Bill Gates' Keynote Speech
at Microsoft's Second Annual CEO Summit
Seattle, May 28, 1998

[Due to the varying sound quality and subject matter of tapes, the information in this transcript may contain inaccuracies. Downloadable Word document are located at the bottom of this page]

MR. GATES: Well, good morning. I'm going to talk about two things this morning, digital nervous system and Web lifestyle.

Digital nervous system is the term that I came up with just a year ago when I was trying to articulate the idea that some companies manage information better than others. And every company is using digital tools to manage their basic operations, to generate customer bills, to manage their accounting, to do their tax work. But I'd say that very few companies are using digital tools to really manage the key information flow inside their company. If you look at things like strategic thinking, how easy is it to go back to decisions that were made in the past and see what the thinking was there? How easy is it to dig into the data and see patterns in the data and share that across lots and lots of employees? How easy is it, when a decision has to be made quickly, to pull together all the new information?

My basic view is that the digital nervous systems of even the companies that have put the most effort into this are very poor compared to what's possible. What's interesting is that the gap is not really a gap created by a lack of spending. In fact, the basic building blocks here are things that most everybody is already investing in: the PCs, the networks, the basic applications. The real gap is created by a lack of thinking about how much information can be available.

People have gotten used to not having at their fingertips a lot of the things that I think they would find very, very valuable. And so part of the exercise is to raise people's expectations. During the last year, I've had a lot of opportunities to go around and talk with CEOs about information flow inside their companies. You know, how do people find out things? And I was a little disappointed that these expectations were so limited. You know, the idea that it's not easy to look at sales results, it's not easy to see the status of a customer account. In some ways, I want people to be appalled by that and to say that's just a basic level of information availability that should be universal.

The term "digital nervous system" is kind of an interesting one. The analogy, of course, is to the biological nervous system where you always have the information you need. You always are alert to the most important things, and you block out the information that's not important. And companies really need to have that same kind of thing: the information that's valuable getting to the people who need to know about it.

There are a few principles that go along with this. The first one is, bad news must travel fast. I remember when we first got electronic mail, I started getting a lot of messages about accounts that we would win. So, somebody would send me mail and say, oh, we won the XYZ account, or we won this other account. And every time I'd get one of those messages I'd think, does that mean we lost all the other accounts? You know, if they're sending me every win, we've really got a problem because we need a lot more business than I can add up simply by these mail messages.

And so I set a rule that people weren't allowed to send good news unless they sent around an equal amount of bad news to go with it. We had to get a balanced picture. In fact, I kind of favored just hearing about the accounts we were losing because, in fact, bad news is generally more actionable than good news. When I got that message saying, we won the XYZ account, it's not like I changed my schedule for the afternoon. I just said okay, that feels fine. But when I get a message about an account that we lost to, say, a new competitor or for a new requirement that we hadn't really thought about, hadn't really prioritized, that could make a very big difference.

And I think large organizations have a real tendency to block bad news from moving around the organization. People don't know how to deal with it. Does it mean that somebody has to be blamed for this bad news? You know, is somebody going to get very upset about this bad news? And so having it just be a matter of course that you get the bad news out there, and everybody is sharing it and talking about it on an open basis, I think is really fundamental. And it's one thing that a digital nervous system is very good at doing.

In fact, we've got it set up in our sales tracking system so you can trigger a mail message to you whenever something is out of line. If there's a product that's not selling, as well as we might expect, if there's a geography that's not doing as well, you get a mail message, and that includes a link that lets you click and then see a complete range of data, and then you can dive down into that data and take whatever you're seeing, whatever you come up with, and then mail that around to other people so that they're aware of it.

A second principle is, the Internet changes everything. I'm sure you're tired of hearing the hype about the Internet, because the Internet is one of these phenomena where most of the predictions that people make for the next two or three years are overboard. But if you take more like a 10-year period, I think most of the predictions in some ways fall short of what will really happen.

As we get this new generation of consumers, who have gone to college, who have used the PC, who signed up for their courses electronically, who've used electronic mail as a way of staying in touch with each other and with their parents, those consumers will expect to be able to interact through the Internet. And whether it's with their doctor or their bank or deciding to buy a car, they're going to use the Internet as a tool. And that fundamentally changes a lot of things about how you stay in touch with customers, and even the rules of competition, who you compete with and how you invest to get ahead of those people.

It's fascinating that some companies are coming around to think about information flow inside their company because their customers are demanding access to information. For example, more and more customers are saying that instead of a bill being printed out on paper and mailed to them, they want to get the bill electronically. Microsoft doesn't want any paper bills. In fact, if you want to invoice Microsoft now, we force you to go online and do that electronically, because we've gone to a paper-free system internally. And it's easy for us to have requirements that vendors conform to that and use the electronic system.

And for a lot of companies, it's that external requirement that's making them think, wow, if our customers are expecting the information electronically, then maybe the knowledge workers inside the company should be getting the information that way. Other companies come about it in the other direction, where they think first about empowering their workers internally, and that makes it very easy to take the subset of that information that you want to provide the partners or out to customers, and get that out and make it easily accessible through the Internet.

The Internet is fundamental, partly because it's a critical mass. The innovation, the tools, the communications companies coming in to provide bandwidth, whether it's phone companies or cable companies, the hardware companies doing the equipment that will move that information at really unbelievable speeds, software companies like Microsoft building these things, the pace of investment is mind-blowing.

And these valuations of new Internet companies that sometimes seem disconnected from the profit and loss statements, the only thing that you can say about that is that it's hastening the investments. People are actually able to go out and build lots and lots of Websites, even if their path of profitability is five to 10 years out there. And so, it may make it tricky to be an investor and figure out which of those people really are on a great long-term path, and which ones aren't, but it is absolutely moving things forward at a level of speed with incredible investment that we've never seen with anything else.

And so, the Internet continues to evolve. If you used the Internet a year ago, or somebody gave you a demo, don't think that it stayed the same. The breadth of material, the quality of the material changes quite rapidly there, because of that investment and competition.

