Steve Jobs on the NeXT Computer: A New Architecture

Good morning. I think I speak for everybody at NEXT saying it's great to be back. I haven't done this in a few years, so I'm a little nervous today. If you happen to see anything you like, feel free to let me know. Thanks for joining us today. I think together we're going to experience one of those times that occurs once or twice in a decade of computing, a time when a new architecture is rolled out that's really going to change the future of computing.

And we've worked on this for three years. It's turned out incredibly great. So I hope that you love it as much as we do. Now we're going to try to take a little bit longer than a usual product rollout today and not just tell you the what of what we've done, but the why behind it, some of the key decisions we had to make, some of the key risks we had to take, to give you a more full understanding of why this is such a revolutionary new architecture. So why don't we get into it now? Now, we started off with a model, and our model said that all computer architectures have about a 10-year life, and that means if they're really revolutionary, when they come out, they don't run existing software.

Now our computer will run most Unix software, but once you see what else it will do, you won't want to run existing Unix software. And so we'll have to build, just like every architecture, a base of application software that takes full advantage of the new capabilities of the machine. Now that starts to get going, the computer takes off, and around year five, it reaches its architectural peak, and then it goes into what you might call a glide slope. Now that doesn't mean its volume, its unit volume sales peak, but it does mean that the architecture is everything it's ever going to be at that point. And an interesting thing happens at the peak. Its greatest asset, which is all of those applications that it's taken five years to build, becomes its greatest liability, because it can't change technology.

It can't adopt new technology, principally software technology that comes along, or else it will break all of those applications and open up the buy decision. And almost universally, people say that's not a good idea, let's let new technology pass us by and ride down a very profitable glide slope. Now as we applied our model to history, it fit pretty well. It fit perfectly with the Apple II. And even though very successful new models were introduced after the peak, still the peak was in approximately 1982, although it's still selling really well after 11 years. Now next comes the IBM PC, and although the clone phenomenon has extended the unit sales peak, our model says that the IBM PC peaked in around 1986. And I think the PC was everything it was ever going to be in 1986.

Now OS II will probably constitute a new wave here, but as far as the MS-DOS PC, our model says it peaked in about 86. Then came the Macintosh. Now our model says two things about the Macintosh. The first one is, about a year ago, it crossed the PC in market momentum. Not unit sales, but market momentum. Now what does that mean? What it means is that all of the new, aggressive, hot software is on the Macintosh first. It means the momentum of the market is behind the Mac. But the model also says another thing. It says that Mac is going to peak next year. It says that you can already see cracks in the architectural foundation, which are going to keep it from being anything more than it will be sometime next year.

Now workstations have made a few ripples, and I think they've made some major technological contributions. Megapixel displays, multitasking, networking. But even taken collectively, they haven't sold enough units because they're not easy enough to use, and so they don't really rate a wave yet. But what we want to do is take some of that technology and make the fourth wave, going into the 90s. And I don't think we'll escape the laws of physics. I think we'll have to build an application software base, and I think we'll peak sometime in 1994 or 5. But I think we're going to make the first wave going into the 90s. Now, where do we find out what this ought to be? How do we define this fourth wave? What we did was collaborated with a group of people that are probably the most diverse and demanding of machine learning as it is known today.

For it is here, in higher education, where boundaries of knowledge are pushed, imagination stretched, and technologies tested. Academia is, and always will be, where the hardest questions are asked and the greatest answers found. And it is from this scholarly community that much of the world's computer invention has sprung. Yet despite the fact that computers are an integral part of academia, they have not yet been the catalyst for the educational change that their promise holds. We collaborated with the leaders in educational computing from research universities such as Stanford and Carnegie Mellon, liberal arts schools like Vassar and Reed, as well as state universities from California to Michigan to Maryland. We asked them not for their list of specifications, but for a list of their dreams.

Not to extend what computers have been, but to imagine what they could be. Only then did we begin to develop the next computer. Have we made progress towards our goal? We'll let one of our advisors answer in his own words. The next computer will become a university... desktop. Since knowledge is a social process, the computer is designed to facilitate communication with others. Since today's scholars build on the heritage of the past, it provides a storage medium which will allow us to recreate the research library on our desktops. Since we use multiple strategies to think and learn, it provides multimedia communications, sound, images, and print. And since each of us learns and creates in a unique way, it is a tool that we can recreate in our own image." We believe that to ignore the needs of higher education in America is to ignore the future. Today, the Next Computer will be unveiled to the world. And if the vision we share with higher education is true, the Next Computer will serve not only as a technological tool, but as a partner in thought.

