Remarks by Bill Gates, Chairman and Chief Software Architect, Microsoft Corporation
Gartner Symposium
San Diego, California
March 29, 2004

ANNOUNCER:  Ladies and gentlemen, please welcome back Michael Fleisher.  (Applause.)

MICHAEL D. FLEISHER (Chairman and CEO, Gartner):  Good morning and welcome back.  Hopefully, we got your creative juices flowing this morning, made you think a little bit differently about the world today, tomorrow and three, five and 10 years out, and that's exactly what I hope to do as well in our next segment.

So without any further ado, let me welcome back, it's my pleasure to welcome back to the symposium stage, the chairman and chief software architect of the Microsoft Corporation, Bill Gates.  (Applause.)

Bill, first of all, thank you for being here, and thank you for taking the time.

We started this morning really trying to get everyone to take a long view, to start to think about innovation and technology and where the world will be headed over the next three years, five years and 10 years.  And so that's where I'd really love to begin our conversation today is hearing from you kind of your view of what is 10 years, not three years, not five years, but what does 10 years look like?

BILL GATES:  Well, I think it's fun to talk about this because the rate of advance is so incredible, and not just in a numeric sense.  The whole way that we interact with systems, the way we write software, the way we administer these systems, the way we collaborate, it will be very, very different.

And we have one thing that helps drive this forward is the magic of Moore's Law, the exponential improvement in the speed and the storage, and then we get to put on top of that the software that connects all those things together.

So 10 years out in terms of actual hardware costs you can almost think of hardware as being free.  I mean, I'm not saying it will be absolutely free, but in terms of the power of the servers, the power of the desktop machines, the network will not be a limiting factor, wireless technologies will have come in and created, whether it's in the consumer space or in the enterprise space, ways that you're connected all the time -- 802.11, Ultra Wideband.

And so in a sense we can just imagine, OK, what would the ultimate empowering tool for our workers be?  You know, if every meeting they attended could be recorded, a transcript created, easy to search, notified people who ought to know about things; could we do that?  The answer is yes, absolutely. 

Many of the Holy Grails of computer science that have been worked on for over 30 years will be solved in this 10-year period.  Speech being in every device, that will be solved.  Having a device that's like a Tablet that you just carry around and use with ink, being able to record different interactions, being able to take business information and see it in a visual way so that you take sales data, forecasting data, get notified when something is surprising, this whole model structure will be important.

In fact, that's one technology that cuts across a lot of these advances.  We'll model IT systems to make them behave in very rich, automatic ways.  We'll model your business processes so that instead of writing lots of lines of code to customize software for you versus some other company, it's just going into this visual model and saying, OK, are approval processes slightly different, are payment processes slightly different?  So we won't be writing as much code, we won't have the kind of complexity we have today.

MICHAEL D. FLEISHER:  So let me pick on one thing that you talked about, speech.  Ten years ago, we would have said speech will be solved today, in 10 years.  So why in the next 10 years?

BILL GATES:  Well, there's an actual curve that shows the error rate of the machine versus the error rate of humans, and we've had, even starting in the '60s when DARPA first started funding this stuff, dramatic improvement at a very predictable rate.  And today if you take either small domains where the vocabulary is modest, results are super, super good. 

And if you get rid of the advantage that humans have, humans are very good at two things that the computer is not yet good at.  One is background-noise elimination and the other is context, that is, knowing, based on who you're talking to and what they've said before, of all the ambiguous possible things that they might have said, it can resolve, whereas the computer foolishly would pick the wrong thing.  This is called the "wreck-a-nice-beach" problem, because if you say to a computer, "Recognize speech" it will sometimes say, "Wreck a nice beach."  (Laughter.)  But if you understood the context you might think that was a pretty silly thing.  And subconsciously your mind is eliminating all sorts of silly ambiguity.

If you take a noise-free chamber and you do just random words, there's no context, the difference between humans and computers today is about a factor of two, and we know how to get rid of that.  Then we have to do the noise elimination and the context.  And we're investing -- this is one of these things where, yes, people should be skeptical because people have gotten up on stage and said this before.  In our case, we're putting our money where our mouth is, so to speak, in that we're investing at record levels in these breakthrough technologies, because we see that roadmap and we have the exponential hardware improvements to help us.

So when somebody says 10 years, I feel very safe saying that your PC, you'll dictate to it; your cell phone, you can talk with.  We had a recent contest in China where they got the very best typist they could find and compared with our dictation software and our dictation software beat the very best typist.  Now, Chinese is helpful because the keyboard isn't very good for alphabets -- (laughter) -- but this is a super-trained typist that's using this.  So you get crossover for Chinese and Japanese, which is just happening now, way before you get it -- probably three or four years before you get it for English.

And it's not say that the keyboard goes away.  A lot of your interaction will be multi-modal, where you'll be typing and talking, and then when you have the small device, the pocket-sized device, then the speech input becomes far more important because the amount of keyboard input or even ink input you can do is a lot more limited.

MICHAEL D. FLEISHER:  So take me now to a world where we have speech.  Why do I need the Tablet?  Why do I need ink, right?  We've created at least two generations of folks who type better than they write.  So if I type better than I write, I've got speech if I want to annotate, why ink and why a focus from you guys so strong on the Tablet and ink?

