IE 11 is not supported. For an optimal experience visit our site on another browser.

How Google's desktop search works

Google's new desktop search takes advantage of the company's core technology as well as a weakness in Microsoft's Windows operating system.
/ Source: The Associated Press

Google Inc.'s long-awaited expansion of its search tools from the Internet to individual computers takes advantage of the company's core technology as well as a weakness in Microsoft Corp.'s Windows operating system.

Google's Desktop Search application, technically released as a preview Thursday for Windows XP and 2000 PCs, uses the same algorithms that have made its Internet search engine fast, accurate and popular. At the same time, it makes Windows' slow, built-in search tool eat dirt.

The key to speedy searches is the construction of an index of information stored on a computer. A number of companies -- including X1 Technologies Inc., Copernic Technologies Inc. and others -- take the same approach in building their search programs.

Windows XP also includes an indexing service, but it slows down the computer while it's running and is often shut off. If that's the case, the hard drive must be scanned for each search -- a time-consuming process as hard drives can hold hundreds of gigabytes of data.

A simple answer and a secret sauce
Google found a simple answer: index when the computer isn't being used.

Once the 400-kilobyte application is downloaded and installed, it starts indexing the PC's main drive. The process, which only takes place when the computer is idle for 30 seconds or more, can take anywhere from several hours to a few days, depending on the volume of data.

After the drive is scanned, indexing takes place in real time with little effect on the computer's performance.

The index is a database that is scoured by Google's algorithms whenever terms are entered in Desktop Search. The technology, based on the company's powerful Internet search functions, is the program's secret sauce.

Most of the tricks that have worked with Google on the Internet behave the same way with the desktop search. Specific file types, for instance, Excel documents, can be searched by entering "filetype:excel" after the keywords.

It also supports more advanced search functions, such as excluding certain words with a "not" operator, winnowing results with an "and" or expanding what's returned with an "or." Searches also can be limited to specific Web sites that have been visited.

Google's local computer searches also integrate with the Google.com Web site. If the option is enabled, it returns local searches on top of what's on the Web.

Currently, the Google application only indexes the content of a handful of recognized file types, including Web pages previously viewed in Internet Explorer, e-mail sent or received in Outlook or Outlook Express, AOL Instant Messenger chats, plain text files and Microsoft Office documents.

It does find files created by unrecognized programs, but it only searches on the name, not the internal content. Those include music, picture and portable document format files.

The company plans to support more types of files in the future. Until then, users must find another search utility or be patient and wait for results from Windows built-in search.

Taking aim at Microsoft, other rivals
Microsoft also is developing a technology that changes the way information is stored on computers, using a database as the foundation of Windows' file system. That would negate the need to index, since the data would be searchable from the start. (MSNBC is a Microsoft - NBC joint venture.)

The new file system -- called WinFS -- was supposed to be in the next version of Windows, code-named Longhorn. But in August Microsoft announced that it was delaying the technology so that it can release Longhorn in 2006. But the software giant still intends to release an improved desktop search program by the end of the year.

Meanwhile, competing search companies are welcoming Google's entry -- and noting that so far it's limited compared to what they're offering.

X1's search tool, for instance, isn't limited to an individual computer's primary hard drive. It also indexes material on corporate intranets and shared hard drives while ensuring information that's supposed to be private is kept private.

"As a startup, the struggle has been building consumer awareness of the technology," said Josh Jacobs, X1's president. "The more people who bring stuff to market help us because we're all pitching in to educate people about how badly broken personal search is."