Now, who exactly is higher education? Three thousand colleges and universities in the United States alone. Over 45,000 departments. Over 600,000 faculty members. And over 12 million students. Now, these institutions are very large. What do you think Stanford's annual operating budget is? Most people guess around $50 million, $100 million. It's $750 million. Michigan's is over a billion. These are Fortune 500 companies disguised by another name. These are Fortune 500 companies disguised by another name. These are Fortune 500 companies disguised by another name. And a lot of myth. Now, we couldn't listen to all 3,000 of them.

So we were very fortunate to have the talents and the energy of 24 of the leaders in educational computing who shared their visions, their insights with us, and they collaborated on us to build this computer. They had a diverse set of interests, but they had one thing in common, which was a deep commitment to realizing the promise of computers in education. They've known all our secrets right from the beginning. And much of what you're going to see today can be traced directly to their vision and tenacity. So what did they want? The first thing they wanted was the power of Unix. They said, we've got to have multitasking. True multitasking. We've got to have networking. We've got to have communications. But Unix is not usable by mere mortals. So you guys have got to marry Unix with ease of use. Okay.

Next thing, we'd like our computers to be fast. Five MIPS, fast floating point, and surprisingly, array processing. Give us lots of memory. Give us 8 megabytes of RAM. Give us at least 100 megabytes of local disk. We want a unified imaging model. Now, what that means is, is that we've got a standard for printing right now. It's called PostScript. But unfortunately, we don't have a standard for what's on displays. We've got QuickDraw. We've got what's ever inside Presentation Manager. We've got two or three other standards out there. And none of them speak to each other. And none of them speak directly to the printer without going through a translator. And higher education realized that we could have a great breakthrough if we unified this and put PostScript everywhere. So could you please put PostScript on the screen? And speaking of screens, if this electronic desktop is so good, would you please give us a big one? Give us one with a million pixels on it. We're tired of scrolling. Give us fast, transparent networking built in from the start, not an afterthought, please. Do for sound what you did for graphics. Give us fast, transparent graphics, great sound. And make it expandable into the 90s. We don't quite know yet everything these are going to be used for. We've got a bunch of creative faculty and students who

want to make things for this computer. Please make it an open architecture. Let's not lose what some of the early PCs had. Workstations are uniformly described as big, hot, and noisy. Would you please make them small, cool, and quiet? And the last thing they wanted was, we like printing, so would you please give us some affordable laser printing? Now, this is a pretty fierce list. So how do we go about turning this into reality? The first thing we did was we looked at PC architecture. We're pretty familiar with it. And PC architecture starts off with some RAM, and then we add the CPU and the video to it. And it works.

It works fine. This is how all the Macs are built, how most of the PCs are built. The problem is, when you try to put ten times as many dots on the screen, and you try to run the CPU ten times faster, it breaks down because the path to memory burns up. You can't get the information through the system fast enough to make it all work. So we thought, oh boy, what are we going to do? Let's look at workstations. How do they do it? What the workstations do is they put a cache on the CPU, and they use dual-ported memory called video RAM on the video, and they make it work. And it works pretty well. The problem is, it stops there. Because when you put a fast network on a workstation, you're burning up memory again. An Ethernet on a typical workstation chews up 50% of the system throughput, trying to get those packets from the Ethernet chip into memory. You put a fast disk on a workstation, and you're in the same predicament. The same thing is true of a fast bus or a fast backplane.

If you want high-quality audio, the data rate goes way up, and you're in the same predicament again. And if you want any sort of an attached processor, like an array processor, you end up with this. And what this told us, one of our key insights in developing the next computer is on this slide. What it told us was, was that MIPS is at most 25% of the performance equation. What everybody's been telling us is 100% of the equation isn't so. Because we could put 100 MIPS processor on there, and if it's waiting to get to memory because of the network and the disk and the audio, it's running at zero MIPS. So that the other 75% of the performance equation is overall system throughput. And our market doesn't just want to use one of these. They want to use all these things at the same time. So before we raced off and solved this problem in our own way, we're in our early 30s now, we asked, has anyone else solved this problem? And sure enough, they have. Now, this is going to sound very funny coming from us, but this is exactly, this is exactly what mainframe architecture has perfected over the last 20 years. Mainframes know how to do this stuff.

