NEW YORK — There's more to search than devising the perfect algorithm.

Two of the newer kids on the search block — Microsoft and Amazon's A9 subsidiary —shared their visions for the future of search during their Thursday keynote addresses at the World Wide Web conference here.

Neither executive was willing to discuss near-term products or strategies. Microsoft is known to be prepping new search technologies that are expected to allow users to search seamlessly across their local machines, corporate networks and the Internet. The new MSN Search part of the equation is expected to debut later this year or early next. A first version of the WinFS file-system subsystem will be integrated into Longhorn when it ships in 2006+. And A9 recently unveiled a beta version of a new search site that builds on top of Google.

Rick Rashid, the senior VP in charge of Microsoft Research (MSR), touched on a number of ongoing search-related projects designed to "empower the individual" in which MSR is engaged. Advances in PCs, high-speed networks and high-capacity disk drives will foster the creation of new applications that will make information more available, more easily indexable and retrievable, and more contextually aware.

Rashid highlighted several MSR projects where search and retrieval play a crucial role. These included the SkyServer, which is a virtual telescopic observatory; the Worldwide Media Exchange, a centralized index of images, tagged by location; and Wallop, MSR's blogging/social networking/document sharing application.

When it comes to making information easier to discover and deliver, the user interface becomes even more key, Rashid said. "We need to model the interface after the way people think and feel," making use of concepts like memory, deep history and dynamic organization, he said.

To illustrate his point, Rashid revisited the MSR project called "Stuff I've Seen" (SIS). SIS relies on Microsoft Search to create an index of personal content, ranging from e-mail, to attachments, files, Web pages, calendar entries, journal entries, etc.

"Search isn't the end goal here," Rashid said, in explaining SIS. "The goal is information management in the context of ongoing work activities. Search happens within the app."

("Search Titans Talk Futures" Page 2)

Rashid showed off a couple of MSR projects related to SIS. One, called the MemoryLens, is similar to a SIS index that is built using dates and times as a key user-interface element. With MemoryLens, so-called "Memory Landmarks" are events such as holidays, news stories and the like. He also showed a variant called "MemoryLens LifeBrowser," another research project that is exploring "thresholds for memorability."

"With more and more data, people need a mental context for categorizing," Rashid told conference attendees.

Doing search doesn't mean you have to try to recreate Google, said Udi Manber, CEO of Amazon.com's A9 subsidiary, who took to the podium right after Rashid.

He said A9's charter is to build new search technologies to improve user experiences, especially in the e-commerce search arena. And Manber — who cut his search teeth developing an Amazon technology called "Search Inside the Book" — was just as focused on the importance of the user interface as was Rashid.

"Ease of use is critical. But it's also a huge barrier to (encouraging users to employ) advanced search techniques," he said. "We are asking our users to play music on one-string instruments."

Manber put A9 through its paces for conference attendees, demonstrating how A9's high-speed history server allows users to store queries and organize by time sequence. Via the history server, users can see the results of queries they performed four months ago, for example, as well as new search results which didn't exist when the initial query was performed.

(When asked by a conference participant whether A9 might allow other parties to share the results from history server queries, Manber declined to comment.)

Manber showed off other A9 user-interface features, as well, including the "Diary" feature that allows users to annotate their search results.

Ease of use isn't A9's only concern, according to Manber. He noted that relevancy is an imprecise science. It's impossible to measure, as it changes all the time and is different for different people.

In closing, Manber listed a bunch of "what-ifs" for those interested in search to ponder — while cautioning that attendees shouldn't conclude that the list had anything to do with A9's future directions.

He asked participants to think about what they could do with an hour of undivided user attention, in terms of teaching them how best to use search. What if everyone became an "author," he continued. "Given all the ways there are to author — Web pages, blogs, digital images — do we have the right mechanisms to publish and consume all of these?" he asked. And what if all published content (books, music and video) could be made accessible from a single place, even if it all wasn't stored in one place?" Manber wondered aloud.