BILL GATES:  Well, ink is super, super important.  Imagine a scenario today where you have a printout of sales results and you look at a number and you say, hey, that's too big or that's too small.  You really want to get people collaborating around that.  You'd like to point to that number and have it expanded by whatever, business group, geography, time period.  You'd like to go in, see some views, do some comparisons, and then you'd like to take that very rich view that you have, write some notes on it and say, "Hey, head of sales, what the heck is going on here in Japan?"  And ink is perfect for that, to let you write that note and then share that digitally.

Now, if you're sales results are on a piece of paper, obviously that doesn't work.  If it's on a Tablet screen device, where you can do those navigations and then share the ink and have that annotation just go with the view so that person -- it's not like today where when you send mail they wonder, "Well, what are you looking at, what number did they give you?" and you just waste a lot of time.  Here you can kind of co-browse, co-navigate either synchronously or asynchronously.

And it's very natural when people meet that, during a meeting, you don't want people taking voice annotations.  Like say you have a bunch of slides, you don't want people going, "And this, remember to tell someone else about this slide."  (Laughter.)  You get speech interference.  And a keyboard in a meeting isn't perfect because the screen tends to kind of block the view and you wonder what people are typing.

We have a lot of protocols for different types of meetings in terms of do the people sitting at the table, can they go off and navigate other information or can the people who are sitting behind go off and navigate other information. 

We've been using pen and paper for a long, long time and getting that skill set in, whether it's recognized ink or unrecognized ink, that's so natural.  This is like graphical user interface where as soon as you get the form factor right and the software right, it's just common sense.  I mean, I don't go to any conferences now and have people tell me, "Hey, what about graphics interface, wasn't that a stupid idea?"  There was a decade of my career where almost every conference like this would be, "Why have you bet your company on this fruit-loop graphics interface thing.  It's slow.  It's hard to program to.  Who needs 59 icons and 11 fonts?" 

Well, in fact, it was overused to some degree and there were things that had to be solved, but things like speech and ink are just so natural that when they get the right quality level, they'll be in everything.

And it's a very nonlinear thing because if the error rate is above a certain amount, only people who, say, have repetitive stress injury or can't have their hands available will be using it.  Then once you pass the magic threshold, say, for dictation, then it becomes something that everybody wants.

MICHAEL D. FLEISHER:  The challenge on the ink side for our audience here is that hardware isn't free yet, right, and so there's a form-factor change as well to the Tablet that has a hardware cost associated with it, as well as sort of a new learning, I think, for folks.  We really have taken information workers and made them typists in the last 20-plus years.

BILL GATES:  Well, when they go to meetings they don't do much typing.  And when you read an article about your industry and you want other people in your company to know about it, you really do want to circle things and write something out in the margin.

You're right, the hardware transition has to be trivial.  It has to be when you go through a refresh cycle on a portable machine, the machine that has the Tablet capability costs $100 extra and there's enough benefit that you see in that and there's no compromise in terms of compatibility, features, speed, anything, just this at most $100 where you say, okay, the idea that we can have annotations, note taking, let's go ahead and spend that small increment and it's all the learning that they've had before, same support infrastructure they've had before.  And that's what we're getting to.  Literally this year that will be about the situation.

MICHAEL D. FLEISHER:  So, we've talked a little bit about speech, we've talked a little bit about ink.  These are two areas that you guys have spent tremendous money on, right, on the research side.  And I want to actually try and understand from you -- I know the figure, right, it's $6.8 billion of R&D, a big number, up from $3 billion five years ago.  How much of that is research and how much of that is development?  How do you think about separating those two?  Because these are more pure research subject areas than they are development of product.

BILL GATES:  Yeah, the boundary between what's research and what's development in the software world is a fairly soft boundary.  Take some of the recent breakthroughs we've had in security where we can take code modules and literally prove that they're correct.  That's been a Holy Grail for, again just like speech, forever. 

And it was sort of a necessity to have that now because you've got really a scenario where you can make the portion of the OS that has to be secure very small but that portion is complex enough that you want machine proofs against it.  And that was some people in our research group who had these ideas, but it only worked for small programs, and it took some of the development people to figure out how to make it work literally for over a million lines of code.

We think in terms of things we can schedule and things we can't schedule.  Take the most extreme thing, our quantum computing group.  We can't schedule those people.  Whereas take the work we're doing on spam, or the security breakthroughs related to isolating systems so you can guarantee isolation, which is far more sophisticated than today's just-the-firewall notion, that's a pretty extreme extension of that.  Those things we have schedules.  I mean, we understand isolation, quarantine, IPsec, exactly how those are going to give those capabilities.

Even things like modeling, these visual-modeling approaches, where you can see programs and business process and edit those, those have now gotten into the realm where over the next -- our scheduling reaches out over about a three-year period -- we know exactly what we're going to do in, say, two rounds of major product releases for those things.

So for us it's we have all these things that are in incubation mode, and they have to have certain proof of concept in terms of does it work and do customers who see the prototype and incubation really get thrilled by it.  And then it gets into a process that's far more engineering, schedule discipline about, OK, which train will this advance fit into?

MICHAEL D. FLEISHER:  So I guess that leads me two places.  One is, so should I think about research as the place where Microsoft creates innovation, and development is the place where it creates product?