And what they do is they put IO processors on each of the IO channels, and then they rigorously control their memory system to squeeze every ounce of performance out of it by not letting those IO controllers in until they're going to do their thing and get the heck out. And it works. And what we realized was higher ed wants a personal mainframe. So at this point, we decided to take off our coats and get to work. And how does this look quantitatively? Quantitatively, we can break down the world into three categories right now. Fast PCs, workstations, and mainframes. Now, fast PCs are like your Mac 2s or Mac 2Xs, your 386 boxes, and they have a system throughput of about 10 megabytes per second.

Your Mac SEs are down at about 4 megabytes per second. So somewhere in that range, we gave them a 10. Workstations, you've got your Sun 386Is and your Sun 360s, about 18 or so. You've got your $35,000 workstations and your low 20s. So we gave them a 20. And mainframes, around a 40. Number of printed circuit boards. To get a fast PC hooked up to a display in a network now is three or more printed circuit boards. Workstations come in at about 3 to 10, and mainframes over 10. Number of chips. Fast PCs all in the 100 to 150 chip category. Workstations generally come in at 3 to 400, and of course mainframes are in the thousands. The cost, you're all familiar with. Now, when we look at this mainframe column, it looks pretty daunting. And rather than build something like this, we decided to build a VLSI group. Rather than integrate on fiberglass, we decided to integrate on silicon. And over the last two and a half years, we have put a mainframe architecture on two VLSI chips. You will see that.

Running today, their performance is staggering. They are totally CMOS, completely designed by NeXT, and they are very state of the art. Now, let's look at this quantitatively. 32 megabytes per second on a sustained basis, we are within striking distance of a mainframe. We are on one printed circuit board. We have 45 integrated circuits. That is half of a fast PC. That is a tenth of a typical workstation.

Now, to accomplish this, rather than blaze a trail to make our own CPU, we formed a partnership with Motorola when we started the company to make the next generation industry standard CPUs. We've been working together with Motorola for about three years now, and rather than build one processor into every computer, we've built three Motorola processors into every computer. We've got a 68030 central processing unit. We've got a floating point unit, and we've got a 10 MIPS digital signal processor built into every computer. All of these are running at 25 megahertz.

On our single board, we also support up to 16 megabytes of RAM using the megabit RAM chips. We actually think it will support the four megabit chips and put up to 64 megabytes on one board, but we can't get any to test it with. We have 12 IO processors, each with their own DMA channel, and that is one for the backplane, one for the serial ports, one for the printer, one for the digital signal processor, two for the audio, two for the disk, two for the network, one for the video. All the top two layers, all of the top two green layers are in our own custom silicon. All of these channels running simultaneously take less than half of the total system throughput of the system.

Let's focus for a minute on the audio and the DSP channels. We wanted to do something great in audio. So what we did was we built in some audio hardware. We have full CD quality stereo outputs built into every computer, that 16-bit, 44.1 kilohertz sample rate, exactly what you'd find in a $1,000 Sony CD player. We've got a microphone input built into every product. We digitize speech. And we've got the digital signal processor. Now why did we build a DSP into every system? You'll hear a lot more about it this morning. Turns out this thing can run two to three orders of magnitude faster than the other processors for certain types of calculations.

And the types of things that you can do with it are staggering. Speech. Carnegie Mellon is porting arguably some of the world's best speech recognition technology to the next computer. They're using the microphone input to digitize the speech. They're using the DSP to do the front-end processing. And they're using the O30 to do the back-end processing with zero additional hardware required. The best music in the world has been done at a place at Stanford called Karma. We were fortunate enough to hire the few best people from Karma a few years ago. And they've been working diligently. And you'll hear some of the most state-of-the-art electronic music synthesis in the world today, again, requiring zero additional hardware. Array processing.

You can do FFTs and convolutions with this thing 50 to 100 times faster than with the floating-point processor. And in the modem and fax area, 80% of a 9600-watt modem and fax is built into every next computer. You have to add software and a box off the back of the computer to attach it to the phone lines. And next year, you'll see these things appearing for very low cost. Now, why did we build all this stuff in? Because we learned from the PC industry that software developers have got to write for the lowest common denominator. They can't take advantage of features which aren't built into every single computer, or no one will buy their software.

