Flippant Calacanis strikes a typical pose.
Search Central Mahalo's headquarters in Santa Monica, California, is currently home to 40 "guides" who build search results by hand.
Related StoriesGoogle is powered by perhaps as many as 200,000 servers (the company won't say) and 10,000 employees. Mahalo, to date, has 20 of the first and fewer than 60 of the second. Each day, Mahalo staffers, who are called "guides," sit before two 24-inch monitors in a converted factory in Santa Monica, compiling search results one at a time. It can take several hours to do a single page of links, depending on the complexity of the topic. Most of Calacanis's employees are young out-of-work novelists, screenwriters, musicians, artists, and actors--info addicts happy to earn $35,000 a year plus health benefits by searching the Web rather than shelving books at Barnes & Noble or slinging chai lattes at Starbucks. Calacanis has promised them 15% of the company when and if it goes public, with the investors getting a third and Calacanis keeping the rest.
Over time, he plans to hire as many as 100 of these Netheads. And while it might be tempting to frame the battle as man versus machine, it cuts deeper than that: It's man using the machine against itself. Mahalo's searchers use Google to gather the raw material to build each page of results. But they also use Google to back up its lists. In other words, if you search Mahalo for a term that has yet to be tackled by the guides, you don't come up empty, you just get bumped to the standard results "from our friends at Google," as Calacanis, reluctant to antagonize the seductress of search, puts it. Mahalo grabs 65% of any revenue Google gleans from the referral. The symbiosis extends even further: Google also provides ads for Mahalo through AdSense.
In essence, Calacanis is using Google to beat Google. He's creating a twofer--a site that tunes and refines the raw machine muscle of Google, but delivers Google at the same time. For people looking for simplicity, relevance, and a little human common sense, it's a pretty attractive proposition. Still, the Web is littered with the carcasses of search companies. And skeptics are legion. Calacanis, they hasten to point out, didn't invent people-powered search: Ask Jeeves did, nearly a decade ago--but Ask.com, as it's now known, gave up on that game, becoming a server-driven Google knockoff owned by Barry Diller's IAC Corp. Likewise, Yahoo's human-powered directory has never been able to gather steam, and Microsoft has abandoned using humans to edit search results. "Wikipedia has shown you can maintain content about popular topics, so it's not hopeless," says Danny Sullivan, editor-in-chief of SearchEngineLand.com. "But assuming [Calacanis] does it, then he still has to persuade people to start trying him on a regular basis. And switching the Google habit for many people is about as easy as quitting smoking--it's not."
Not even Google seems convinced of this last point. "Search is one of the most competitive industries around,"says Matt Cutts, a software engineer there. "Users are only ever one click away from an alternative search engine." Indeed, people switched from Yahoo to Google in a span of four years, and from Friendster to MySpace to Facebook in the same amount of time, and there is zero cost associated with switching. But Google doesn't seem too worried about this latest people-powered upstart. Cutts doesn't even consider Mahalo to be a search engine--he calls it a directory, with links compiled by humans. Besides, he says, Google isn't just cold algorithms and circuit boards. But it relies on people in the way that Mahalo relies on computers, as subservient to the master, as mere inputs to calculate the "relevance" of a site largely by gauging how many people link to it and how many others link to that.
In short, those machines are pretty hard to argue with. They don't need health bennies or take vacations, and they can instantly search for virtually any term you throw at them. The human-powered model, on the other hand, seems an anachronism, as quaint as using monks instead of a scanner to copy documents. But Calacanis is unfazed. And, typically, he has an answer for everything. He calculates that it costs Google approximately $4,000 to operate each of its servers for a year. So for the price of a dozen machines, he can have one human (who not only receives a base salary and benefits, but costs him payroll taxes) who will write some 500 terms a year and keep them up-to-date. Multiply that by his target of 100 workers and Mahalo would be able to generate 50,000 terms a year. What's more, Google's staff is 100 times larger, and precious few of them take home as little as 35 grand. In other words, however machine-driven it may be, there's nothing cheap about running Google. "Search folks don't understand editorial," Calacanis says. "I'm not afraid of editorial costs, just like machine-search folks are not afraid of computer servers." He adds, "CNN was crazy to think they could fill 24 hours with news--let alone around the world in 10 to 20 languages. Reuters or AP with a thousand people around the world covering news? Crazy. But what if I get to that level. Is it possible? Would it change the world?"
Calacanis is in act three of his business life, when the protagonist either achieves his dream--or dies trying.
Email
Digg
StumbleUpon
Facebook
Buzz Up!
LinkedIn