BILL GATES:  Well, development is the group that is scheduled to actually release things, so all of the complexity about compatibility and deep testing methodologies and things like that, those are in the product-development groups.

Research will put out some sample code that people will play with and the nice prototypes.  It's kind of nice but the real value in the IT business is when you reduce things, you integrate them in so they're not just a thing off on the side in terms of the user interface and programming model, they have total upwards compatibility and they've been through this testing thing.  And so that release process is in the product groups, which means the balance, 75 percent of the money is spent actually in the development realm and only 25 percent is pure research and incubation type activities.  Now, the two, we make sure there's no wall between those two things. 

MICHAEL D. FLEISHER:  So it's a seamless flow from innovation through to development through to product release.

So let's look back then.  If I look back and you think about the research you've done that's led to product development, what are the one or two things that you look backwards at and said, boy, those were places where we did great research and created great innovation?

BILL GATES:  Well, one of our biggest contributions was when we were tiny is the PC model, the idea that hardware is compatible and you get a software industry that's taking this high-volume, low-cost approach.  That was just sort of the founding principle of the company.

Then we got into graphics interface where Xerox, Apple, ourselves really bet on that.  That's also way before we had a research group.

Then we finally said, OK, linguistics, vision, speech, we can afford to do some of these things and started to build that up.

One of the incubations was a guy named Dave Cutler who did VMS, writing what came to be known as the NT kernel.  And so Windows NT, Windows 2000, Windows XP are all based on that.  We actually transitioned from our Win 9x code base to this far more secure, reliable, rich code base in a compatible way, and that was a huge effort that started as an incubation an then became a huge focus of the development group.

Tablet is an example that was incubation and then got committed into the product groups.

A lot of these modeling technologies that you're going to see in Visual Studio just in the next year, we were so ambitious in terms of how modeling could be used for workflow and IT administration and app customization that there are some pieces of modeling that are still in incubation, but 90 percent of the vision is actually now in a development mode.

MICHAEL D. FLEISHER:  Define what you mean by modeling.

BILL GATES:  OK.  The idea is that if you ask a businessperson what their business processes are and how do they do forecasting, how do they do sales analysis, how do they do HR review, you could capture in a very precise manner what those things are:  these people provide input, this person does this rating, if you don't accept the invoice you go through this procedure.  It should be easy to sit down at the PC and navigate those processes and when something new comes along, say you have a new partnership with somebody, to just -- without writing a line of code -- describe how that affects the various things that go on.

And this is true for describing how applications ought to run in the data center, it's true for how you do all these typical business processes.  There should just be a little visual model and that visual model should let you dive in and see detail, it should let you see the state, which stage of the process.  Some of these are very formal models like your monetary collection process or your project management process.  Some are very ad hoc, where somebody, let's say you're having a big conference for your customers, somebody wants to have people get together and have certain milestones and deadlines and they should just sit down and draw that out.  So it's expressing without code exactly what the steps are that need to take place.

MICHAEL D. FLEISHER:  But this sounds like enterprise applications?

BILL GATES:  Oh, absolutely.  The heart and soul of this is to take what has required large amounts of code and say that a business analyst can do these things.  And the beauty of it is it means if there's a new version of the application, because you've worked at this abstract level, you can take that new version in without a conflict between what you've done and what they've done.

The key breakthrough in coding is to write less code.  I mean, there's nothing magical that's ever going to make a million lines of code a pretty thing.  Corporations, governments need the platform to move up to be so high level that with these modeling tools the amount of code they're writing -- and let's take an ambitious target -- over a decade we should be able to reduce the amount of code the write by at least a factor of five.

MICHAEL D. FLEISHER:  So you're painting a version that completely undermines today's enterprise application providers, right, because you're painting a vision that says where today it's sort of this much software and this much services to get that enterprise app built, now I'm going to create a world that says it's this much software and a little bit of sort of user work.

BILL GATES:  No, I'm not undermining -- I mean, things like SAP have a very strong future in this vision, because their basic -- they are the engines underneath.  So when you're doing these visual diagrams, you're essentially customizing without writing code the big engine underneath.

But in terms of the idea that you're paying huge amounts of money for people to write glue code or you have to have an IT staff that's doing the hundredth reporting application, which is basically just presenting information in a certain way or intercepting certain pieces of your different applications and writing glue code between those, that work should be done visually.  The tools should make that so that you're expressing those things in a way that a non-programmer can understand.

MICHAEL D. FLEISHER:  So the enterprise app doesn't necessarily go away, but all of the services that it takes to make it work in today's model?

BILL GATES:  The services get refocused away from writing a bunch of glue code to whatever the true business value is, you know, designing what the resource allocation algorithm should be, designing how when you work with a new partner how the model should come into that.  And so it's a real shift in terms of that services piece.

I mean, the whole game here is to let people take high-volume, low-cost systems and use them for 90 percent of their needs. And so enterprise applications have been really complex, where you had to buy queuing software, pub-sub software, transactional software.  You had to learn all those different models, different manuals, different security things.  Getting it so that the platform, just the $50 operating system on the desktop, includes all those things, it lets you when you're going to write code be at a much more abstract level, but even that code, a high percentage of it should be done through models.

Now, we're taking the 10-year timeframe.  This does not happen overnight.  But this is what the IT industry owes to its customers.

MICHAEL D. FLEISHER:  I don't think you'd have any argument from the people in the audience that that's the case.