A third principle is, a customer should be at the center. That's pretty basic, but how many companies really find it easy to browse and bring up all the information about a customer? Not just the billing information, but also who at that customer company makes the decision, what types of use do they make of your product, what is the history with them? If they've had any problems, any support issues you need to track down, were you responsive to those problems? Is all that information available in one place, so that different parts of your organization that touch that customer can work with each other, can know who the others are, and always make sure they're on the same page in terms of what's going on with that customer? I think there are very few companies that even meet that basic level of information availability.

A final point is, every worker is a knowledge worker. What that means is, if you can get more information to the different employees, you can often take decisions that would have involved many people and many steps and get those made right at the interface to the customer, whether its granting a loan or deciding to give a credit to the customer for something that's gone wrong. If you can empower people by giving them information, and making sure that they see the guidelines for how those things are done, they can do a better job of being responsive as well as being more efficient.

I don't want to focus too much on the hardware technology, because I think a simple summary of what's going on with the hardware side is that the advances are coming fast enough that even the wildest fantasy you might have about information availability inside your company is feasible. The only kind of data that's actually difficult to move around is video data. And that still today requires a pretty substantial investment in your network. But making sure that other types of information, text information, numeric information, even still images of any kind, are universally acceptable to your employees who are on the road, to all our branch offices, and making sure that it's up-to-date at all times, that is very straightforward using the technology we have.

There is a constant trend to use PC technology more and more. PC technology is different because there are many suppliers of the technology, and because it's high volume, it tends to be lower priced. The main drawback that it's had is, it hasn't had the kind of reliability of very high-end performance that the mainframes or minicomputer type systems have had. There still is a gap there, but that gap is closing year-by-year. And I think most people would be surprised at how small that gap is today. It's one of the exciting challenges that the companies involved in the PC industry have, and it will probably be two or three years before that gap goes away altogether.

Another key element that I think is important to keep in mind is how approachable they are. We're starting to see new form factors, machines that fit into your pocket. In fact, I think most of you received one of these new pocket-sized devices, palm PC type devices, which really represent the smallest form factor that we've got. Inside that device is a processor that's faster than the best PC had only three or four years ago. And what we're trying to do is make it easy to move the information from that pocket device, to your PC, so that you don't have to enter the information twice. Your schedule, your contacts, your mail messages, all of that is easily available, and whenever you go near to your PC without giving any commands, that information synchronizes, and so it's always uniform and consistent.

As we get those new form factors, and as the screens become higher resolution, and we use the flat screen technology, people will change their behavior from reading information on paper to reading it off the screen. Today there are some types of information that should have already moved off of paper. If you look at the encyclopedia business, that has shifted very dramatically toward the electronic form. Partly, it's easier to update, it's lower cost, you get the pictures and sound. In the last year, the CD encyclopedias outsold print encyclopedias by 10 to one. And that's actually made the economics of the print encyclopedia pretty difficult. That's something Warren and I talk about from time to time, because he owns the World Book operation. But, he does have this one big advantage, which is that leather smells better than our CD. And we're still working on trying to get the CDs to look at prestigious and smell as nice. But, we haven't conquered that yet.

Another type of information that should be moving off of paper is anything that's a form, anything where you're filling out data. Forms that are inside companies, or even between companies, they are incredibly inefficient. You end up entering redundant information, you don't get any help, as you fill out that form. There are inaccuracies introduced by it. Those things should simply go away. And in the last year at Microsoft we have gotten rid of all of the forms. In fact, we'll be demonstrating some of the online systems we've created that are a replacement for those.

The shift toward doing things on the screen won't be a total shift until we get far better screens. And we've been doing a lot of experiments in our research group to see what kind of a screen does it take, before you'd be comfortable simply even to take the extreme case, a book of fiction, where the computer adds very little value, since you're going to read it in a serial fashion. How nice would the screen have to be before you'd want to read a book off of that screen. And it will probably be 5 to 10 years before we have the screens that have the right weight and the right resolution before that happens. But, the key point is, that it is going to happen, even if you take a pessimistic view, you'd have to say it's 10 to 15 years before reading will make that kind of shift. It will be small enough that you can take it to bed, and you know, use it any way that you've used the book in the past, but still have all the richness that comes with the electronic information.

During the last year I made an interesting shift. I used to read all the magazines on paper. Now, there are some that I read in a leisure way, like The Economist, ignoring this week's edition, but I still generally enjoy reading The Economist. But, all the computer trade magazines that I used to read on paper, I don't touch them anymore. Somehow they're still sending them to me, so they pile up, and I toss them in the garbage can. But, they're so much more up to date, easy to browse, on the screen. And I've got a better screen than I had a year ago, that I've made a shift there. And that is a very interesting thing, because most companies when they think about the Internet, have been thinking about it on a marginal basis, where they say, it doesn't cannibalize any of the existing economics. And so you just try to make sure your Internet site covers -- the extra revenue from that, on the margin, covers that special expense there. But, it's very  uncommon that it's actually paying back for the basic creation of material.

When you get to the point of cannibalization, then you have to actually think of the economics a little differently. And so although for most periodicals that's three or four years in the future, three or four years in a planning cycle is not that long a period of time. So the bottom line on the hardware side is, I think all the companies involved here, whether it's Intel on the chip side, or the system manufacturers, the advances are coming fast enough that there's not going to be any kind of bottle neck in putting together a pretty amazing system.

As I said, the building blocks are straightforward. You could not benchmark companies in terms of how they spend money in these things and get a very good correlation to what benefit they get out of them. There are a few things that are expensive that are worth spending money on. Making sure that the network that connects all your offices together is a reliable network, with enough bandwidth. That's hard to do, it requires very good employees, good relationships with the communications companies. But, I view that as something that's worthwhile.

Having a directory, which is the listing of all the employees and the group names, including the security access privileges that the employees have, that requires keeping it constantly up to date. If you're going to put sensitive information on this system, then you've got to use the directory in order to control who has access to those things. A critical element is to make sure you don't end up with multiple directories, one for file sharing, one for the Web, one for databases, because it is such an investment, that you want to make sure that it's only done one time.

In the extreme case, people are only using their PCs for document creation, and they aren't using the PCs for this rich collaboration and information sharing. It's absolutely the same PC that you need, whether you're using it in a broad way, or not in such a broad way.

