Three themes will shape the development of the Internet in the next few years. You need to be aware of them because they will affect how we use networked personal computers at home and work. And if you're looking for new business opportunities, there are plenty of them here.

We can call the first theme "the Internet as desktop," a change in Net usage that is already visible in tools such as Google Apps.

Offloading common computing tasks to the Net is Google's way of challenging Microsoft's hold on office software such as word processors and spreadsheets, as well as e-mail. Google Apps (www.google.com/a/) offers the advantage of using tools online, saving companies money and letting them standardize software easily across their business network. But Microsoft has just thrown Google a haymaker in the form of a new browser plug-in, Silverlight.

I call Silverlight a plug-in, but for developers, it's the chance to leverage the Net's flexibility while offsetting some of its disadvantages. Though developed to provide multimedia experiences on the Net, Silverlight is powerful enough to offer desktop-like performance with Web applications. It's also capable of working with Internet Explorer, Firefox or the Mac's Safari.

Down the road, I can see Silverlight tying in to Microsoft's Internet office tools to make the existing Google applications look positively prehistoric.

Take a look at the demonstrations available at the Silverlight site (www.silverlight.net), where you'll see this speed and flexibility on display. Then ponder how Google will respond when Microsoft turns this rich platform in the direction of an office suite that can be used online, with much of the functionality of the desktop equivalent intact.

Silverlight makes the Internet as desktop look a step closer, at least in terms of performance. The security question -- Do you trust your data online? -- is still a problem.

But we won't start using the Internet as desktop overnight. What will happen is a gradual adoption of useful tools.

It's handy to have Web-based e-mail such as Google's Gmail, because you can retrieve messages from any computer. Try what I do -- keep multiple e-mail accounts, feeding mail from one to Gmail, while keeping any sensitive e-mail relegated to desktop software, such as Outlook. The idea is to draw the best experience out of desktop and Internet alike.

Second theme

That experience is materially enhanced by the second theme, which I'll call "the Internet as code." This isn't old-style code-writing. Tomorrow's Net will be laden with graphical tools such as Yahoo Pipes that let you drag and drop modules into easily defined relationships, connecting the output of one to the next, and building your own set of customized tools where you need them. It's a user-based programming revolution, one that sees the Net as a repository of data objects that can be shaped into new configurations.

In my research, I use Yahoo Pipes to extract data from a particularly useful Spanish-language site, feed the information automatically into the Babelfish translation engine (www.babelfish.com), and send myself the output as an RSS feed. I can subscribe to it so that I receive updates any time the site's content changes.

I get the daily postings of a researcher whose work I follow, translated into readable (if seldom idiomatic) English, by dragging and dropping data elements into relationships that create a new Net resource.

Pipes (pipes.yahoo.com) is just the beginning of this trend, and I see that Microsoft is working on a visual programming tool, PopFly, that will extend these ideas.

But once we take today's notion of a "mashup" -- a site created from data found in other sites -- to its logical conclusion, we've freed Web users from the need to rely on others for all their content. We begin to see how malleable connected Net data can be as we learn to build our own connections.

Theme of the future

Still missing is the third theme, which I might call "the Internet as database."

Because the Net gives us vast amounts of raw data, we have to use that data through brute-force methods, such as the Google search engine algorithms that deliver thousands of hits for any search. We need the combination of rich data with database-style structure to help machines improve the search experience. Ultimately, we would like a Web that gives up a few precise answers for each search, rather than thousands of vaguely on-target ones.

It's asking a lot of computers to master human intentions, but the semantic Web will one day tap tools such as FreeBase (still in its earliest stages of construction), combining user input to flesh out information with database structure to find relationships between the key word elements of a search.

How smart does such a Web get? Philip Tetlow just wrote "The Web's Awake," (Wiley-IEEE Press, $49.95) which makes a startling point: Complex entangled systems can lead to "emergent" behaviors that no one has programmed.

A living, evolving Web? Not any time soon, perhaps, but it's a future no one studying Net technologies should rule out.

Get $150+ in coupons in every Sunday N&O. Click here for convenient home delivery.