No one's going to take advantage of a DSP in their software if, to use their $200 program, you have to go buy a $1,000 DSP board. So it won't get used. You will never see programs for the Macintosh that escape the orbit of working perfectly on that 9-inch electronic screen. So you are bound in by your lowest common denominator. This is our lowest common denominator. A one-foot square board containing the entire system. It's the most aggressive use of surface-mount technology in the industry. Let me take you through it. 16 megabytes of RAM up here, custom memory driver chips to get the highest performance out of the memory system.

Two mainframe-on-a-chips right here. The 68030, the floating-point processor, the DSP with its local memory. Our backplane. We've adopted the new bus protocols for the backplane, but we've made two changes. Number one, it's all CMOS. And number two, instead of running at 10 megahertz, we're running at a 25 megahertz. And we put the entire interface for the backplane in one custom chip. Everything. And we're going to be selling it to third parties for about $25.

Here's our video memory and a custom video chip that does all of the 100 megahertz video in one chip. We've got our connector to go to the monitor. We've got our Ethernet connector. We did all of our own Ethernet electronics. This is the first true 32-bit Ethernet ever put to market. Printer port. This is a SCSI port. It's MAC-compatible at the pinout level, but it goes to a chip that runs three times faster. So we're seeing burst transfer rates off our SCSI drives above 4 megabytes per second. MAC-compatible serial ports and a connector that goes to the serial ports of the DSP in case you want to hook something up to it that we haven't thought of yet. Thank you.

I hope you get a chance to look at this a little later. It's the most beautiful printed circuit board I've ever seen in my life. Now, there's one more fact about this board we want to share with you today. It's built completely untouched by human hands in NeXT's automated factory in Fremont, California. We realized that to achieve the quality levels we wanted to, the reliability, and the volume levels that we think the market is going to demand that we meet, that we had to automate the entire process. And I tried to capture that on a slide for you all today, but I couldn't. So we decided to make a film. It's the most advanced factory of its kind, building the most sophisticated computer of its kind.

Could we please run the film? ♪♪♪

It's about 20 minutes. fully build the next board. First, it goes to the screen printer. This robot stencils the solder paste that will hold the components onto the board. It's actually very similar to the process used for silk screening t-shirts, except it's about 1,000 times more accurate. Next, the board stops at the paste inspector, a robot designed and built at NeXT. It's a laser measuring system which does what no one had ever thought of doing before. It checks the height and skew of over 1,700 solder pads, making sure they're perfect to within 1 ten-thousandth of an inch before any parts are put down. It's given NeXT a defect rate ten times lower than any other surface-mounted board. Next, the board goes to the first of two pick-and-place robots. This one places the smallest transistors, capacitors, and integrated circuits on the board from locational dimensions programmed at NeXT. Those silver solder pads you're looking at are about one-hundredth of an inch higher.

This isn't only the fastest, it's also the most consistent and accurate robot of its kind, placing parts at the rate of 150 per minute. Next, the board travels to the second pick-and-place station. It places fine-pitched parts using a vision system that can accurately mount components within closer proximity than the width of a human hair. Next, the board is sent into a computer-controlled oven, which first dissolves any oxides that may be on the components. This leaves only pure tin and copper to ensure that the NeXT board has the best solder joints. Then the board is uniformly heated to 215 degrees centigrade, making a perfect metallurgical joint with the solder. When the NeXT board emerges from the oven, the surface-mount process is complete.

It's emerged with the highest density of components, combined with the highest production quality controls in the computer industry. Next, the board arrives at the through-hole insertion robot, which places, as its name implies, any through-hole components onto the board. It has two specialized arms. One chooses the part, the other inserts it into the board. Together, they can handle about 70 parts of all different shapes and sizes and perform a number of complex operations. Next, the board arrives at the wave solder station. Simply, it's Niagara Falls with 700 pounds of 255-degree centigrade molten solder. But before the through-hole components are soldered, the board is deoxidized with liquid flux to ensure high-quality bonds. Then it's over the falls. Next, the fully constructed board enters an environmentally sound cleaning

system that removes all flux and residue. The board emerges ready to power the most advanced, most sophisticated, most efficiently produced computer workstation of all time. Next. And we liked the monitor being separate a lot, but we didn't like the tower underneath the desk. And so we decided to do a modified tower where we've taken that tower with all of its expandability and power, and we've put the whole thing into a one-foot cube.