There is one trend that's worth mentioning, and that is, there's a huge trend away from writing your own applications, to buying packaged applications. Finally, the packaged applications are starting to be flexible enough that with a modest incremental investment you can customize them to your particular requirements. And that's really a breakthrough, because in software there is only one secret, which is to use software that somebody already wrote and not have to rewrite it yourself. If you end up writing software, it gets to be massively expensive, not necessarily the first time you write it, that's when it's easy, it's maintaining it over a long period of time.

And of course, this year 2000 issue is just reinforcing in a big way what a problem it is to have a base of code that's unique to your company, that you've got to go back into and make sure that when the tax rules change, when you get something like 2000, you have to go back through that.

Well, here I've got some techniques to rate digital nervous systems. As I said earlier, I think the average grade in this type of scorecard today would be like a C- or a D for companies overall. The first question is, how good is your corporate memory? If there was a product introduction that was done very well, and somebody did a post mortem on it, and talked about what worked particularly well, and what didn't work particularly well, can an employee who wants to access that in, say, 60 seconds, sit down at their PC and call up that post mortem? Have it on their screen, be able to mail it around to the other members of their team? That's common sense.

I mean, after all, you paid for that product introduction. Your company learned something, even if some of the people involved moved on to other positions or they left the company. That learning is an asset of the company. And if that's a memo that's hard to find, or it's stuffed away in a file drawer somewhere, the likelihood that people are going to take the time to go and find it diminishes to zero very, very quickly. Whereas, if it's online, and you can type in a few keywords, you know, product introduction, give a date range, and that document comes right up, then people are likely to take advantage of it. And it's something that's not very hard, but it's not very typical that people have that.

Do managers have the same information you do? It's very typical that CEOs have lots of people who sort of scrub the numbers and present them to them in a nice way. The problem with this is that then if you see these numbers and you don't like them, and you call up your managers, they may not have the information in the same form. And so they have to take time and say, whoa, what are you looking at, whoa, send me a copy of that, geez, I didn't know that was going on. There used to be this thing called an executive information system, which was really bad, because it propagated this notion executives should have information that the other people in the organization wouldn't necessarily have. And so you get that impedance mismatch.

The ideal is to have an information system where everybody is looking at exactly the same thing. If you want to go to the highest level and see the companywide summary, it's there. If you want to dive in and go down to an individual transaction level, that's there as well. It's all one system, everybody has the same system. If you mail the view around you can navigate through it all the same. You know, the worst thing is to go to a meeting and have somebody present data. And then you're looking at your data, and it's somehow different. If you have just a single place where all those numbers are, that's very straightforward. All the information on a customer, I mentioned that.

Perhaps even more important is capturing customer feedback. Let's say your sales force is going out there, and the customer is saying, there are certain things that aren't high quality enough, or certain new features they'd like. What is the process for that information to get back to your product planning people? Is it standard that any customer comment would become immediately available to the people who do that product design? Unless that's institutionalized in the digital nervous system, it will all happen on sort of an anecdotal basis, and people would be impressed, wow, I happened to run into the guy who talked to this customer, and I heard they want such and such. And they'll be impressed that happenstance pulled that together. That's simply not effective enough if you want to respond the way that competition requires. And so having as part of this digital nervous system the easy ability to take all that information and have it get back to everybody and really ensure that people are reading it, I think is fundamental.

In our case, we have a lot of this, because we have people calling in, we get over a million calls a month where people are asking questions about our software. And many of those questions are clues that the software should be simpler or better. And so we really force the product design people to browse through that information and take the top issues and talk about what the heck are they going to do about that. And so having that customer feedback really should allow a leader to stay in the lead, because you're getting all that guidance if you use it in an appropriate way.

Making partners part of the system, this is to enable you to outsource things, without giving up the same kind of information flow that you've got within the company. Setting up the security system so that it works, making sure you're using the Internet standards the right way. There are some basic things to enable this. But, again, it's something that most companies don't have.

And finally, if you take a business crisis, and every company has, probably, it's own definition of what a crisis is. You know, is a currency change a big thing, is a competitor activity, a new governmental regulation, what are the things that might create a discontinuity, where you'd want to respond, say within a 24 hour period? And how easy is that to do, particularly as you're getting employees across many different geographies, who need to respond to those things?

I want to quickly give an example --of where Microsoft went through a crisis and I think it's a good example, because having some good digital information flow I think was absolutely critical for us to be able to respond on a timely basis. It was in November 1995, just after we'd shipped a wonderful product, Windows 95, people started saying that our strategy had not incorporated the Internet enough. And that was a valid criticism. You know, we had it on the list of things we were doing. We were doing some work. But, we hadn't given it the importance that it required. And so it was recognized as a crisis. And there was an unbelievable amount of electronic mail that flowed around this topic. You know, once people said, gosh, what do we do? You know, who has any ideas?

Then all the people who had sort of been thinking about this, at all levels of the organization, particularly some of the younger kids, who had come out of universities where the Internet was becoming common sense, you know, they were the ones saying, oh, you guys just don't get it. You've got to go look at this, you've got to try this. You've got it all wrong. A lot of the email was not very respectful. And you've got to have a culture that can deal with that a little bit. And you want to make heroes out of a few of those people who were particularly outspoken in saying that we really weren't doing the right thing.

We took that, and found the people who were sending a lot of that email. We went out on some retreats, so that reinforces the notion that just electronic systems alone aren't good enough. You've got to get the people face to face, particularly when you have a debate, when people are disagreeing. Typically on email, if we go back and forth more than three or four times, the wise thing is to send a piece of mail saying, hey, let's get together and discuss this, because it's very hard to really come to an agreement in an electronic mail exchange. You know, it's not a good debating forum. In fact, things can go on and on and on. And that  can actually become a time waster if you're not careful about knowing where electronic mail works and where it doesn't.

This story is a good one, because in the final analysis, it's got a happy ending. The ending is almost ironic, you could say, which is today we've got a situation where the U.S. government thinks we have too much Internet support in our products. And they're trying to convince us to take our Internet browser out of our Windows system, which is exactly one of the things we decided to do in response to people's demand for the Internet.

