RealTime IT News

Mahalo Testing The Opinion-Driven Algorithm

NEW YORK -- In a keynote address here at the Search Engine Strategies conference, Mahalo founder and CEO Jason Calacanis announced a new set of features that he promised would bring the social graph to search.

My Mahalo will be a sort of mashup of a social network and a search engine, where people's reviews and opinions appear in a box to the right of the search results.

Launched last May, Mahalo promotes itself as "the human-powered search engine." Mahalo focuses its search efforts around some of the top vertical categories, such as travel, products, news, sports and health.

Mahalo employs a distributed workforce of some 400 paid guides who comb the Web in an effort to separate the good content from the mediocre, the legitimate from the spam. The idea is that through the process of human filtering, low-quality sites will never get indexed, but that of course means that the indexing process is much slower.

To date, Mahalo has about 40,000 links, and is growing at a rate of about 1,000 per week, Calacanis said.

Save for the sharp Brooklyn accent, listening to Calacanis criticizing the current state of search is uncannily similar to the talking points of Wikipedia founder Jimmy Wales used on his press junket announcing Wikia Search. Spam-free, human-powered social search also sounds uncannily similar to the niche search engine Earthfrisk.org launched late last year.

More recently, Yahoo announced that it would begin allowing developers to annotate search results and support various Semantic Web standards. This came shortly after the launch of Yahoo Buzz, a news aggregating site similar to Digg.

Yahoo's developer platform aims to augment search results with user-supplied data to make them more relevant to the searcher. Just as with Wikia Search, this is the Web 2.0 concept of the "wisdom of the crowds" at work, the belief that human involvement will produce better search results than algorithms alone.

Similarly, Calacanis' idea is to add a semantic dimension to search results by incorporating the opinions and experiences of individuals, either drawn from within a person's social circle from the Mahalo guides with the highest reliability scores.

Mahalo is chasing the same concept as the social networks are pitching to advertisers: the trusted referral. But having the wisdom of the social graph bottled up within the Facebook or MySpace complex doesn't make the best use of it, Calacanis argued.

"The time when you want to know what your friends think of a product is when you're searching for it," he said.

So using the Mahalo toolbar, users will be able to add the Mahalo Top 7 -- the reviews and opinions of top friends or guides -- to a Google or Yahoo search results page.

Mahalo added a social feature to its site in December, giving it a dimension of community similar to Delicious, but for the volume of data needed to provide consistent, relevant results tagged with the wisdom of the crowds, the niche site will have to look outward.

Calacanis outlined a vision for My Mahalo that would scrape user data -- with permission, always -- from outside sites. Imagine, he said, the reviews from people's Amazon profiles or their Netflix queues fueling the recommendations that show up when searching for a book or movie as a Father's Day gift.

Perhaps down the road. The only real partnership that Calacanis had to announce today was with GoodReads, an online book recommendation site.

Calacanis admitted that he was outlining a vision, and that much work needed to be done to achieve his vision -- and that of many others -- of a Web where seamless data portability is the norm, and search is about more than just machines.

"We don't know what the new algorithm is, but we'll figure it out," he said. "My philosophy as an entrepreneur is 'start, and figure it out.'"