And this cube is pretty remarkable. It is a totally open architecture. We have room to plug in four boards of identical size. We ship with one, leaving three of the slots vacant for future expansion. We have a large power supply in the bottom that works anywhere in the world. No switches, no fuses. You find the power cord, it works. The unit is extremely easy to manufacture and to service. And we allowed room inside for two full-height, five-and-a-quarter-inch mass storage devices. Now, why did we do this? Why didn't we go to half height or three-and-a-half? The reason is because all of the advanced storage technology comes out in full height first, and two or three years later, it's available in one of the smaller sizes. So right now, we can fit 1.3 gigabytes in this one-foot cube, and that will go to three gigabytes next year.

And as you see with the software later on, to do really serious stuff, it comes in handy. Now, what does the user interact with? We've designed our own megapixel display from the ground up, all our own electronics, our own mechanical packaging. And this computer is going to be used a lot, so we decided to spend the extra energy and make a really great built-in stand that raises the display to the proper ergonomic level, and it also allows it to tilt to exactly the right angle to get the minimum glare and to get the minimum eye strain. In addition to that, on the back of the monitor, we've built in a speaker underneath.

We have Walkman stereo headphone jacks. We have gold-plated line-out jacks to go to the line-in of your stereo. We have the keyboard and mouse connector, and we have the microphone connector. So we've tried to put the connectors where the user is, which is just basic common-sense good design. And everything connects from the display to the cube with one three-meter cable, and this three-meter cable is pretty remarkable. It supplies not only 100 megahertz video, but also power to the monitor so that the monitor doesn't need its own power cord. You've only got one cable leaving the desk. In addition to that, since there's no moving parts inside the display, all the noise is three meters away.

In addition, it supplies all of the digital audio, takes back the keyboard and mouse data, and takes back the digital speech data. So we've got a complete system. Well, not quite. What are we missing? Well, not quite. What are we missing? We're missing a printing strategy. Now, hardly anybody seems to have a coherent printing strategy out there. And our marketplace told us that we better have one. So what might one be? Well, a typical thing to do would be to make a $7,000 laser printer and a $500 impact printer. The problem is that once you use laser printing, you can't go back. So no one wants the impact printer anymore.

What they want is an affordable laser printer. And not very many of our market could afford $7,000. So we worked very hard, and we came up with our own printing strategy. This is it. It's the next laser printer. And it's got four really great things about it. One, it's about 60% of the size of existing laser printers. Two, it is full PostScript output. No compromise. The third one is, it is twice the resolution of any laser printer on the market. It gives you 160,000 dots per square inch versus 90,000.

You can really see it on text, and you can really see it on graphics and images. And it's switchable between 300 and 400, so you can pick which one you want. 300 is kind of draft mode. And the second thing is, as you'll hear later on, we've achieved this at a real breakthrough price. We're going to be far under 50% of that $7,000 laser printer. So now we've got a complete set. It's been designed from the start to be a complete system. And you've listened to me talk about it for about 40 minutes now. What I'd like to do is have the first computer of the 90s show you itself the kind of thing it can do.

Thank you. It's running. It ain't running. Come on, come on, come on. Come on, damn it. Okay, good. Pause. Okay, ten seconds.

Turn the mic off. Pause. Pause. Pause.

Pause. Pause. Pause. Pause. Pause. Pause.

Pause. Pause. Pause. Pause. Pause. Pause. Pause. Pause. Pause. Pause. Pause. So now you know. This is what we've been up to for the last three years, and we're going to have a chance to get into the software quite a bit, which is even more astounding than the hardware, after our intermission.

But before we go, there's one last thing about the next hardware that I wanted to brief you on, and you'll be able to see a little bit later on today. We haven't talked about mass storage. We thought, what should we have for mass storage? And we looked at floppies. The problem is, floppies are pretty small, and every time we want to update our system, it seemed crazy to us to ship 50, 75 floppy disks. And the more we looked at it, the more we realized that floppies were a technology of the 70s. So then we looked at small Winchester disk drives. And Winchester disk drives are pretty small, the 20s, the 40-megabyte versions.

In addition to that, they're not particularly portable. And we decided not to use small Winchesters. We then looked at new technologies, emerging technologies. Tape, digital audio tape, looked pretty interesting. We looked at CD-ROMs. We looked at read-write worm optical media. And none of it really measured up. And so two years ago, we made a decision. We saw some new technology, and we made a decision to risk our company that we could pull it off and bring it to market about four years before anybody thought possible. And we've done it. The next computer is the first computer in the world to ship with read-write erasable optical technology storage.