But, certainly, the response we got from the people in the organization was fantastic. So what were the keys there? The sort of open exchange through electronic mail, giving people access to the Internet. I think there are a lot of companies that are worried about this balance of, if you give employees access to the Internet, aren't they going to be out there goofing around, trying out things that don't relate to their job? The fact is that there are very easy ways to monitor these things. You'll have to decide explicitly what your policies are. But because all the Internet requests go through a central point, which is called a proxy server, you can actually monitor the things, as little or as much as you choose to.

And I'd certainly come down on the side of saying that, because there is so much information about competitors out there, there is so much news out there, and you want to get your employees kind of thinking of this as a key tool, really giving them pretty broad access with some guidelines is the best way to go at this. And even employees of a pretty wide description, your sales force, your product designers, anybody who is in that knowledge worker category, I think it is very, very important.

Most companies today still don't have electronic mail as sort of the standard way of communicating, and it's a pretty binary thing. It either is or it isn't. It has to be very reliable, and you've got to get everybody using it, or it's just sort of there as an adjunct. And it's probably more overhead than it's worth unless it becomes the primary thing. That's a key thing. And it's often the senior executives that are a little bit of the bottleneck in terms of making that switch. It's a case where leadership from the top can make a very big difference.

Certainly, in our response to the Internet thing, just having these tools alone wouldn't have been enough. We had to encourage a lot of crazy ideas to be put forward. We had to have the face-to-face dialogues that went with that. And then we had to zero in on what the important priorities were and get people marching towards a singular vision. None of these things were a substitute for the leadership that all good management comes from.

Now, let me give you a quick little tour of the kind of things that we do in terms of having information inside Microsoft. This is what's called the intranet, information that's just internal. And so you only get this information if you log-on and you give a password. And, in fact, the information that you see is completely tailored to who you are in the organization. The breadth of information you get, and we'll even get into some cases where you'll see that differentiation.

There are quite a few things on this page here, non-work things, the ability to search, events, go through and see the product groups, their schedules, their specifications. Let's go into workplace services. Actually, what I've done here is, I've taken a snapshot of our intranet and brought it down onto this laptop. I didn't want to have the risk of the network connection between here and the campus going down. So you'll see this is actually a few days old snapshot here.

Here all the basic things are set up. If you want to order a pizza or sign up for a training event, all of it is here. If you want to check something out from the library, if you want to send in a question. Let's go ahead and look at the head count tracking thing. When you have a company that's reorganizing all the time, and growing pretty rapidly in multiple locations, and yet you've got electronic mail, you often get mail from people where you think, who is this? You know, what group do they work in, who works for them, and you want to be able, just with a few clicks, to see their background, understand where they are in the organization.

Also, sometimes when I want to send mail out, say I want to send it out about the spreadsheet, I think who is the Excel product manager now, and you know I don't want to send it to the wrong person, but I want to make sure I copy the right people there. And so having this information online is a great thing. You know, when I come into this, it reminds me of all the people who work for me. And I can see the basic information, ACS's contingent staff. That's a big thing in our head tracking system, is that we used to only track full-time employees, and so it meant temps and contingents were this unbelievable loophole, where you could hire as many as you want. And then people would come along and say, well, oh, these are contingent staff people who have been here for years. We just want to convert them to full-time, it won't cost us any more. So it was a complete perpetual motion machine in terms of head count. So, now that's one of the things we track pretty closely.

I can go in here and click on anyone and see who works for Steve Ballmer, and what the numbers are there. I can go out and pick somebody's email name, in this case Deborah Willingham, and have her come up. Then, if I want to dive in under her organization, I can do any kind of filter. In this case, I can ask for everybody who is in Building 22, and it will go down and show me everybody who is there.

Any one of these employees, I can click on them and see quite detailed information. You know, the way this is set up, I actually don't have permission to click on the tab that shows their compensation. People in the HR department do. For some reason, they didn't give me that capability.

I can even go in here and say I want to change somebody's position. I want to move them to an organization, change their title. Now, the system understands the approval process, so most people don't have the ability to unilaterally do these things, but once you enter in this, it will trigger off an electronic mail that will then cause the right approval process to kick in. So here, we can just do these things. On all these things, there used to be a massive number of paper forms that related to all the things that I'm looking at.

Well, now let me go over to my electronic mail. This mail package is actually called Outlook, and these messages in bold are the messages I haven't read. I see I've got mail from Warren, he wants to play bridge. I've got mail from Nathan, he's got a hot new research idea. We'll hear about some of those things tomorrow. And then I've also got some mail messages here that relate to expense forms. When somebody files an expense form, it triggers a piece of email to say that's there for me to approve. And so, I just click on that, and I get in, in the browser, I get this little thing here, and I see that Nathan works for me, and is known as a bit of a big spender. He's got this expense report here. And so, you know, I go in and see, does that look okay? And then I have two choices. I can approve it and say, wow, that was a nice restaurant you went to, or I can reject it, or whatever I want to do. So, there I just click on approve. And, you know, usually on most of these expense reports, I just spend 10 seconds seeing that things are sort of in line, and then it's completely done. I see that now I've approved all those things, and so I'm done with that.

For somebody who wants to create one of these things, it's actually done with a spreadsheet. There's a structured spreadsheet with some macros, where you just type these things in. We try to make it pretty easy. And the macros actually can kind of notify you if it looks like it's an unusual expense. You can still submit it, but it just kind of warns you in advance that the normal sort of logic has highlighted something that's a little surprising there. And so, if you want to do the submission, you just go in and see the summary and push submit. So that's all Nathan or anybody else would have had to do to create something like that.

It's also very easy for him, at any time, to look at the status. He can see where it is. Has it gone through the system, did I reject it? Has the check been sent out to him, or anything like that? Actually, here's the status of all the different things that he's submitted.

Let me now go to another part of this Web, which is related to sales statistics. And everybody can customize this page, and particularly take the accounts that they particularly like to track. You can actually call up any account here and see lots of different information. I'll just pick Chrysler here as an example, and you can see the kind of things that we keep online. The account profile, what kind of hardware they have because we're always interested in the PCs, what's going on with their software. So, I'll look first at desktop applications and say, that looks good. It looks like we've got the share that we like there. I can go in and look at browsers and say, whoops, looks like we've got some work to do. There's only about a 20 percent share there. And then I can go over here into the opportunities and see what people are doing about those things.

