Bill Gates's Keynote Speech at Microsoft Professional Developers Conference Beijing, 11 December 1997
Good morning. It's a great pleasure to be here. Today is a major milestone for Microsoft as our
first Professional Developers Conference here in China. The key partnerships we build with software developers around the world are only to the success of Windows but also to realize the possibility that PC technology provides. It's through applications of every variety that businesses will be using the personal computer as the tool of the Information Age.
It's rather amazing how fast this innovation is moving. Even to keep the like of myself who are deeply involved in the industry to go and see the improvement and every element that are taking place on a yearly basis is quite fantastic. Of course one of the driving factors of this business is the exponential increase in processor performance. There is no doubt that the magic of chip capability has delivered through the advance in microprocessor allows us to think of application which never would have been possible before.
The PC industry is one of the few industries that can deliver lower price equipment at the same time as improving the capabilities. The storage systems are now delivering Gigabyte of storage as the standard capability. Over 8 million of PCs are being sold a year. And the server market, the higher performance machines that these PCs networked with, are the fastest growing part of this business. The performance of those servers is increasing not only because the individual processors are faster, but also because we are using multiple-processor machines, so called SMP designs and clustering nodes together.
We are also improving the high end software on the server level. So a year ago, the fastest transaction benchmarks are using PC technology, that is about 6,000 transactions a minute. Today that is more than double to over 14,000 transactions a minute. And I can say with great confidence that next year we'll more than double that again.
It's fair to say that even the most demanding applications now can be handled with PC technology. At the same time as we have this power, we also need to improve the simplicity of the machines and to make interface easier to work with. I will talk today at the end of my presentation about some of the research that we are doing, in areas like handwriting, vision and linguistics. These will make the machines far more natural than they are today. These tools of the Information Age are for everyone, not just people with particular expertise in computer technology.
Microsoft has a vision for where the PCs are going. That vision says that PCs will be a central element of how companies share information. The name of that vision is the so called Digital Nervous System(DNA), allowing companies to reduce paper work and make better decisions. The DNA means that not only the PCs are connected together, not only you have standard elements like electronic mail and also if you really think carefully about what information is important - all the processes like order taking, sales planning, personnel management, project management - all of those have been set up to take full advantage of the capability of the computer.
By empowering every one in the company to have the information they need and allowing them to focus on what really counts, not only will customer service improve, but the ability to collaborate with partners around the world will be very straight forward. It's fair to say that the key platform for DNA is the combination of the Internet with the PC. The benefit of this kind of system is even clearer when you are dealing with surprises.
In this economy when there is all kinds of surprises whether it's new regulations or customers who want a product in a different way or something that the competitors are doing. These systems are particularly good at allowing one company to do better than other companies in dealing with these unplanned events. So truly saying that modern companies will not only have the technology but use it in a very deep fashion for competitive advantage.
The building blocks that allow companies to have the world's best DNA are very simple. In fact, most companies are investing in these things today. The up-to-date computers, the productivity tools, the email which I think will become very standard. Today's business card is not that common to have email address. I think in five to ten years here in China, most business people will have an electronic mail address. It won't just be simple messages that would be exchanged but all the things that would have done in paper forms today where you have to fill out information. Those will shift over into the electronic form.
Since every business has particular means, a lot of business applications are important to fill out the picture. These are not applications that Microsoft will be building. We and other companies build the vertical applications. On a global basis, that includes companies like SAP, Bond or Peoplesoft. I think there are important opportunities for lots of companies in this new area. In fact, one of our big priorities in China is to help companies build these new business applications. In some cases these are applications that in the past would have been built on Unix, that are now moving into Windows NT. In other cases, these are new applications that are to be build in the very beginning to take advantage of NT and its rich server capability.
I've mentioning that the Internet many times. Of course the Internet takes on many forms. The share of information using Internet inside an organization which is often refereed to as Intranet. There's the use of the Internet to allow different companies to work together and that eliminates distance as a problem. Companies no matter where they are located can do that. The Internet will really help promote global trade, not just trade in manufacturing goods, but also trade in services. A Company has engineering skills that it wants to make it available. It can find somebody who's interested in those skills across the Internet because the will be directories of customers and suppliers. Once they have been matched, the actual delivery of that service can use the Internet. Using of video and audio type of capability are becoming standard features built into the personal computers.