Every single computer comes with 250 megabytes of mass storage, removable. And these cartridges cost 50 bucks apiece. So as we move into the 1990s, we really can take our entire world in our backpack, pop it out of the computer, run across campus, run across town, run across the country, and pop it into another computer and have everything we've probably ever done in our lives with us.

We're going to take a prompt 20-minute intermission. We've got some machines set up in the lobby. We've got some disks. We've got some PC boards. We'd like you to see firsthand what we're doing, and we'd like to be back in here promptly in 20 minutes. Thank you very much. Thank you very much. Thank you very much. Thank you very much.

Well, what did you think? We wanted you to see the computer up close for two reasons. One was so you could see the beautiful, ultra-sharp megapixel display, which even this light valve that we used can't quite do justice to. And secondly, so you could see the beautiful industrial design, which is done by a friend of mine named Hartmut Esslinger. And Hartmut did this stuff personally. I think it's some of the best stuff he's ever done. And I had a story I thought I'd share with you.

When we first started the company, we weren't working with Hartmut, and we had retained this firm in Europe to do our packaging. And we knew then that we wanted to put the computer in a one-foot cube. And so we told them this, and they said, great. And about two months later, they called us and said, you've got to get on the next plane to Europe and see what we've come up with. We've done the obligatory cube, but we have something far better that's going to change the shape. of computing forever. So myself and Bud Tribble and the rest of us got on a plane, flew over to Europe, and drove out to their studio. And they gave us sort of a lackluster presentation on their cube design, which was about a tenth as good as what we ended up with. And then they had the major unveiling where they were going to pull off the curtain. And they gave this design a really fantastic buildup. And it really was, every computer was going to look like this in five years.

And they pulled off the curtain, and it was in the shape of a human head. From the neck up. And the boards plugged in the back of the skull. So Hartmut rescued us. And I think he's the best product designer in the world. And I'd just like to say thank you, Hartmut. I know you're out there today. Now, when we started our software effort, we had some big challenges. We had to get an operating system together. And Unix is a lot more powerful than most of the other operating systems, but it's also a lot more complex. So getting a Unix together is a fairly large challenge.

But then our market told us we had to make this usable by mere mortals. So we had to break ground in the user interface to go further than a graphical user interface has ever gone before, to embrace multitasking, to embrace transparent networking. But it turns out that that wasn't even our number one software challenge. Our number one software challenge, if you zoomed out of it and looked from a distance, was that Macintosh was a revolution in making it easier for the end user. But the software developer paid the price of that revolution. It is a bear to develop software for these things. And the newer systems that are coming out are even harder than the Macintosh. So we decided that this was our number one software challenge. Because our market doesn't want to take six months to write a program.

They want a statistics program for next Thursday's statistics class. So what do we do? First thing was we adopted this as our number one software challenge. Now, if you look before 1984, an application was 100% core. If you will. Very simple. All of a sudden in 1984, something happened called Macintosh, which reduced the core part to 50%. And this user interface gobbledygook invaded. And half the code of a modern application is this user interface stuff. And this is hard stuff for most people to do. But it gets worse than that. Because if you look at the time it takes to make this application, we hear from everyone, from the most professional developers to university developers, that the user interface takes 90% of the time.

I know there's a lot of software developers out there. Is this correct? And so we decided that we had to do something about that 90%. And we did. What we've done is, starting with that 90% of the time for the user interface, we have invented some technology which, even for very complex apps, reduces this to 10% of the time. Now, how do we do that? This is what you need to know to program a Mac. It's about 400 subroutine calls. And you can layer stuff on the top. But you end up having to know this stuff before the day is done. And we have replaced this entire way of doing things with a thing called our application kit.

It is totally object-oriented. And there's about 18, 20 objects that the programmer uses. And they do all of the complex event loop programming inside themselves. So all you have to do is connect them up, if you will. As an example, you can take a text object, a rectangle of any length of text, any font, any size, any style, full cutting and pasting, done for you. Drop it into a scrollable view object, inherit full scrolling behavior. Drop that into a window object, inherit full window behavior, resizing, go away buttons, et cetera, in a multitasking operating system in this much programming. And because it's object-oriented, you can subclass any of these objects and change 10% of them without knowing about the other 90. You can add objects to this and build your own libraries.