And the account team, you know, for somebody like Chrysler, there will be multiple people involved. So we track the owners, and people keep this up-to-date. And there is no paper system. They're not sending paper around to each other. This is the system that they use to track all of these different things.

Let's go switch now to take the sales data and see it in a summary form. Now, this is the finance home page, and they have a lot of different reports. You can look at information in an incredible number of ways. We do our P&L by product, we do them by channel, and so there's a lot of rich capability that's built in here.

Typical standard reports are kind of -- you can use the standard reports, or you can go in and customize the different reports that you're interested in. So, I'll go and click on tools, and let's see, what do I want here. They reformat these pages often enough to keep getting results. And what it does is, it opens up the spreadsheet, and the beauty of the spreadsheet is that it gives you a lot of flexibility to play around with the numbers, put out charts, anything you're interested in. This may be a little complicated to explain. It's fantastic once you get used to it. It's called a pivot table, where you can take any kind of information and see it in different formats. I'll pick a little bit of additional information, and then I'll go ahead and ask it to create a report.

And so here we can see there's a report that is by product. If I want to take some locations and break those down, as soon as I do that, you can see it puts the geographies there. I can take my product stuff, and break that back up. And so, I can see things any way that I'd like. So now I'm just seeing things by geography.

And so, when I go off to visit, say, Japan, I'll go into this tool, I'll call up the information on Japan. I can see how much they're selling compared to the budget, and I can dive into it by time period, by product. I can have it jump out for me the things that aren't doing particularly well. And this is actually kept up-to-date on a daily basis. I have a tendency to only go in and look at a few top level things once a week. And then, on a monthly basis, I'll go in and browse around a little bit more than that.

One of the nice things is, whenever you see something that catches your eye, you can just go in and use the command that sends mail here, and you'll be able to share exactly what you're seeing with somebody else.

Two other final things I'll show, one is the purchasing. This is where if you want to buy something, you just go into all the different categories. One of the things we buy a lot, of course, is PCs. Here I'll say that I want to get this Compaq Workstation. And it's a standard configuration, but I can change and say I want a little bit more memory, or I want to get a monitor to go with this thing. And then as I get down to the bottom, I'll add that into this order. And a lot of different information here. And, again, when I submit this, it understands the logic. And so, if that was Nathan who submitted it, then you'd get in my mailbox, because he works for me, I'd get a mail item just like this, and then I'd have the ability to see what he's doing. I can browse down into it. And if I want to approve it, I go ahead and do that, and it goes off.

And so, we've been able to take the interface between ourselves and our vendors, whether it's our paper suppliers or the people who do catering in the lunchroom, or the people who bring in these PCs and put that all up here electronically. And the prices people are seeing, of course, are the special discounted prices, whatever we've arranged with that vendor because this is an internal Website where all this information has been put together.

The last thing I wanted to show real quickly is how we're using our Website to work with customers. And we're spending a lot of money today on advertising and direct mail. But we expect that those categories will go down somewhat. Particularly direct mail we think will go away altogether. And the Website will become the primary way that we interface with our customers. It's very low cost, and it's very customizable.

This is the microsoft.com site. This is what you see if you just come in for the very first time. But when you come in, you can go to what we call the personal information center, where you can pick all the different things that you might be interested in. So, you can see we have all these different newsletters. If you give us your zip code, then we'll talk about seminars that are in your local area. We have things on different products that you can choose or not choose. And as soon as you go ahead and submit that, you can decide exactly what format you want things sent to you in, and then you're all set up.

And so we're keeping track. Every time you come to the Website, we know that it's you coming back again. And so we're profiling what you're interested in and, therefore, presenting things to you as well as letting you choose those things. This has become our primary way of touching our customers. Developers in particular, when they come to our Website, they spend on average about 15 minutes, and our good developers are coming to our Website about one-and-a-half times a week. And so it really has substituted for a lot of things that we used to do on a paper basis. In fact, if we look at the phone calls we get to our support center, what's happened is, all the simple calls, all the short calls, have gone away completely. Now that means the average call length has gone up because we're stuck with the ones that are fairly complex. But it is a very substantial savings for us, and I think a great customer satisfaction thing.

Just another thing about this Website, because it's our primary way of interacting with customers more and more, we want to get a lot of statistics about what's going on. Here we can see, was the site up 100 percent of the time? We can see how many people came in and looked at it. We can go in and see for any part of the site what the traffic looks like. You always see a huge decline in traffic over the weekends, which is logical, it makes sense.

You can go down for any of the different areas, and you can see where people are coming in from? For example, what country are they coming in from? So, if I pick a site here, and go, and then if I go in on the user attributes, I can see for this particular part of the site what country people have come from. And, of course, because it's an English page, it's overwhelmingly the United States, and it's not too many people from Belarus or Malta or different places. It is fascinating, though, that because it's a worldwide system about half the hits we get in total come from outside the U.S. And so it means all the work we do to prepare material up here is leveraged on a global basis. We take about a third of the site and translate it into different languages. But for some of our customers, like developers, that's not very necessary.

So, anyway, that gives you a quick glimpse of how we're using the intranet as a tool. The fascinating thing about all those applications I showed you is that the development time for each of them is about three months. Now, that doesn't mean we only spent three months, but we can cycle through a version in three months. And then every one of those pages has a peg at the bottom where, if you don't like the way it looks, you click and send electronic mail in. And so, after three months, they have a lot of feedback to go back and do the next rev of the system. On a lot of those, we're on about the third rev. I don't think they're as good as they need to be yet. The interface is still a little bit clunky, but a year from now those systems are going to be dramatically better, and they're not dramatic projects to get those things going and working in the right way.

I wanted to mention two other Websites that are pretty interesting. One is an electronic commerce site that we worked with the Gap to do. And it's really a fun one. I encourage you to go out and play around with it. When you think clothes buying, isn't that sort of the ultimate thing where you want to go and see the merchandise and, of course, a lot of customers will always want to do it that way. But they've pushed the technology to the limit. They've got the ability to interactively dress up a model, see what it looks like. They've got a style advisor if you don't know what things should match with what things. This site has all of that taste built into it. They can explain all these different fits of jeans, and things, which are way beyond my understanding. You get 3D views.

