Today's super-size hard drives don't come close to storing the billions of gigabytes of information found on the Web, and yet federal workers often have a significantly easier time locating an obscure document in cyberspace than they do finding a dated but much-needed e-mail message in their inbox.

A simple reason explains this disparity. Web search tools build an index of content and then search that streamlined index using keywords. Most hard-drive search tools, however, are application-specific and rely on what's known as "raw search" or searching file by file, word by word.

As a result, "even the most organized workers can't find the information they're looking for — or at least not very quickly," said Susan Feldman, research vice president for content technology at IDC. "People end up spending hours and hours looking for their own information, and it makes for a huge waste of time and productivity."

That no longer has to be the case. Companies that brought dynamic, indexing search tools to the Web are finally focusing on the desktop.

They include Web search mainstays such as Google, Yahoo, America Online, Lycos, Microsoft and AskJeeves, and other niche players in the larger search market, such as Copernic Technologies, Blynkx, Intellext, X1 Technologies and Autonomy.

Often available free by download, desktop search tools promise to cut through the clutter of word processing documents, e-mail messages, spreadsheets, PDFs, locally stored Web pages, images and video files with toolbar-based functionality and a click of the mouse.

The price may be right, but before agencies turn these tools loose among their users, they should understand how they work and their potential security implications, which can be significant for the risk-averse.

Building an index

Desktop search programs differ in the types of applications they support, but they basically work in the same manner as their Web-based counterparts. They scan hard-drive contents first and build an index of the information by keyword or phrase.

User queries are then run through the index, returning results in seconds or less. Most of the programs present the information by file type, date and relevancy. Users can immediately see a summary and then double-click on the result to open the file.