Let's go back to the $6.8 billion.  Where is that getting allocated today?  And so now let's start to shrink the timeline down a little bit from 10 years.  Let's get into sort of the realm of sort of our lives of the next three to five years.  Where is that getting allocated?  What are the big projects that are being worked on?  And where's the real innovation come from?

BILL GATES:  Well, one of the biggest areas of innovation is actually something, if we do it right, will be something that people just don't talk about, and that is the issues around security, reliability, spam, privacy.

Today there's a tiny bit of a dilemma for people where they say, yes, we should move these systems to be very digital, finally with the Web-service standards, the idea that the way you do business could be extremely digital is coming about, and the fact that we can connect to the workflow collaboration things that we're doing, can connect these back-end systems to the individuals.

But whenever you think, boy, the more I make this digital, the more some sort of Internet flooding or somebody guessing our passwords or some virus coming along would shut that down; it's almost like, hey, maybe we should keep doing things some old way because they're less subject to these attacks on digital infrastructure.

MICHAEL D. FLEISHER:  So the concern, because you've put a big emphasis on spam, right, you've been very public about it, so the concern is really at a corporate enterprise system issue as opposed to sort of I get a lot of junk e-mail?

BILL GATES:  Yeah, well, spam is awful in many ways.  It's a waste of people's time.  It overloads your mail.

MICHAEL D. FLEISHER:  It's small.

BILL GATES:  Oh no, it's a significant waste of time.  And people miss real e-mail. 

The most dangerous thing about spam, though, is you can send into any corporation today a piece of mail that appears to come from the IT department that says to people, "When you get messages that look like they're coming from the IT department ignore them.  This message says download this piece of software and if somebody tells you not to, that's a really bad person, so please ignore them." 

And so the fact that you can do this social engineering to fool people, like your password needs to be verified, type it here again, or download this thing, you shouldn't let communication come into the e-mail system that appears to be authentic.  It should only come in if it is authentic. 

And so the very design of the Internet protocol, SMTP, didn't have this verify-sender approach.  Now, it turns out that verify sender, what we call Caller ID for e-mail, is key for solving spam.  It's also key for making sure people don't do these social engineering attacks against your employee base.  And so everything that comes in has to be verified that it really is the person who sent that thing.

And so we don't have like a spam effort that's separate from our security effort.  The security effort, which is the biggest thing we're doing, the highest percentage of the $6.8 billion, are on things related to that.  That's an area that we have to solve and there are certain issues that are industry-wide issues, you know, preventing Internet flooding, getting rid of passwords, changing the mail protocols so that these systems verify each other. 

Then there are things that are Microsoft-specific, making it easy for our customers to know that they're properly isolated, making it easy for them to know that they're properly updating their software and making it so there's no manual effort involved in that at all.  That, which when we get it done it just enables all the other cool things to take place, that's actually the biggest category.

MICHAEL D. FLEISHER:  So security is a game of cat and mouse.  It's never going to completely go away.  But what is the timeline?  When does this audience sit back and say, boy, that's off my top-five list?

BILL GATES:  I think within the next two years it will get off the top-five list.  We have pioneering customers today who we're going down the learning curve with on, OK, let's do the isolation, was this easy for you, did it work; let's do the updating process, was this easy for you, did it work.  And so there's already best practices being established around this.

It's probably two years until all the issues around easy quarantine and everybody being educated and having the really great auditing tools are out there.  Isolation is a very key technique.  You can't assume, given the variety of systems out on the Internet, that every one of those would be reliable, and so you have to make sure that your corporate environment and even sub-partitions have degrees of isolation.  That is the magic thing that even if there are isolated outbreaks it prevents them from becoming these gigantic earthquake events like we've seen over the last several years.

MICHAEL D. FLEISHER:  And that's where a big chunk of the 6.8 is going?

BILL GATES:  A big chunk, yeah.  In terms of the development activity, you know, at any point in time I can always say to you what our number one priority is.  And one thing we're very good at is getting the IQ onto that No. 1 priority, whether it was the Internet in the late '90s, whether it was the Web-services protocols and XML in 2000, 2001.  These late few years and for these next several years these security related issues are the thing of greatest import.

MICHAEL D. FLEISHER:  And how does your customer see that come to them?

BILL GATES:  There's no lack of interest in this.  What they'll see is that as part of the platform the software updating capability, SMS with the current version of 2003, will be very prescriptive in terms of how they can use that, how they can make sure they've got that right.

The isolation things, that has to do with how they administer their firewalls, how they use IPsec, how they don't let these various port attacks come into their enterprise at all, and we'll give people visualization so they understand exactly what are we letting in and what are we not letting in and we'll let them see logs of those things.  So there will be very strong practices around isolation and updating.

MICHAEL D. FLEISHER:  On that front though, it hasn't been Microsoft's greatest strength to sort of help these folks from a services standpoint make stuff work, and isolation and some of those sound more like we're going to show you how to do this as opposed to we're going to provide you a product that does it.

BILL GATES:  Well, we're in the solutions business, and there's a certain learning curve as we get better and better at those things.  Office has always been a solution, but even there we could do better showing people how you use it effectively.  We're always stunned when people come and see how we use Office and they go, wow, oh that's how you do personnel reviews, that's how you do budgeting and sales analysis.  So, yes, we can do a lot better.