They also make it very personal. If you ever mail something off to an address, they track that. And so, if you have some people you like to send gifts to, when you come back in, you just select that out of your personal address book, and never have to re-enter the information. You could also fill in special dates, birthdays or anything, and a couple weeks in advance they'll send you a piece of mail, and there's a simple link in there. You can click on it and go out and simply buy a gift and have it sent with plenty of time.

Here's what the home page looks like. And you can see, they've done a very good job retaining their style, their unique look. And then here's just a few shots of the model you can take and make it look any way you want. You can try out different kinds of clothes, summer clothes, different things, and then anything you like you just click and you can buy.

Electronic commerce really is exploding, and I think that's the leading-edge application. Part of the beauty of electronic commerce is that it essentially pays for itself. It's clear that it's just another store. It's a store with a little bit different way of spending money. But in the case of the Gap or Amazon or many others, they're seeing not only a lot of traffic, but an opportunity to be in a leadership position.

The other thing I thought worth mentioning is some of the work we've done with Boeing. They've gone in the last couple of years from being a company where electronic mail was there, but not the primary means of communication, to now using it as central to how they stay in touch. In fact, it was pretty educational for both us and Boeing, because when they first put it up, they had some of the servers just around the company in different locations. And over one of the holidays a janitor came in and recycled the power, and one of the servers went down. And a lot of the employees were using mail, then found themselves not very productive. And so that really pointed out that this has become a totally mission-critical application, an application that you want to measure the up time, and be right up there at about 100 percent, as much as any system in the company. And so there's a lot that's gone into that, and we have some shared metrics that we use between ourselves and Boeing, where we can see constant improvement in getting to that 100 percent level.

This is a great example, because this one went very quickly. It went over just a two-year period from a couple thousand users to 130,000 users, to now over 3 million messages a day. And these aren't just simple messages, like yes, that plane looks good. These are messages that include engineering drawings, schematics, test data, very critical things, and they're employing not only inside the company, but to partners, as well. So they'd even seen on a statistical basis things like overnight packages go down, and really redefining the role of partners, letting, for example, partners in Japan participate in a very new way.

Well, now I want to switch a little bit and talk not about the company centric view of this, but the user centric view, which is what I call Web lifestyle. It's very typical that there is no single application that drives you to the Web. You know, you're not going to decide to learn how to dial up, and remember your password, and the muck you kind of have to go through to get used to this, just to do your banking. You're probably not going to learn how to do all that just to get sports scores. There are very few applications that alone can justify it. But, once you have a critical mass of applications, where you are getting online on a regular basis, that's when you expect to be able to use it for many, many different things, whether it's studying a disease, or learning a new hobby, or even looking up the new movies or restaurants that you might want to go to that weekend.

The top Web sites today are all what are called portal sites, they're sites where they help you find things. They started out as search engines, and now they're trying to be as broad as possible, to let you find different information. AOL, of course, is the most popular site. They include the dial-up capabilities, they're billing you monthly. The second biggest site is Yahoo, that is a pure Internet search site, and doing very well.

Some of the others up there, like GeoCities, are places where you can create your own home page, and so a lot of people are taking simple editors and putting up personal information, and kind of enjoying the fact that people come in and comment, and send them electronic mail. If we take out these portal sites, and look at the next top 10 sites, it's kind of interesting. Blue Mountain is a site that lets you send postcards to people, kind of neat interactive postcards. And it's really come out of nowhere to be very popular. ZDNet is a site with computer information, it's not surprising that's popular, because of course, everybody who uses the Internet uses computers. And so those sites have been among the first to catch on.

Amazon's the bookseller, Disney has done a great job for kids. MSNBC is a big news site, that we're a joint partner with NBC on. ESPN is today the top sports site, although there's a few others like SportsLine that do very well. And then you've got entertainment sites, Sony and Warner. And when I say these are top 10 sites, these sites are getting literally 300,000-400,000 unique users coming in every month, even the least popular site that's up there.

There are 24 percent more shopping sites, just in the last six months. Amazon's traffic has gone up 81 percent. Barnes and Noble came in and it's new and doing very well. There's an interesting type of selling that's emerged, this auction style classified, where you basically put something up there, and you put a date by which it's going to get sold to someone. It started out with used computers, which was a very poorly mediated market, but now it's spread to lots of different products that are being used.

If you track it as a percentage of these product categories, there are very few product categories where it's gigantic today. In cars, about 2-1/2 percent of transactions are closed over the Internet. Now, that understates the influence. Because, you have a lot of people going up and looking at the blue book trade-in value, or looking at the new models, or getting the comparisons at sites like Carpoint. But, the actual transactions are still a few percent. Travel will be crossing over about 1 percent in the next few months. And I expect that one will grow pretty dramatically. Advertising is probably the most disappointing thing on here. A lot of these Web sites are assuming they're going to get big advertising dollars, and that's still a pretty distant hope. If you took all the business plans for Web sites, and took the expected ad revenue, it would probably be about 10 times what people will get in the near term. So there is still some disappointment in that.

The most explosive category has probably been retail stock trading. A little bit over a quarter of retail trading, and I am distinguishing that from institutional trading, is now done through the Internet, whether it's E-Trade, or Schwab, or other people. And I think that is just an indicator of what will happen in other categories. Those people are a little higher income, a little more interested in doing things on a rapid basis, and so that's the one that has really gotten big, before any of the other things.

Probably the biggest block to making this all happen is making it simpler. And you know, it is still way too complicated to install an application, to get a PC, to deal with the funny error messages when you try to connect up to the Internet and it doesn't work, to remember all these passwords. You know, we are pretty guilty when it comes to not simplifying things. In a sense, you can say it's good news, because we have new versions of our products coming out that are a lot simpler, and it's a good motivation for people to want to get in and take advantage of those new things.

Part of the simplification is going to be eliminating concepts. Today you have to learn different ways of searching your files, and searching your mail, and searching the Internet. It's all different. And so building the browser in, and making that just one set of commands for everything, is going to be a very big deal. We've been going through our error messages recently, and we have some really great error messages. This one, even in an engineering crowd, people are a little confused. It says, the DHCP client could not obtain an IP address. If you want to see DHCP messages in the future, choose yes, otherwise choose no. Well, the nice thing about that message is, it assumes you know what DHTP is, but it's helping you understand yes and no. And, you know, what are you supposed to say? Well, it's such a great message you'd probably say yes, because you definitely do want to see that again.

