Searching For The Best Engine

Editing. No matter how clever a computer program, it will never match a human brain for determining quality and relevance. Some new search engines, including Mahalo and ChaCha, rely in part on human editors or guides to pre-cull the most relevant pages for some searches. You'll probably get more select results than on Google—but only if your search terms are among those the editors have explored.

Focus. Google searches everything—but you don't want everything. You'll actually get more relevant results with a search engine that indexes a much smaller number of pages, as long as the pages are on-topic. Trulia searches out homes for sale, Healthline lets you plug in symptoms to track down possible causes and treatment, Globalspec's searches are aimed at industrial engineers, Like.com offers pictorial product searches, and Spock specializes in information about people.

Guided queries. It's hard to guess which search terms will do the best job, but some search engines help by suggesting terms, as do Yahoo and a start-up engine called Accoona, or by grouping results into categories that focus on the desired topic, as do Ask.com and Clusty. Type "spears" into Ask.com, for example, and it will suggest you steer the topic in either the pointy or pop direction; Google just mixes them up. A number of cutting-edge engines, including France's KartOO, KooltTorch and Quintura—founded in Moscow and now based in Virginia—display the categories in graphic maps that visually suggest which categories are likely to be the most useful.

Community. NosyJoe, Squidoo and Sproose allow other users to help determine which pages are most useful, cutting down on the often irrelevant and spam-ridden results that come up via Google's link-counting approach. Wikia, which has ties to the online, everyone-can-chip-in encyclopedia Wikipedia, is working on a search engine based on user contributions, and the Web-page bookmarking service Del.icio.us, bought by Yahoo in 2005, allows searching through everyone else's labeled bookmarks to find relevant pages.

Right now all these underdog search engines (except Ask.com, the No. 4 search site) have a combined share of less than 5 percent of all queries, according to Knight. But even if one or more of them starts to gain traction, does Google really have to worry about being bested by some obscure search engine, given its longstanding, widespread popularity? After all, Microsoft continues to dominate software, in spite of persistent claims that better alternatives like Apple and Linux are out there. Google's dominance, however, is different from Microsoft's. The costs of dumping Windows can be intimidating, between setting up new hardware or software, retraining and lost productivity. What's to keep someone stuck on Google? "The moment someone proves themselves better than Google, people will switch in a heartbeat," Srihari says. Just ask anyone who was at AltaVista in the late 1990s.

Google isn't waiting around to be AltaVistaed. Its smaller challengers can't hope to match the company's massive investments in computing infrastructure, said to include more than 450,000 servers. So be prepared to wait an annoying three seconds or so for results on some of the wanna-be search sites, compared with Google's blink-of-an-eye speed. And with $12 billion cash on hand, Google can buy hot companies that pose a threat, much as it plopped down $1.65 billion last year for YouTube, whose video search crushes Google's popularity. "Google was buying tiny search companies at the rate of two per week at one point," says Knight.