So DNA is a big vision - one that won't be realized overnight, but one that will be the direction of the industry that modern companies will go. People sometimes ask how far along has the personal computer come. Are we reaching the peak, now that we have over 18 million machines a year? The answer is that we are just in the very beginning. The machines in terms of their complexity and limitations are not fully penetrating the home and business market. But with this innovation I have no doubt that will change.
This slide shows some other technology and how quickly they caught on. We see radio, electricity, telephone and TV. We can see it in the past most technology have taken 40 to 80 years before most people are using it. My view is that the Internet and the PC are moving faster than that and so here we see some of the new things like VCR and TV. The final two things I've added are the Internet and PC. We have moving very quickly in just about 10 years see from the chart that a lot of people are not using the PC and the Internet. Even in the United States today, only 40% of the home have personal computer. There is no doubt in my mind that it will become as popular as any of the other technology. It will be based on the work that the industry is doing today.
Another major vision that Microsoft has is that writing the applications for these machines needs to get very easy and we could write programs that run across the entire Internet which is millions of machines. This is a new approach that programming draws on what was done previously. DNA says that developers should be able to focus on their particular task and not have to learn a lot on management of the machine resources.
Looking back to previous programming approaches, in the 50's and 60's,the mainframe - the central machine - was how things were done in a single machine. The focus was simple financial management - general ledger and billing materials processing. All the applications are around one machine. The exciting breakthrough was simply being able to use one language like COBOL to do the development.
As we move into the 70's and 80's,we started to have multiple machines and some set of batch processing and on-line transaction came along. We have terminals that were connected up and for the first time some intelligence in those terminals. Resource planning was the key application. Here we move up to not just use high level languages but also to use databases and graphical interface. We made it possible to split an application into two parts with the data on one machine and the interface as well as the application execution on the other machine. This is a two-tier approach - a client/server type of approach. It's a huge step forward and was very successful.
What does the future look like? What are we going to be doing in application from this point forward? We are going to have computers everywhere, not just in offices but in home, cars as you travel around. I must say it's an interesting challenge because those machines will be connected at some of the time at very high speed, some of the time at very low speed, and some of the time not connected at all. We need to allow for occasional connections.
Applications won't just work for a single server but they'll work for thousands of servers. In fact, we have to write them so that the application doesn't even know how many servers that it will be working with. The network needs to automatically take care of that. The key applications are electronic mail and bringing in a wide variety of form factors to let you connect up to the application.
The great breakthrough will be letting people create their business rules and not have to worry about the programming models. There is still a lot of work to be done here to make this application architecture easy to work with. It's not simply a case of taking what's good on the mainframe and putting it on cheap computers. Even mainframes with have all their nice capability aren't powerful enough to deal with this Internet environment- the number of transactions, the need for great interface... We are talking about inventing something new - inventing that by building on the strength of today's personal computer industry.
The architecture that Microsoft is providing to developers is what we called DNA, an architecture that includes all our development initiatives. It is based on the idea of different users in different locations with different types of execution and different kinds of databases. The key part here is cooperating component, letting different programs to work with each other. Standing out from simply thinking about databases to thinking about any program working with any other program, extending from a headquarters or a branch office to the customers of the company. If we use banks as an example, it moves from having systems that went batch to the headquarters to systems that ran out in branch offices. Now we are saying that any customer that is connected to the Internet should be able to connect to that bank, not only to see their information but do transaction. So this application architecture has to take advantage of personal computer and yet maintain compatibility and security.
What are the key abstractions that let this happened? Cooperating components are the key that draws very much on the object oriented techniques that have been explored over the last decade. But it goes beyond simple object orientation. That goes to the point of letting programs run on different computers without knowing that they move from computer to computer.
The application architecture let you separate things up not just into two pieces but into three... So that the data, the business logic and the presentation can all run very separately. What needs are these aim at? First, taking advantage of the Internet and all the technology around it. The phenomenon will simply continue to grow - communications cost coming down, more and more information is made available... We have achieved critical mass and so the pace of adoption will exhilarate in the years to come.