We've actually gone out and looked at households and seen some of the confusion that's taking place as people are trying to install these devices and learn different things. And we made a little video that chronicles some of the challenges. So let's go ahead and take a quick look at that.

(Video shown.)

MR. GATES: There's a fair bit of work to do before we can say this is a seamless, integrated system. One of the interesting dialogues we have with a lot of companies are about IT spending, how it breaks down, what are some of the best practices. There's probably more data on here than is easy to read and absorb, but this is just a basic outline of what we consider the key categories, and it sort of shows what the initiatives are that are going to bring you to those categories. And then in key areas we picked partners that are going to be fundamental to us as we leverage off of their particular expertise, whether it's SAP or Compaq or (inaudible), people who are particularly focused in that area.

Our IT spending as a percentage of our corporate spending is a very substantial thing. It's about $500 million a year. It's interesting that the biggest single category is actually the wide area network, connecting up the global operation and making sure that runs on a very reliable basis. Our PC percentage would tend to be a little bit higher than other people. We have over two PCs per employee, which seems wild, but if you include the amount of test labs that we have, and the fact that people want to work in parallel. In our case, that makes sense, in most companies you wouldn't want to do that. And it's a really valuable exercise when somebody comes in for us to compare this framework and look at different people's trend life.

The goal is reasonably straightforward, it's to be able to use the new technology to drive down some of these costs, while at the same time getting more value of out the system. There's a tendency to very much focus on the cost because that's the most measurable thing. One of things that's frustrating about digital nervous systems, the empowerment of the employees, is that it's a very hard thing to measure. And, in fact, if you do it and you're getting a lot of value out of it, and your competitors do exactly the same thing, to some degree that's a neutralizing effect. You would have been worse off if you didn't do it, but it's not like it jumps up in terms of your P&L because it simply raised customer expectations throughout the entire business.

I just have this one slide that mentions a particular products because there is a big break through in terms of managing PCs that will come some time next year as we roll out the next version of Windows NT. One of the big problems with PCs is they're all different. And you can't make them the same. And so, you get surprised in terms of what works on one, what works on another. And being able to centrally make them identical, control what goes on, that's been a holy grail, and people have vacillated back and forth because most of the approaches to that in the past would have forced you to give up the benefits of the PC, the fact that it is flexible and affordable. And that's where we think, with this new approach, we really give people the best of both worlds.

And we have a lot of pioneering customers who are working with us on this, and we hope to have some very concrete numbers for what it can do in terms of that PC support number, which you saw was 10 percent of the IT budget that we showed there.

The opportunity in this business to make better products is quite incredible. Our R&D has been growing at over 20 percent a year. Fortunately, some of our other expense categories have been far better behaved. So, I get the latitude to increase R&D. We're up at about $2.6 billion, and next year we'll certainly be over $3 billion spent on the R&D. In fact, the only thing that holds that back is actually how easy is it to hire people into the organization.

That's why we've been a big proponent of relaxed immigration rules. Although we've been doing most of our development in one location, we actually have been forced to go outside the U.S. and do more and more. We opened a site in Cambridge in the U.K. We have a pretty big site in Israel. We're doing a site in India. We're doing a site in China. The beauty of that, of course, is you get to tap the global IQ pool. The drawback is that coordinating people in those different locations is very tough. Throughout the years we worked with IBM, the inter-site rivalry and the inefficiencies they ended up with there was something we said, okay, we're never going to have that. Well, now, here we are, we're at the size where we don't have a choice. We'll just try and manage it a little bit better.

The vision is a pretty straightforward one, and that is that PCs can be natural to work with. The idea of doing speech recognition, having visually seeing who is sitting down at the PC, what gestures they're making, these are things that are clearly within the next five to 10 years. And so, the PC form factor we think of today will be quite different. It will just be pervasive in the environment.

Some of these problems of making the PC natural have turned out to be pretty tough. For example, speech recognition is something that the industry thought it had conquered or was close to conquering quite some time ago. But it turns out that unless you have sort of common sense, there are a lot of ambiguities in speech. A good example of this is just the phrase "recognize speech" sounds a heck of a lot like "wreck a nice beach." And yet, if you say "recognize speech," and on your computer screen it prints out "wreck a nice beach," you're going to laugh because you'd never make that mistake since you understand what the topic of the conversation is. Well, computers haven't had that benefit. And so that's a lot of what we've been doing, and other companies like IBM and some of the start-ups have been doing, to build in speech capability.

It's tough to say, because people are a little cynical about this now, but I can pretty much guarantee in the next five years, this will be a standard capability. Some people are already using the dictation products that are out there successfully. I tell you, about one out of three of the people who buy the dictation products end up using them on a long-term basis, which is a huge improvement. It used to be, you'd buy it, and 99 percent of the people were back to the keyboard. They couldn't really explain why. It was just sort of a sense of frustration that they couldn't predict when it would make mistakes, and the errors were kind of surprising and difficult to correct. And this is one of those threshold problems. A 96 percent accurate rate, there's very little market, but then you get to a 99 percent accuracy rate, there's a gigantic market. And I think we are definitely very close on that.

So, my overall picture here is a very optimistic one. I think doing digital nervous systems is just good strategic thinking, stepping back, looking at what other companies are doing and trying to gather the best practices there, some initiative from the top down. I do think that there's some time here for companies to get ready for the customers living this Web lifestyle, because it won't happen overnight, but it is happening, and there are certainly things in almost any industry that that are going to change. And I do think it's important to keep in mind that the tools we're talking about will be dramatically better than the ones we have today. Even if it didn't change, the popularity would increase because you're just getting more and more people involved. But the thing that will accelerate that is that the ease of use will be so much better than it is today. That, even more than the fact that the price is going down, is what will drive things in a big way.

So, it's a pretty exciting time. There are even more things, fun things, ahead for this business than all the neat things we've done in the past. Thank you.

(Applause.)