Now, the Web, in terms of having profession-specific things, videos, there's lots of ways to make sure people get the solution.  E-mail, we've always been a solution.  When somebody puts Exchange in, they expect us to work with them on how many servers, how do you set up the network, all the things that make e-mail a key solution.  And it's one of the most complicated things because it's global in nature, it's got to run all the time, and so you do get very involved in sometimes saying to the customer, look, your network doesn't have the right capacity to do this or you need to back that server up or you need to have a clustered server for those things.

And I totally agree with you, the progression of Microsoft is from, hey, here's great technology, you go figure out how to use it, to, here's your problem, it just happens that if you apply certain of these technologies the right way it solves the problem.  So that's a very important evolution.  We do that without giving up the magic technology.

MICHAEL D. FLEISHER:  The distinction versus others is it's showing folks how to use the technology as opposed to installing and using the technology for them?

BILL GATES:  Well, that's right.  We're not in the pay-us-$300-an-hour-and-we'll-show-you-where-your-brain-is-type business, or that we'll be your brain.  We want to enable organizations to be self-sufficient and use very high-volume, low-price tools, where in the past, they either wouldn't have gotten that productivity at all, and that's often the case, or they would have had to use something that is lower volume, higher price, more complex, and requires outside expense.

MICHAEL D. FLEISHER:  But it's a different skill set.  And we've talked about it here today, certainly we've talked about it before with Steve; there's a different skill set and a culture almost of helping the client make it work, as opposed to engineering the best product.

BILL GATES:  Well, from the very beginning the vision of the company was about empowerment, and we ourselves wanted something that would let us communicate and collaborate in a better way.  And so it's not science for science's sake.  These are things that every discussion about the next version of Office is in terms of the customers who come in, give us the feedback about those things. 

And one of the things that's phenomenal for us nowadays is we have through the Internet the ability to have a constant feedback loop.  So take Office today.  The latest version has the Help actually on our servers.  So if you're offline it will go to the local Help but if you're online it will go to the Help which is 100 times bigger that's up on our servers.  Not only is that more up to date, but we also see what you're accessing and you rate, if you're willing to, that experience.

And so every week we're going through the topics that are rated low and saying, OK, what do we do with the next version of the product, but, more importantly, how do we take that help and update it that next week.  And the things that we've found and seen as we're in this constant improvement mode are quite phenomenal.  And so the ability to be in touch with, OK, are people using this feature, why aren't they using this feature, what is confusing, what's frustrating, what error messages are people getting; basically there's a whole group at Microsoft that's devoted to the notion people should never get error messages.  And so they take all of the data about the error messages people are receiving and try and understand what is behind that and then drive those changes.

So the visibility of what's being used, how it's being used, how to improve it is something that's at a whole new level because of that Internet connection.

MICHAEL D. FLEISHER:  And as I know you know, there are some who are concerned about sort of that constant feedback loop, right, because there's a little bit of sort of show us everything that's not working right, and sort of trust us.

BILL GATES:  Well, certainly in every one of these products there's a little box you check, which is never call home, never send a bit to anyone.  And when the CIA buys our products they check that box.  And corporate customers, they can choose to do that or not.  We get enough people who are willing to have us see this profile that we get a very statistical sample.  I mean, 75 percent, if you get the thing that, God forbid, your machine has some crash or something, 75 percent  of the people do send that error report in, and about 75 percent, when we ask on Help, just a simple check, did this do what you wanted, 75 percent check that.  About 20 percent then will actually fill in some text to give us a little bit more of a model of their frustration.  And so that's the beauty of having a user base of, for Office, hundreds of millions of people is they want you to make it better.

MICHAEL D. FLEISHER:  In our sort of talk here about innovation and the future, two subjects got touched on that I want to follow back up.  One is rightfully noted that maybe the earliest innovation was the separation of the hardware and the software, right, that this was not a consumer electronics device, right, where they are embedded together but rather a common platform and then the software.

The business model of that for you all is that you make that software product better and better, add more and more, and therefore people buy updates.

I don't want to spend a lot of time on it, but the EU challenge to that business model, how do you see that and how is that affecting what you're doing in the R&D side?

BILL GATES:  Well, it is a very fascinating business model where we take the $50 operating system and we spend a high percentage of that $6.8 billion a year is going to put new features in it.  And when we put those new features in, the price is $50.  So you spend billions and billions and billions and it's way better, but it's the same price.  That's our equivalent of Moore's Law is that it just gets better.  Another equivalent is that if you want to keep using the old one, it costs you nothing. 

And so it's incumbent upon us to innovate enough that it's worth it; take Office, it's worth it to install and use that new version.  And it's fun to be put to that test.  It means we've got to make it easier to update and lots of things that come out of that.

The fact that Windows will get richer, one way or another, believe me, Windows will get richer.  People want more capability in Windows.  There are some legal issues about how we package that up, how we license it, how we engineer it, some models of which would slow down or be, I think, a real problem for customers, many models of which are very livable, even though they might create some complexity for us.

And in everything we've had along those lines we've always come out with solutions that work well, let us continue to innovate in Windows.  We had that here in the United States.  We solved that.  We're moving full speed ahead.  That's been a very good experience for everybody who was involved in that.  And at some point the same thing will happen in Europe.  It turns out that it's going to be several years of more process before we might get to that point.

