CBC does not endorse and is not responsible for the content of external sites. Links will open in new window.

At a presentation Thursday on the future of technology, Paul Lee, director of Deloitte Research in the U.K., made a number of predictions on what 2006 will hold for communication, entertainment and technology.

Among them, Lee predicted that in a few years, search will surpass e-mail as the most used online application.

As the amount of information online increases, he said, its value will decrease just because it will be harder to find. The rise in what he called amateur content – blogs and online photos – will mean search engines will have to keep up to give good results.

Lee also said searching online will move beyond using text searches to just find more text. Text will be used to find other forms of media, such as pictures, audio and video.

Some search engines have incorporated these searches already. Yahoo Search has images, video and audio as options in its search results. Google Image Search has also been around for a while. (Google Video is still in beta and is more of a searchable database of videos than a search engine for video on the web.)

Results from these kinds of engines can be unpredictable, though, because they use information such as file names and surrounding text to match words to media. A Yahoo image search using the word "cow," for example, yielded not only photographs and cartoons of cows, but also a still from the movie Chariots of War, abbreviated as "cow" in the file name.

A better way to search, Lee said, is to use text descriptions of the media, or metadata. The metadata for a digital photograph for example, may include the photographer's name, the names of people in the photograph, where it was taken, and the model of the camera used.

But technology can only supply some of the metadata. People have to provide the rest.

"Metadata is now applied manually, which is a laborious process," said Lee. "You only apply metadata to information that you think might be useful, but you never know what might be useful."

One method developed to make writing metadata easier is tagging. A tag is a short description, usually just one word, of a piece of media. A picture of Niagara Falls, for example, might have the tags: Niagara, falls, waterfall, water, Ontario, mist, rainbow.

Among several websites that rely on user-created tags to organize their information, Flickr, an online photo sharing site, is one of the most popular. Searching Flickr for the tags "Niagara Falls" brings up not only several pictures of the falls themselves, but also some of thebutterflies at the Butterfly Conservatory in Niagara Falls, Ont.

Tagging on Flickr has become so popular that it has spawned a game, called fastr, in which you're given a series of photos and you have to identify the tag that's common to all of them.

Del.icio.us is a online bookmarking site that also uses tags. You can add a web page to your del.icio.us list of bookmarks, and add tags to describe that page. You can also search through the bookmarks of all the other users on del.icio.us, using their tags to narrow the search.

Lee called tagging part of the evolution of metadata, but he also had reservations about it.

"Tagging is inaccurate because it depends on people writing their own tags," he said. "What's orange to one person might be yellow to another."

Some technologies have been developed to automate the creation of metadata. As an example, Lee recommends Podzinger, a search engine that uses speech-to-text technology to create transcripts of audio from podcasts.

A search for "Harper" on Podzinger brings up the CBC Metro Morning podcast of a series of interviews on the expansion of the Toronto island airport. You can even click on the word "Harper" to fast forward to the point in the interview that mentions the prime minister-designate. The transcript is pretty good, but not perfect. "Toronto" is transcribed "to auto" at one point.

Eventually, said Lee, one goal of searching is to take text out of the equation and instead use media to search for media. As an example, he said instead of searching photos for the word "dog," you would give the computer a picture of a dog and it would give back pictures of similar-looking canines.

Lee also mentioned a technology that's available on cellphones called Shazam that will identify a song based on a 30-second clip.

Users of Flickr have also picked up on this trend of searching without words. Colr Pickr is an experimental search tool that uses colour. Select a colour from the wheel in the middle and it will return photographs from Flickr that best match that colour.

And retrievr is even cooler. It will give search results from Flickr based on a sketch you make using a simple painting tool.