MR. HERBOLD: Why don't we take about five to ten minutes of questions. And then we'll cut for a break. But if you've got questions, just raise your hand. We've got floating microphones here to make sure everybody understands what the questions are. Are there any out there?

QUESTION: Bill, you mentioned e-commerce is about to really explode and take off, I'd be interested in your thoughts about how important security of the e-commerce is. And if it is real important, how you see that problem being solved.

MR. GATES: Well, it's interesting, when people use credit cards, you know, they give the credit card to the gas station attendant, they give the credit card to the waiter, and so their credit card number is widely available. Actually, when you do transactions on the Internet, we do them in a special way so that the merchant never sees the credit card. That's encrypted in a way that only the bank can see the credit card. So creating an artificial charge slip is actually harder when you do an Internet transaction than it is across the normal kind of transaction.

The interesting thing about e-commerce is, if somebody puts something on your bill improperly, you can just get it taken off. The thing that's the most difficult, the most irreversible, is if somebody reads your electronic mail. You know, the lack of privacy of seeing salary data, or seeing a plan, or some personal thing, that's not easy to reverse. And people are using these systems today for electronic mail. The password is a weak link. Because people tend to pick passwords that are easy to guess or they write them down, or they'll use the same password on a very insecure system that they use for a very key system. And so we need to move away from password only, to having a smart card, plus a password, which is similar to what you do with the automatic teller machine today, where the PIN number is the equivalent of a password.

So, in the not too distant future, the PC keyboard will have a smart card reader. And so they have to have both that card, and the password to get in and authenticate yourself. I don't think that security is going to hold back electronic commerce. There's this unfortunate regulation in the U.S. that doesn't let us export software that has reasonable encryption. And so the countries outside the U.S. are really complaining and actually buying a lot of third-party stuff, and that's making things quite messy there. But, it shouldn't hold things back.

Also, when we talk about electronic commerce, there are two categories that are important to think of. One is simply taking buyers and sellers who would have done business anyway, and they're simply moving from paper ways of doing those transactions to doing them electronically. With that, you can get big numbers very quickly, because you're just shifting across. More profound is the matching of buyers and sellers who wouldn't have done business, wouldn't have found each other if it wasn't for the Internet and the ability to reach out and grab, and locate each other. And most of these reports that predict things, you know, show the big numbers without really distinguishing those two from each other. And so they get the really big numbers, mostly from the shifting of business away from paper.

QUESTION: Bill, in your talk you mentioned that the gap between servers, mainframes, and PCs was closing pretty rapidly. Would you mind elaborating a bit on that?

MR. GATES: Well, the PC, of course, started out as a low-end device. But, the improvement in the chip technology for the PC has been far more rapid than any other part of the computer industry. And so today PC servers are a huge percentage of the volume of computers people are using. Most Web sites are based on PC technology. Most branch banks are based on PC technology. The one place PC technology hasn't been used is in the central data center, where people have their primary database. There is still very much a tendency to use mainframes, or very high-end UNIX systems for those central systems. And there is enough interoperability, and ability to exchange data that you can actually move all your new software development onto the PC and use the tools there, that are very Internet focused and very modern, and still keep the data on those larger systems. And so you can move software development before you actually move the data.

There are companies that are moving their primary databases onto PC technology. Those are pioneers today. The key elements that the industry are working on are things like clustering. It turns out that clustering a bunch of machines together, so that if any one of them fails, the rest of them can continue to work, or so you can upgrade one of the machines, without taking the system down, that's a very fundamental technology. And it's really the last fundamental architectural things that mainframe has -- mainframes have had that PCs have not. And that's just coming into PCs, people like Compaq, and HP are being very aggressive in taking the clustering technology and moving that down there.

So in the next few years, you will see people who run their companies purely on PCs. Microsoft is fairly unusual in that respect today. We run our SAP system, which has about 5,000 active users, on NT-based systems. People will continue to have the old systems, which is why we work a lot on interoperability. But, the raw performance is definitely not a problem. The things that are being filled in, much more of the 24-hour a day operation, and a lot of the utilities that you need for this high end data center type capability. If you ask Gartner, they would say it would be five or six years. I think they're wrong. You know, I think they're always conservative, they're not tracking the things that are going on. I'm much more in the --you know, it's in the next couple of years those pieces will fall into place.

MR. HERBOLD: Let's make this the last one, so we can keep somewhat on schedule.

QUESTION: Bill, I'd be interested in your comments on best practices, with the email and intranet systems, you said just a little about it possibly, in some cases, being overused. And I have a very specific question, within that, and that is, do you, for example, have your email address available to all your employees for any comments, or do you find you have to screen them in some ways not to be overloaded?

MR. GATES: Yes, my email address is well known to every Microsoft employee. And they can easily look it up in the directory and send me mail, whenever they want. And that does not end up being a problem whatsoever. I mean, people will sometimes send me mail about their favorite charity, or some random thing that's going on, but the way my mail interface works, you know, I can look at that and forward it on to someone, or just send them a quick reply that says, hey, I don't have time to be able to do that.

I get on a typical day a couple of hundred email messages. And the surprising thing is, I don't get that many more than other people in the company, 200 is not that unusual. The average might be closer to 100. But, there's a lot of people up at that level. And you get very, very quick at processing the email. Also, there is a lot of etiquette about sending email. People, you know, they think twice before they send a message. And I do find out sometimes things about morale in a group, or a problem in a group that I wouldn't find out otherwise. So it's a kind of open-door policy that works fairly well.

Now, outside the company my email -- if you send mail from an arbitrary mail address, it actually does go to an assistant's folder. And then that assistant looks at it, to try and see, you know, what category does it belong in. If a person writes it as though they know me personally, she'll go and drop it in my folder, and it will come to me. If it comes from certain domains, like if it comes from Compaq or Intel, then it passes through to me immediately, just like mail from inside Microsoft. Also, if I ever reply to a piece of mail, somebody sends me mail and I reply to it, then that person's email address is automatically promoted to be one that comes to my inbox directly and it doesn't go through the assistant. So we have had to go to a managed approach for the external email, but not for the internal email. And, you know, I'd be very reluctant to do it, because it is sort of a barometer on what people are unhappy about, or thinking about.

MR. HERBOLD: Thank you very much, Bill.

(End of presentation.)