MICHAEL D. FLEISHER:  You also touched on Web services, sort of mentioned it in passing.  Everybody has a slightly different notion of what this is.  So define for us sort of your notion of what Web services really are.

BILL GATES:  Web services are super, super important.  Essentially in the '90s, with the Internet, we got the ability to flow bits from any computer on the planet to any other computer on the planet.  And people sort of thought, well, wow, we can flow bits between all these things; doesn't that mean we're going to have electronic commerce and buying and selling and all the software will magically work together?  Well, the answer is no.  Unfortunately, just having the bits flow, then it just leaves the question of what the heck do these bits mean and where do I find the bits, how do I make sure they're accurate, how do you deal with it if software on the other end fails.

And so there's never been a protocol that lets software connect to another piece of software, find it, look at its capabilities and have really complex flows of information back and forth.

And the starting point for doing this was XML.  In the mid '90s Microsoft and some other people got very excited about taking that, which was really a document standard, and turning that into a data standard. 

And that has flourished.  XML is such a phenomenal technology and the tools around it, it's been amazing.  Everything we dreamed XML would become, it is becoming.  When you have nuts in your shop who say XML is magic, well, basically they're right.  This heterogeneous data, data mapping, data-viewing stuff is all coming out of that.

But XML is just a way of expressing the data; we still needed protocols for finding each other and exchanging it.  And amazingly, starting at about 2001, both IBM and Microsoft decided it was important to us to have these high-level protocols in a completely system-independent way, so that the mainframe could talk to anything, Windows could talk to anything.  And this was because we wanted customers to be able to develop this next generation of applications where data flows very easily.  E-commerce is just one example of that.

And we both put super-good people on it, and this year the spec for even the super high-level stuff, the security, transactions, reliable messaging, these specs are almost final.  We've had a lot of interoperability fast, we've drawn in lots of other companies.  Those things will go final this year.

And so Web services gives us the ability of any software component, written in any language, running on any operating system, to find and connect up to and exchange in a secure, reliable, transacted way information with another piece of software.

And so that will be built in to the server operating system, to the client operating system.  That will be the basis on which all of our admin stuff is done, our management stuff is done.  This architecture pervades everything.  If you go home and plug in a new printer and it gets located, underneath there's Web services.

MICHAEL D. FLEISHER:  So two things there.  One is the business model for you is not selling something new related to Web services; this is just your products getting richer and supporting this model?

BILL GATES:  Well, the protocols themselves are just out there and available, like TCP/IP, HTML and those things.  So the WS Security, WS Reliable Messaging, WS Transaction, that's all available.  And so we compete on having the best tools, the best implementation of those things.  And so there are new versions of our software like the new Visual Studio, the add-ons to the Windows Server that implement these capabilities, that we'll be going out to people, saying, hey, take advantage of this.

It's been a really good progression.  People today are using the simple stuff, the SOAP, WSDL stuff, and starting this year in a widespread way you'll get people using the more sophisticated stuff. 

And the impact of that is people aren't trying to write their own protocols, which is very hard to do, it's a lot of work, making those things secure is really hard.  You're just using this protocol sweep, which is very composable, that our runtime in every copy of the OS has that capability.

MICHAEL D. FLEISHER:  So what's the action item on Web services?

BILL GATES:  Well, the action item is that anybody who does any development or thinks about architecture application, you've got to look at what's happened with XML and say, OK, databases will be XML, Office will be XML, and what sort of schemas are important to your organization, are they being standardized horizontally or just in your industry group or does your organization have to do extended schemas around those things.

Then you think, wow, can we publish our capabilities as Web service, can we connect up to other Web services.  It's a whole mindset change that lets you compose things in a much richer way.

And all the new applications that get developed will be Web services and even existing applications you can put a layer on top of them that connect them into this Web service world.  It's not a rip and a replace kind of thing.  In fact, when we've gone into do government portals, we can take these huge, crazy applications that, say, do taxes or social security, and build a Web services layer on top so that then you can create a portal that gives the illusion that all these government applications really have a common architecture and work very well together.

MICHAEL D. FLEISHER:  Let's switch subjects.  You said before that the difference between research and development was development was when there's sort of a timeline associated with it.  So is "Longhorn" research?

BILL GATES:  No, no, it's development.  Now, there is a difference.  "Longhorn" is not a date-driven release.  We have things where we say the train is leaving on this date; whoever has their act totally together by that date, the train will leave.  And the train could have a lot of people on it or it could be fairly empty.

The "Longhorn" release, there are some fundamental breakthroughs in terms of how trivial it is to click and install a rich application, how trivial it is that there's no state on the machine, that if the machine breaks you can plug in a new PC and you're just up and running, you can go work on another PC and if you authenticate you're just there up and running.  There's a lot of things that have to do with IT complexity that these fundamental breakthroughs that are must have for "Longhorn."

And so, as long as you have must-haves, until you're getting really close to completion, your date will be somewhat variable.  And if you want customer and developer feedback that you're really willing to listen to, your date has to be somewhat variable.

So all we've said about "Longhorn" is that we'll have kind of alpha release out this year that everybody can look at and then as we get through that, we'll give a precise date.

MICHAEL D. FLEISHER:  So alpha in '04 and then a precise date to follow?

