The personal touch
Can human expertise enhance automated web search?
(FSB Magazine) -- A new wave of web 2.0 software geeks believe in the former and are offering services in which experts recommend the best sites based on search terms the user provides. In addition to ChaCha's network of human guides (chacha.com), Mahalo (mahalo.com) creates hand-edited results pages for the most popular terms (Britney Spears was an early target). Eurekster (eurekster.com) allows users to build custom search portals that tap the expertise of online communities.
All these sites are slick, but success won't be easy: Not only is the search market blisteringly competitive, with hundreds of alternatives to market leader Google, but there is the problem of scalability. Can a service such as chacha.com grow large enough to handle potentially millions of questions an hour? ChaCha already has 30,000 experts on hand to answer questions, but keeping quality high as volume grows will be a challenge.
Human-aided search has a long and not particularly happy history. Yahoo! started in the late 1990s with human-aided queries, and dozens of so-called people-to-people interactive portals such as exp.com (www.exp.com) were the rage in 1999. They all withered under Google's all-conquering algorithmic approach, in which the computer rank the relative importance of sites within a given category based on the number of links to them from other sites.
Human-aided search "fundamentally focuses on the user experience, and humans have more computational ability than computers," says Steve Campbell, the CIO of Bezos Expeditions (bezosexpeditions.com), Amazon founder Jeff Bezos's venture fund, which has invested in Chacha.com. "In many ways we are only just beginning to realize the full benefits and impact."
Yet it's not easy to draw a clear distinction between machine search and human-aided search. After all, community-driven search products such as StumbleUpon.com, Digg.com, and Del.icio.us rely almost entirely on human-created tags and links. And even Google's fully automated search engine tracks links that are created by people. "One of the biggest problems with social search is it keeps getting defined more and more loosely," says David Berkowitz, director of emerging media for 360i (360i.com), a search marketing agency in New York City.
And all human search ventures face the problem of scalability. At the moment, human search seems most promising in targeted scenarios requiring narrow but deep expertise. In the marketing realm, for example, services such as Guideline (guideline.com) and CIRadar (ciradar.com) provide clients with custom market research gathered by trained analysts scanning the web.
But there is a larger issue lurking here: the difference between information and clutter depends on what is being asked, who is asking it and when. It remains to be seen whether injecting humans into that dynamic will refine or blur that boundary.