Companies want to extend their investment in their core application and we're talking about a new approach. An approach that lets you migrates from the current software and hardware without giving up compatibility. You can move one application at a time. Certainly the simplification of this computer environment is key. Today the computer exposes too much the details to users. Thing like registry, customers shouldn't have to think about that. Keeping software up-to-date is a lot of work today, we need to make that automatic. Some of the error messages that come of these machines are incredibly confusing and certainly Microsoft has been guilty in many cases. So we need to step back and say how do you use the power of the machine to hide the complexity and yet let these programs run anywhere.
Finally, customers want lower cost. When they think about cost, it's not just the hardware, it's the communications cost, the development cost, the data center cost and those costs that go with it. In fact, the hardware is the one that we've seen the most improvement. That means it's become a real priority to focus on all those other costs.
The great thing about that is that's a software problem. We should be able to do it very rapidly and there is no difficulty in delivering that while keeping the investment level very low. The environment that people are developing for today has separated between the Internet and the classic data processing environment, which will decide the client/server network. There are bridges between them but they are not seamless. The way that security and directory are done as well as the key product costs are different. We need to get the Internet much more embedded into what it's been the corporate client/server network. We need to do that by improving the tools and Windows itself to have all of that available.
The idea is integrating the Web and PC computing. What's good about PC computing? Graphical application, they are fantastic and they allow people a lot of rich interaction. They need to be here and improve on. Our choices are in hardware and software, Why has the PC industry outdistanced the rest of the computer industry? It's because of the choice that is provided. If we go forward, we need to make sure we retain those choices.
Finally powerful economy of scale. Eighty million systems a year means that companies are having incredible incentive to improve every element of the system, from storage to screen, graphic chips and certainly to the software application. What's good abort Web computing is that simple access of information. You don't have to think about deploying software. You don't have to think about the state of the machine. It feels like you go somewhere and the information appears to you Even though underneath, it may involve downloading software and updating things. To you, that is completely invisible. That is very exciting to data processing department because they don't have to spend so much time updating software and making sure that they run in the same way. That's the key that Web computing has done in the right way.
Also there is simple navigation being able to search very easily without knowing very confusing names that's something that the Web has but that the file systems of PC do not have. Let's take the best of both of these and that's what Windows DNA is about. What it leads to is a new class of application. Applications that are more flexible to change. Applications that will improve the access of information. Applications that are easy to administer and change. And it has built in scalability.
When I say scalability I don't just mean that they come on a more expensive machine but rather they can automatically take all the advantage of many machines. The scalability is not just limited to the most expensive piece of hardware out there. Let me be very concrete about our message to developers here. We are taking three different levels: storage, business processes and user interface as well as bringing new things to each of those.
For user interface, you have to think about code and pages. Those two should be unified. For business processes, the Web approach and the client approach should be very different, although we should bring them together. Finally in storage, things like file system or Web pages have been a completely separate world from the SQL environment and we need to bring the two together, again using the object oriented approaches.
When we do this, we want to support the mobile network and the connected network. We need to support distributed approaches as well as connected approaches. This is the framework we are creating here. To do this means updating the tools. One key philosophy we have is allowing people to use any computer language they want. That's very important because languages like C, COBOL or BASIC continue to improve. There are a lot of co-authors of these languages. We don't want to tell people to throw that away. We also want to support new languages and there will be many new languages like Java, the newest language and it's an incredible phenomenon. People are very excited about it. But it won't be the last improvement in computer languages and so we need to design the architecture so that as new improvements come along, they are all available to developers as well. Supporting any language is the center of this strategy.
When you write the code, I have been saying, developers don't have to worry about where it's going to execute. I will simply use high level specification to talk about what data or what components you want to interact with, locate it for you and determine where application should run. That requires a much richer platform than ever before. I will take a very specific concept, the concept of transaction. This is a very important concept. The ability to update information and making sure that a set of updates is all done together, so you don't have inconsistent information.
