A Search Engine That Relies on Humans
By JOSHUA BRUSTEIN
Published: February 4, 2010
A report this week laying out a strategy for social search has been getting a good deal of attention in tech circles. The paper, "Anatomy of a Large Scale Social Search Engine," was written by Damon Horowitz and Sepandar Kamvar of Aardvark, one of several companies working on creating social search engines. As of October 2009, Aardvark had about 90,000 users.
Social search aims to connect people with questions to people who can answer those questions. By contrast, regular Web searches take questions, break them into keywords, and then find Web sites that have the most relevance to these keywords. The idea has been floating around tech circles for years. Yahoo, among others, has tried to develop social search as a way to challenge Google.
The idea has gained momentum with the increased use of Twitter and Facebook, where people rely on their networks for information, blasting queries to their social networks and, if their networks are good enough, getting useful, personalized responses. Aardvark and competitors like Mahalo are trying to create better tools for people with questions to connect to people with answers. There are some people who think social search has the potential to go beyond Google and fundamentally change the way people use the Internet.
Indeed, the authors draw a parallel between their report and another that the founders of Google published in 1998, "Anatomy of a Large-Scale Hypertextual Search Engine," which hinted at its own paradigm shift. The paper described a prototype of the Google search engine, which is based on the theory of reading hyperlinks and other data to determine how relevant Web sites were to a user's search.
Aardvark uses various factors to identify who it thinks are the best people to answer a question, then poses the question to them. Among the things it tries to determine are the expertise a potential answerer has about a subject, how closely connected the two people are, and how quickly the answerer is available.
From a technical standpoint, Aardvark's task is easier than Google's. As the tech-news blog TechCrunch put it, "On Google, when you type in a query, the engine has to pair you up with exact websites that hold the answer to your query. On Aardvark, it only has to pair you with a person who knows about the topic -- it doesn't have to worry about actually finding the answer, and can be more flexible with how the query is worded."
But there are also some significant shortcomings to Aardvark's approach. Getting answers through social search requires someone else to do something, so it cannot produce the instant gratification that comes from typing something into a Web search box and watching a page of results appear. For Aardvark to be successful, it needs to enlist the participation of competent answerers. (Aardvark says that more than half of the questions posed received an answer within 10 minutes.)
There is also the question of whether or not to trust the answer one gets through a social search. How do you know if the person who answered your question is qualified to answer?
Social search will not replace conventional search, say its proponents. Instead, it will become another tool for Web users, like other specialized search tools such as Wolfram Alpha. Aardvark said in its blog post:
We demonstrate that there is a large class of subjective questions -- especially longer, contextualized requests for recommendations or advice -- which are better served by social search than by web search. And our key finding is that whereas in the Library paradigm, users trust information depending upon the authority of its author, in the Village paradigm, trust comes from our sense of intimacy and connection with the person we are getting an answer from.
Mac Slocum wrote on O'Reilly Radar that another factor -- the increased use of mobile computing -- may be the final piece of the puzzle for social searches. "Mobile search has to be concise and targeted. Results that emanate from a trusted network of friends and associates certainly fit that bill," he said. "Toss in more geolocation features and improved speech recognition, and the utility of mobile-based social search could get really interesting."
Aardvark's report gives an idea of what niche it sees social search filling. For one, mobile Aardvark users are more active than desktop users. Questions on social search also tended to be more complex and more subjective than the average Web search question. As for the issue of trust, the closer a responder was to the questioner's social network, the more likely the questioner was to be satisfied with the answer.
According to Aardvark's report, the most common questions being asked are restaurants and bar recommendations, product reviews and help, local services and travel. And, as with all new buzzy technologies, a sizable chunk of the questions people are asking are about Aardvark itself.