BILL GATES:  Yeah, and people are speculating that we're out in '06 sometime and that's probably valid speculation, but this is not a date-driven release.  People are amazed how far along we are.  When we had the developers' conference, there were a few people who said, "Hey, that looks good, why don't' you ship it tomorrow."  Well, in terms of getting all the pieces together so it's a clear IT breakthrough, a clear developer breakthrough and a clear end user breakthrough, we've got quite a bit more that we're doing.

MICHAEL D. FLEISHER:  So there isn't another Windows release before the "Longhorn" release?

BILL GATES:  Yes, there is a release called SP 2.  This is a great name, Windows XP SP 2.

MICHAEL D. FLEISHER:  Say that 10 times fast.

BILL GATES:  And this summer, that release is all about security.  It's a breakthrough in terms of it turns the firewall on automatically, it blocks certain types of Internet downloads and files things.  It's an XP update that has to do with security.  It's not about new features or new APIs, new user interface, any of those things.  It's purely security focused.

And so we will have a strong message that as you're buying new systems or if you're looking at updating, that SP 2 is the best client version to be using, but the really big breakthrough release, the next one is the "Longhorn" release.

MICHAEL D. FLEISHER:  Of that $6.8 billion, a huge portion of that is people in Redmond.  And we're spending lots of time with our clients talking about taking advantage of talent both from a quality, talent and price perspective in other parts of the world.  It strikes us that at Microsoft that isn't the model.  Why is that?

BILL GATES:  Well, we're partnered with companies in India, China, the Philippines, to make sure that their training skills and understanding of our platform is super high so that if customers are utilizing them they can do a super good job.  And so people like Infosys and Wipro, man, those people know our technologies, give us feedback on technologies.  It's quite amazing the quality of the people they have and the kinds of complex projects they're taking on.

For us, we're not about doing "Longhorn" 20 percent cheaper.  We're about doing "Longhorn" six months earlier or making sure that the fundamentals are exactly right.  And so our core development group, the bulk of it will always be in the one location where they're meeting every day, talking about the integration every day.

We do have development groups in India, we've got development groups in China.  One of the most incredible groups in Microsoft is our research group in China.  I mean, when we start a new research lab, we expect within four or five years they'll be contributing.  This lab had been going six months and in terms of graphics breakthroughs and protocol breakthroughs and a large number of things they were already making contributions.  So the energy and the talent pool there really blows the mind.

Research is easier to do on a distributed basis than development.  So we've got research in Europe, which is on the Cambridge University campus in the U.K.  We've got it at our headquarters and then we've got it in Beijing.  And that's worked out very well.  We use some methods, like this thing called TechFest where the researchers come in and show their latest advances and about 10,000 of the developers can go and look at those things, so we make sure that the communication works well.

But the bulk of Windows development and testing is going to stay in one location where we can do the very best job, even though that means our cost structure is high, and we overcome that with the volume model.

MICHAEL D. FLEISHER:  But ignore costs; you don't think that you can get quality and speed?

BILL GATES:  Well, the question is, how many locations do you want to do this thing in?  A lot of what we do is compatibility.  This is something -- someday when the free software guys get an installed base, they will have to think about binary compatibility.  That is they break it every new release they do.  It's a very hard thing and yet it's super critical if you have an installed base.

And having the understanding and the discussion about what's going on, what are the customers saying, how is this looking; having all that pooled in one location is very advantageous.

And so, yes, if I had been born in India, maybe the one location would be in India or maybe it would be in China.  It could be done there.  Now, you wouldn't be closest to the world's biggest market, most demanding market, but you could overcome that.  You'd have to have a liaison group and five people in to see those customers a lot.  It turns out we are in the world's biggest, most demanding market and I think that's some help to us.  But the key point is much  more are you going to fragment it across a lot of different locations.

And some of the other companies in the technology industry do things across multiple locations and, boy, have we seen the inner lab fights, the lack of common architecture between labs and things like that.  We have one common architecture, everybody working towards one way of doing modeling, one set of Web services protocols.  That uniformity allows us to make the advances as fast as we can.

MICHAEL D. FLEISHER:  Before we run out of time today, I do want to touch on and talk a little bit about some of the work you're doing in your foundation and on the philanthropy side, because you've clearly decided that you're going to make an impact in a particular place around world health.  So I guess spend a couple minutes and tell us why world health, where did that come from, what was the epiphany for that and a little bit of sort of how you see that world changing and the impact you're trying to make there.

BILL GATES:  Well, I see two areas where the world is improving very dramatically.  We talk a bit about how in IT tools it's going to improve dramatically.  And people are so underestimating these improvements in the devices and software and the way they get administered.

The other area is medicine, biology.  And the fact that we have basic systems understanding today, basic tools, those will be applied to the diseases of the rich worlds, cancer, heart disease, mental problems, and we will see very rapid progress that people are underestimating.  If you take a 20- or 30-year timeframe, there will be a huge number of breakthroughs.