In the past, application has to do a lot of work to manage this transaction capability. That was not built into the operating system. SAP spends over 20% of their development budget doing things like transaction that indeed the platform should do with that. They are very interested to have us take over that so they can spend time on the business specific work that they really believe in. We need to let application talk to each other and transact as an automatic defined capability.
What we are showing here are the three-tier model. The database information up at the top. That can be many different systems, many different systems, many different stores. Finally the different presentation machines down at the bottom. That can be as simple as an Internet browser client with HTML or a full blown PC that's either connected or mobile. In the middle, we have the applications. That's the business logic that is specific to a particular industry or company's requirement. That's the part that developers should be able to spend all their time working on and thinking about without having to write a lot of rude codes and connect to the other systems. It's about distributed application. It's about one architecture and this can work with the existing computers and new computers. All of our tools and all of our work in Windows are aimed at delivering that as a Distributed interNet Application Architecture. That's what DNA stands for. We usually use this three-letter acronym because Distributed interNetArchitecture is quite mouthful.
I've mentioned several times that objects are key part of this. We need more than objects but the object approaches are necessary keys. There has to be a standard of how these objects talk to each other. For all Windows systems, desktop and server, that object standard is called COM. It ties together all of these pieces. I am showing here many of these pieces. The yellow pieces already exist today. The other pieces, the blue pieces, are being filled in as we complete this picture. They are blue, shown in purple here, is this way of objects talking together.
COM will not be the only way that objects talk to each other. But it will be the primary way because it's the only thing that is built into the popular Windows applications. We have gateways out to other approaches, approaches like Cobra and the other ones that are coming along. COM has already has an install base of hundreds and millions of systems and a lot in the industries is selling these COM components. We need to take that and continues to enhance it. We are talking about new enhancement which is called COM Plus that extends back.
I've mentioned that people are going to switch only to PC technology and higher and higher percentage of the new systems will be PC technology. But the bridges out to the existing world are critical to make that easy. For example, the transaction support that's built in the Windows is called MTS, Microsoft Transaction Server. It can work together with the mainframe transaction software, of which the most popular is the IBM CICS. I've already mentioned the bridge out to the object request broker, or ORBs, which is what Cobra is. I've also mentioned the support for standards like SQL are ODBC. This is a very rich integrated member of the computing environment and it makes it easier to write applications when you just take advantage of what's integrated.
What is the timeline for all of these? Today, we know NT. We are shipping version 4.0.In fact, we recently did an update with an option pack with many of the programming elements of NT4.0.There's an update of the Internet Information Server(IIS),moving up to its version 4.0.The Transaction Server has been updated. The Message Queue has been updated. The Distributed COM has been updated. All these fit in with the Internet Explorer version 4.0.The key pieces I am talking about are already here. People can build these Windows DNA applications today, integrating the client paradigms into application services.
The next big milestone will be the shipment of Windows NT5.0. We've already planned to offer the beta product in the first half of next year. We will have the final beta of that product shipped soon after that. It's a major advance bringing in a rich directory which provides better security and much lower cost of ownership. Manageability has been a key theme there and we are very interested to get that part done because it really is a breakthrough in managing systems.
I've mentioned ease in updating software. What NT5.0 does is it takes everything on your PC and it makes sure it keeps a complete copy of that upon server. So you got the speed up to these PCs because the information is almost stored as local. You also get the ability to go mobile because we store locally. Because we also store a copy on the server, the effort to operate your machine or go borrow someone else's machine and do everything you want requires no special activity because the information is up on the server. In fact, if your computer breaks and you get a new one connected it up, as soon as you log in, it will immediately be up and running because we bring down the information off the server. That is what we call intellimirror, the built-in feature to NT5.0.
The final phrase in delivering DNA is the enhancement to COM,COM Plus, and this is where we build in richer services, more connection and automatic component execution. Over the next two years all of these pieces will roll out in not only in Microsoft tools but also other people's tools as well.
One key point to make is that when we talk about sharing data and rich application like this, there needs to be standard. This standard can't just be about data formats that are common to every business. We need to focus on very specific business data type. These vertical formats for things like manufacturing, healthcare, insurance etc, we need object standard using COM for all of these industries. We have initiatives working with key developers to find interchange standards using COM.
