail of queries to Google.

It's the latest iteration of the a Yahoo directory or About.com--back to the future--with editors selecting results and organizing the information on Web pages. Users can offer input on the results to Mahalo's staff of 40 editors. Of course, Mahalo is in alpha and encompasses less than 5,000 results. Calacanis said that 10,000 search terms is about 24 percent of searches.

Mahalo, of course, will be advertising based. Calacanis compared Mahalo to Wikipedia, which he said sucked in the first few years and then took off in year four or five. In the first few years, Mahalo will get to 25,000 search terms and then go into maintenance mode, he said. "We can help to dictate good behavior on the Web," Calacanis added, citing sites with spam that would not be selected by editors for inclusion in the directory pages.

"We have enough money to run the company for five years," Calacanis said, noting his blue chip investors. "Why not build a Reuters or AP organization to handle 24 percent of searches," Calacanis said. His pumped up metric is to save two or three hours across millions of searches 24-hours a day. Wikipedia and About.com have shown that human edited services can generate revenue, but scaling turns out to be adding more people and more machines and bandwidth. At some point, human powered search directories will run into the Semantic Web, in which the machines interpret human behaviors online with greater precision, moving closer to an intelligent Web, but Calacanis has a lot of runway with which to work.

Editorial standards