In both this IT sector and this medical sector, the sad thing is that the benefits of these advances are not being experienced by the world at large, that it's so concentrated on the 20 percent of the people on the planet who are rich.  And so a theme of my philanthropy is, OK, how could you take these advances and make sure they're available to people.  Now, it's Microsoft's role to reach out to all those countries and governments of the world to make sure that in their schools, in their IT systems, that we're helping to get those IT advances to every country.  And so my philosophy in the foundation is to say, okay, what's the world's biggest inequity?  Well, it's health.  It's that 10 million children a year die who shouldn't die.  And that leads to all sorts of awful things.  It's not well understood that because you have poor health, families choose to have a lot of kids because they need to have enough kids to support them in their old age.  So, actually as you improve health, you reduce population growth, and therefore your ability to take the resources and feed and educate, get you on this positive cycle.  But solving health is always the first thing before you get that positive cycle, before you get decent government, and all the things that we take for granted.

And, so it was stunning to me to see that these new science advances, biological understandings, were not being applied to the most popular diseases, or the most common diseases.  Popular is not quite the right word.  And so you have a statistic that way less than 10 percent of the money in solving health problems is spent on the things that cause 90 percent of the problem.  And to be concrete, tuberculosis, malaria, HIV, and various childhood diseases, respiratory and diarrheal diseases get almost no money at all, and yet that's where you could save these lives.

So, the foundation has two parts, one is this health, which I think is the greatest inequity in the world; the other is, I decided to pick something that is more focused on giving back to the U.S., and that was education, being involved in helping out with that.  And so, at this stage, there's about $750 million a year going to world health, and about $400 million a year going to education, scholarship, library related things.

MICHAEL D. FLEISHER:  Here, too, you've picked places with very long bets.  So, how do you ‑‑ a) how do you parse where that money goes; and b) how do you decide whether you're successful or not, what's the measure of success?

BILL GATES:  This stuff is fun because it's a lot like software, where you pick really bright people, back approaches that have some risk, and you need to mix skill sets together.  It's not just people who can come up with the right candidates.  The whole thing about, how you get them tested, how you make sure they really work.  Delivering medicines is very complex in the developing world, you have to have approaches that fit in with that.  You've got to have relationships with the governments in those countries.  So, getting teams of people together and seeing how they're bright, how they can take on ambitious goals, it's a lot like my day job, my chief software architect job.  When I go over to the foundation, and we'll sit and talk about malaria for five or six hours, it's the same kind of thing about, okay, what are the early milestones, and how would we know, and who do we need to draw in?  Do we understand what the need is like?

Now, there's not the same economic reward at the end.  Here it's lives saved is the measure.  It's still pretty numeric.  When you say to somebody, hey, there's one person dying, let's go save their life.  People go, wow, let's go do that.  Then, when you say there's 10, they think, okay, show me their pictures, okay, let's go save them.  When you say, let's save a million people, it almost has less impact.

MICHAEL D. FLEISHER:  Harder to comprehend.

BILL GATES:  Yes.  And it feels impersonal.  It's not that direct.  The stuff we've already done, we've already saved way over a million lives, but we've just scratched the surface in terms of what's possible here.  And so it's betting on breakthroughs in medicine, bringing together all these people who have sort of ideas.  One example is, when you deliver vaccines, right now they have to be refrigerated, and so we wrote down these 14 what we call grand challenges, which are things that, if they were solved, would save millions of lives.  And one of them was being able to have vaccines not have to be refrigerated, doing something to them so that they're sort of sugarized, or something, so that you can carry them around and they don't denature.

And so now we have all these proposals, a lot from India and China, some from the United States, where we're funding people to go do that, and really thinking through, OK, is their approach going to succeed, would it work in the real world?  And so backing, almost like venture capital, backing some of these approaches, and these problems are so important, we have to back often multiple approaches that each might win.

MICHAEL D. FLEISHER:  To wrap us up here, and this has been a great discussion.  On both sides, give me the 10-year innovation, on the technology side, the Microsoft Research side, on the Foundation side, what is the one innovation that if we sit back down again and check in, that's the one we should check in on how we're doing?

BILL GATES:  Well, on the IT side, it's really the magic of software, and the ability to surprise people, whether it's administering these systems, or developing customized applications around these systems, or empowering their knowledge workers to see information early, to really know what's going on in the market, to collaborate at a distance within and outside the company.  The hardware platform is wedding the software industry with standards like Web Services, and advances in security, really letting us go and do magic things.  There's nothing holding us back.  And that's why we're at these record investment levels, even faced with, yes, free software can do some stuff, but not the really good stuff, and the high volume/low price model allows us to innovate and invest at those record levels.  So, I would say it would be the magic of software, and how that's been underestimated across all of the ways that it will be used.

In terms of world health, this notion of does the rich world ‑‑ is the rich world aware of how four billion of the six billion live?  If we were aware, we would want to help out, we'd want to get involved.  And so, it's creating the visibility, taking the latest science advances, which the U.S. spends almost $30 billion a year on medical research, which is a huge contribution to the world, very unique that goes on.  But then, letting that be mapped into things that in Africa and Asia really count, and having the world think of the U.S. as discharging its position as the most advanced country in a way that's beneficial very broadly.  I think some of these breakthroughs will help with that.  We'll see a positive momentum factor, and we'll be able to get many countries out of the trap they're in today, and get them on the cycle that countries like Korea, Taiwan, many went down in the last 20 or 30 years, and be able to think, boy, equity in the world has improved very, very dramatically.

MICHAEL D. FLEISHER:  Thank you very much.  Thanks for spending the time with us.  (Applause.)