First, we had to learn how to write a story in Cyberspace. We had to embrace multidimensional journalism, by which I mean narratives that the reader could follow in any of myriad directions. One of the most difficult things we had to learn this about the new paradigm was that it was essential to give our readers the right to move through a narrative following their own paths. We could no longer assume a vertical story that began with a lead, followed by reporting that tapered down in relevance and importance.

All of our training had taught us that we were the gatekeepers of information, and we dictated by the layout of the paper, especially the front page, what was important. We had trained generations of readers how to read The New York Times -- whatever was in the upper right-hand corner was the most important news of the day, anything below the fold was less important than anything above the fold, etc. Now, we had to formulate our reporting in ways that asked the reader to learn a new set of clues as to the our judgment of a story's importance and that allowed the reader to navigate information in ways that were never possible before. We needed not only to learn how to make it easy for people to do this but to respect their right to do it in the first place.

The second thing we needed to do was to let the medium suggest its own paths and destinations. This was a little unsettling at first, but we soon discovered that on the Web we could help provide context for the news stories that we covered and deeper understanding of issues and ideas. I'm going to show you some examples of how one such new-media reporting paradigm, the computer simulation, evolved in CyberTimes through four iterations.

• We began with a relatively simple simulation, showing people what would really happen to their own income and taxes under the plan that Bob Dole was proposing in the 1996 presidential campaign. Readers could enter their own tax facts -- income, number of dependents, marital status, etc. -- and the simulator would produce a close estimate of their taxes under the proposal.

• We did a simulation of the spread of HIV in a heterosexual population, which was our first Java program. I couldn't think of any way to put it up on CyberTimes because my only mandate was to cover the social, political and cultural aspects of the Internet, so I framed it around "computer simulations, new-media tools for on-line journalism," which was how I saw it but I'm sure must have mystified our readers. This simulation, written by Peter Wayner, was far more complicated than the Bob Dole tax calculator and more expensive to produce. But in the end, it was far more valuable to the reader because the more you played with it, the more you understood the measurable impact of different levels of promiscuity and what you were up against out there in the world of heterosexual encounters.

• We created a program about breast cancer, which was designed to complement our Web version of an extensive special section in the newspaper on women's health. At that time, I realized that women must feel like Ping-Pong balls, because one week the paper would quote some experts who said women over 45 should have a mammogram every year, and the next week, we'd report that another group insisted that only women over 50 needed annual mammograms, and yet another said that women in their late 30's should have one every year. I asked Alison Stuebe, my producer at the time, a talented programmer who is now in medical school, to create a computer simulation that would explain the factors involved. She produced "Why Doctors Can't Agree," a simulation that went far beyond what I had asked for, listing and weighting all the different factors that experts used to determine at what age and how often women should have mammograms. She set it up to let readers assign their own values to all the criteria and then see what age and frequency of mammograms their own values would dictate. The interesting thing about this is that it not only educated readers about how such recommendations are made but forced them to think through the issue in ways that merely reading about the criteria never would have.

• Finally, we did a program, again by Peter Wayner, to simulate segregation. It was based upon a controversial survey which concluded that most people feel uncomfortable living in a neighborhood whose makeup included less than 13 percent of their own race. The simulation showed that it is the natural inclination for people to segregate. Of course, we had to present warnings in the overview that this was just one theory. But it help readers understand how you can anticipate areas to segregate based on different variables, including degrees of comfort with various levels of racial diversity in neighborhoods and geographical factors like rivers, highways and railroad tracks that present natural or man-made boundaries to a neighborhood.

The point I want to make about all these experiments was that they engaged readers in ways uniquely enabled by the medium. If we continue to give people reporting in traditional newspaper forms year after year in this medium -- basically transposing the printed word to the computer screen -- I don't think that newspapers can survive for long. Now we have a tool that can do many more things than simple text. Yet, it's sometimes very difficult to let that medium dictate its own rules and parameters. It goes against all of our traditions and training and depends on skills that most of us have never developed. This is a major problem, because at The New York Times, for example, our competition on the Web is not just The Washington Post or the Wall Street Journal; it is also television networks like CNN, hybrids like MSNBC and Internet information services like CNet -- news organizations that bring very strong multimedia assets to the medium.

Yet, we surely can compete, and we can compete well. Instead of just telling someone how something works, with computer simulations, we have the opportunity for the first time not just to share information with our readers but to help them achieve a greater understanding of complex issues through the experience of testing theories and variables and relationships in their own ways. I think that simulations are an easier way to understand things, and that belief has been reinforced, at least anecdotally, by the feedback we have received. We found that the number of people who spontaneously wrote to us about the value of simulations exceeded the feedback and response we got on more traditional reporting. As journalists, it is our responsibility to our readers and our publications to learn to use these tools.