We saw big advances in natural language. That's an area where for almost a decade we've been making major investments, and just now the speed of the machines, the size of the memory, all of the elements have come together with this new software to allow us to bring this into the office and allow it to be more effective.

And, finally, we saw a component architecture where people use the pieces that they're interested in, including the ability to plug new things in that leverage off of the richness that the Office platform is able to provide.

Now, there's a number of enabling pieces that have to come into this besides the software. We do need faster processors. Fortunately, we have Moore's Law working for us as well as billions in investments that the chip industry is making. So, in the next few years we will have speed that's more than adequate for everything we saw.

We also want the presentation to be easier to work with, the animation, the motion video. And so the graphics capability has got to improve, and that's on a very fast track as well.

Some of these new input techniques will also be important. The video camera that lets you video conference, and it will also let you make gestures that the Office software can understand. Voice input, pen input, we believe those are all complimentary to the keyboard which will continue to be a primary way of getting lots of text into the system.

Now, to make using all of this easy, the communications hardware is going to have to be built in. There's many different generations of communications. Within an office in the local area network, we're moving up to a hundred megabit speeds and we're getting the kind of quality of service guarantees that allow you to pass audio and video across that local area network. But the hard problem is when you move to the wide area network and you're trying to reach out to other people, what kind of connectivity do you have?

If we just use today's phone system with modems, that's what we call narrowband. Sort of the ultimate speed we'll get out of that is about 28.8 KB that's moving into the mainstream. But an important characterization to it is the simultaneous voice data, DSVD, and that really lets you know whether it's discussing a contract or getting product support or ordering something out of the catalog. It really makes it far more social and yet gives you the richness of the computer display as part of that conversation, and being able to hook up those connections by finding a Web page and saying that you want some advice, you want someone to talk to you. That will all become very seamless now.

We also need to take narrowband communication out and make it easily available on a wireless basis for people who are working in a mobile fashion, and that will take a couple more years before the infrastructure is in place.

During that same time period, though, we expect a very rapid move to the next generation that we call midband. We're very excited about ISDN. The ease of setup is improving. The price of the hardware is coming down, the monthly charges are coming down, and ISDN, because it's about five times faster, really makes a qualitative difference.

Today those images that seem a little bit slow out on the Internet will be very, very snappy. Still image performance at midband is really, really excellent.

Another flavor of midband that's not quite as far along but will also be important, is the cable modem, taking the coaxial infrastructure that the cable TV companies built for broadcast video and using that to tie your PC in at data rates about the same as ISDN, or perhaps even a little bit better.

At midband, we can do some video capability, but it's a little bit like still images are in narrowband. It works, but it's not perfect. For example, you wouldn't want to watch a movie sitting there with a midband connection. For a business meeting it works, and I think it would be very popular.

The ultimate is still broadband, and that means providing data rates of several megabits or more. Now, this has been the Holy Grail of the so-called information superhighway. A couple years ago some communication companies were really talking about how many millions of people they'd connect up in the next few years. Well, a little bit of reality has snuck in here and they're realizing that the revenue opportunity for connecting broadband up to homes is going to take quite some time to develop. The prices have got to come down. The applications have to be far better than they are today.

And so the concrete evolution is moving towards midband, and then using that as the base point to eventually get to broadband. Broadband will happen, and it will be important. Using technologies like the optic fiber, asynchronous transfer mode and ADSL. We've got that in front of us but not as a revolutionary step, not as a big bang, simply as something that we'll move to step by step.