But we've gone one step further. We have written a revolutionary piece of software called Interface Builder, which reduces it to zero. And Interface Builder allows you to graphically lay out and connect all of the application kit objects without any programming whatsoever. And I could not make a slide for this, so I will personally show it to you in a few minutes. Now, our software architecture, then. We start with our hardware. We add our operating system. We are using a version of Unix called Mock that has been developed at Carnegie Mellon University under a grant from DoD.

It is fully compatible with Berkeley 4.3 Unix, but its underlying technology is not We've added to that full NFS file system support. Now, what are we going to do about graphics? As we said earlier, we wanted to adopt PostScript. A lot of people said it could never be done, that PostScript was way too slow to drive a screen. So we formed a partnership with Adobe, and it's been a great partnership. We have worked with a number of companies to develop PostScript for a number of years. And it's been a great partnership. We have worked for greater than two years with Adobe to define the extensions to PostScript to make it interactive on the screen and to make it an order of magnitude faster.

And everything you're going to see today goes through PostScript. It is full PostScript support on the screen, the exact same PostScript that drives the printer. And so we can add a layer here that is full PostScript support. Now, we had to do a window system. We wanted our window system to be compatible with PostScript, so we did our own. Here's the application kit, interface builder, and a program called the Workspace, which manages your electronic workspace for you, helps you navigate through the file system, and then, of course, applications. This is our software architecture. Now, the four middle layers are really key.

The four middle layers constitute what is our development platform moving into the 1990s. And because of that, we've given it a name. It's called Next Step. And Next Step is what accomplishes that number one software challenge, as you'll see in a little bit. And it provides software developers, I think, the breakthrough platform that we've been looking for for the 1990s. And so programmers, instead of concentrating on these layers, can start at the 10th floor and concentrate on their applications. Now, what I'd like to do now is do some live demos for you. We've got a machine up on stage here. For those of you that aren't software developers in the audience, I'd like to remind you of the first two laws of demoing.

First law of demos is that demos will always crash. And the second law of demos is that their probability of crashing goes up with the number of people watching. So if something goes wrong today, have some compassion for the demoer. OK. Now, what do you see when you sit down to a Next machine? First thing you see is this. This is what happens when the machine is booting off the disk, off the optical disk. Now, of course, we can hook up to a network and we can boot off a network as well. So if you're booting off a network, you're going to see something a little more like this.

Now, once the system's finished booting, you're going to see something like this. And it says, why don't you log in for me? So I go by the name sjobs. And if I manage to get my password wrong, it'll tell me. But if I get it right, it'll disappear and drop me into what we call the workspace. And this is the working environment that you will see when you first sit down to use the Next computer. And there are three elements. First one is the menus here. The next one is the browser. And the third is the icon dock over here. Now, let me take you through these things one at a time. First thing I'd like to do is take you through the browser.

What is a browser? Well, we believe in icons. So we have icons. And you can show your programs as icons. The problem is, even with our large megapixel display, sometimes there just isn't enough room. And you find yourself scrolling and scrolling and scrolling to find that icon you want. And in a system as sophisticated as Unix, sometimes you think there's got to be a better way. So in addition to having icons, we have this thing called the browser, which lets us say, look at my disk, and lets us look at demos, and lets us look at all of our demos and scroll through and select things with text. And as we select them, their icons will change.

I can also grow the browser. If I want and trade off screen space for scrolling Make it as large as I want so I can traverse my directory with a click and See everything underneath Very very simply so I like more screen space and I don't mind scrolling a little bit. So we'll take our browser down like this Now the next thing I want to show you is menus We thought about pull-down menus. The problem is they don't work on a large display. You're constantly going to the top of the screen So then we thought about pop-ups like the workstations The problem is users want to see their top-level menus. And so we came up with something that is really great These are our menus. We can put them anywhere on the screen. Nothing can ever cover them up we have our sub menus right here that we can click through and As you see we can just pull them up like this if we want to use them. We can just tear them off We want to leave them around somewhere So that we allow the user to trade off making use of this great Megapixel display and use it for the things that they use most frequently if we don't like them We just send them home with a click of a switch here Okay, that's the menu system the dock. It turns out that When you're running applications things can get lost These icons can get hidden and you might want to read your mail at a moment's notice So we allow you to take any icon and take it over to any one of these dock positions And it'll snap in and dock and the minute it docks. Nothing can go in front of it And so it's a place to always have the applications that you use handy. You can customize it any way you want to and