I am showing some of these examples here that we've done. There are many more that are under development, e.g. financial services, CAD, process control etc. That's just to make sure when a customer has multiple applications, they can work together and share information.
So in driving the next generation, we have a lot of pieces other than doing this. We got to improve hardware. It's important not to underestimate that. It's very critical. We got the improvement for management. Some of which these come from NT5.0 which is a breakthrough.
Then we have the new interfaces, the ability to use handwriting and speech. I will make these things pervasive. I will talk about some of our researches because there have been some breakthroughs, not only by Microsoft, but many other companies. We can paint a more optimistic picture of when we won't just have to use the keyboard which has been the only choice in the past. Particularly for Chinese, that's very limiting to force everybody to input their information that way.
Other key drivers include the popularity of the Internet, the richness of those Internet standards, audio and video networks, Internet security. Microsoft is one of those companies that participate in doing all of that. A lot to be done in integration, making people not have need to buy different pieces and put them together.
Finally I think we are going to see a lot of innovation in different forms of machines-small machines in your pocket, tablet-sized machines, intelligent TV sets and all those that will make sure Windows connect up and is the platform for these applications. We have a major investment in China to work with developers much more than we have in the past. That will be critical tasks to drive those relationships.
Let's look a little further up into the future and talk about some of the research work that we're doing. We have relationships with a number of universities here in China. Our key location has been our headquarters, Redmond in Seattle. We just open up in Cambridge, England and more in San Francisco. We will be adding new locations and are getting a lot of encouragement here about what location in China would be the best place to come.
Research is critical to Software Company because today's software will never be the final software. Even in two or three years' time, people expect much better software, it's so limited compared to what we have right now. The price of the software is coming down even if it's been improving. Having that volume allows us to invest very heavily in research-over2.6 billion in the last year.
What are some of the key areas? Speech. I put that on the very top of the list. Vision, handwriting, graphics, news user interfaces, better ways in creating programs, new ways of testing program and scalability are just some of the central topics that we are studying.
Today's computer speech recognition is not nearly as good as human speech recognition. Human are amazing. Twenty years ago, people are optimistic. They will teach computer how to do this. They underestimated how good humans are in understanding speech. These are areas that humans are a lot better than machines no matter how much the machines have improved. Continuos dictation is hard but conversational speech is even harder because you don't have well-formed sentences to deal with. There's still a lot to be done.
It's ambiguous to what's been said unless we have some context .It almost requires common sense to understand what these sentences mean, what are they trying to do. A good example is doing searches on the Internet. You go to the Internet and say who sells the fastest chip. What results will you get? You get a lot of things like stocks, which is sometimes called blue chips, companies whose names include chips, and then you get some are actually accurate matches.
But how could you distinguish between these? You have to pause between these sentences and see what it's meant by that sentence. Then it will let go out and find the article on the Web that matches it. So here we see it's about selling chips and that should eliminate the case where the chips are the subject. When we apply linguistics to search in the Web, we go from having 20% of the results being meaningful to over 60%. That's a huge difference and we can do even better than that.
Let me talk about another area which is complete revision .I actually got a video that actually shows some of the progress we are making here. We are just at the beginning of this. But with a powerful chip ,I feel confident that's got to be a standard built in PC feature. So let's take a quick look at it-a little video that shows some of the computer vision work.
The key summary for the research is that interacting naturally with the PC has become a standard. By that I include handwriting, talking, listening, and finally the most difficult is learning. This means having a computer the same way that human systems work, what we are doing and making all these connect up to these future applications that I talked about today.
Great chips, systems developers, partners who are sponsoring this event, making this all possible. There is an incredible opportunity for developers. The applications that are written today will sell to an even larger base of machines out in the market. There is a lot that we're doing to increase the work of good developers-make sure they understand where the PC is going and how tools can help them now, more and more marketing type of activities making sure they got in with the customers. This in something that we are going to increase year after year.
The overall DNA message is one about helping developers seize that opportunity by bringing together the different architectures, making things automatic and allowing this to be done in an evolutionary fashion. I think it's a fantastic time to be developer and we appreciate being here and look forward to the opportunity to work with you more.