Nothing will ever keep these things from being a glance away That's what the docks all about and if you decide That you need to use that right part of the screen for an awfully big window and you don't want to undock things you can Just slide it down and everything but the little logo will go off the screen Now what happens when you throw things away What do we do? Well, let's say you have a file that you don't want to keep around anymore. You can just drag it over here We don't have a trash can we thought we'd get a little more sophisticated about it. We have a black hole and When you drag something into the black hole, it just decides to take it into hyperspace No trash cans here Okay, next thing I'd like to do is Show you a program that was written in about an hour. It'll just demonstrate a very simple facet of the machine It's called stopwatch and you can run it and the neatest thing about stopwatch is that When you move the window if you notice we move the whole window not just an outline We also let you put windows off screen which makes your virtual screen area quite huge And of course, it's not giving any cycles to the second hand of the stopwatch as I move the window But it smoothly catches up after I release it and the person that wrote this rather than just writing code wrote an animation object Which does all the complex mathematics of that smooth catching up and put it in the library and now everyone that uses

Animation uses the animation object and we're seeing from this object oriented programming a tremendous amount of reusable code But one of the neatest things about stopwatch is I can bring up more than one copy because this is a true Multitasking system and why don't I go ahead and bring up three or four of these things? Here we go, we'll bring up a fourth one here Alrighty Let's move this off screen and here we go No problem These are four completely different programs running here, okay Let's get rid of the stopwatches And go back to our workspace now, I want to show you something that is quite revolutionary and that is interface builder and It's so revolutionary that even I am going to build a program with it today and we're going to start by building a program that Is very doing something it's very simple and then we're going to go do something that's a little more complicated so the first thing I'm going to do is Open a new window and Interface builder titles it and gives me a window to use and gives me a menu to make a little bit later an

Interface builder lets me go up and grab the objects from the application kit So let me go up and grab some radio buttons Just tear them off and place them in the window And if I don't want them, they'll just fly back and I can take the radio buttons And of course I can grow them I can have as many radio buttons as I want to simply by growing them like this and If I like radio buttons, this is perfect They'll all adjust in in size based on resizing for titles and I can also go up and take a field object Stretch it out And I can have as many fields as I want to so they'll just grow database program something like that I can stretch them out if I want to put a little more space in between them or I can tighten them up Which is kind of nice if you want to Do something where you're just going to enter a few numbers like this. So this kind of stuff is very very easy to do So, let's go build a real application now to build a real application I'm gonna do the user interface, but I'm gonna use an object that somebody created Maybe a grad student maybe a professor at another University and what they've done is they put that in a physics palette They've made their own palette as one can do for interface builder So here's the physics palette that they've made and it's got a few objects on it Let me go ahead and make my window a little bit bigger here And the first thing I'm going to do is grab the main object Which is an ideal cylinder with a gas molecule in it And what we're going to do is we're going to see how that gas molecule behaves. We're gonna make a simple simulation

So let's go ahead and do that. Now. Let's make our gas molecule a little bit bigger here See that okay, and because of Postscript we can resize all these things and Let's go ahead and say we want to stop and start this thing. So let's go grab a button Let's make the button a little bit bigger here and let's go change the font and We'll go ahead and Helvetica 12. We'll go ahead and make it nice and big so everybody can read it Helvetica 24 point and So now we've got a button and let's call that start And let's make some more buttons make this one a little bit bigger Let's just make some more buttons like this. We just drag out and we have more buttons Make as many as we want and let's make these a little bit further apart And let's relabel a few of them let's say stop and reset So here we've just laid out something fairly standard fairly simple Make our molecule a little bit bigger Okay, now, how do I connect these buttons up to the molecule? Well, what I do is I go get a connection panel, of course And here's a connection panel and what this connection panel allows me to do Is it allows me to say okay?

Let's take my molecule object and let's take it up and put it as the target And now it's showing me all of the things that the molecule object is showing me So let's take my molecule object and let's take it up and put it as the target And now it's showing me all of the things that the molecule object understands Over here. It understands reset gravity mass sound whatever the programmer put into it So that's fine. Let's take my button my start button and say I want to connect my start button up To say start to the molecule object and I want to connect my stop button To stop and I want to connect my reset button Up to reset and now what I can do is that these are connected I can flip my big switch here to test And let's go ahead and test it you think it'll work there we go So Now what you're seeing is really interesting you're seeing that the oscillation of the top piston of the cylinder And as you see as the cylinder gets lower the gas molecule hits it more frequently driving it up As it gets higher the gas molecule hits it less frequently bringing it down And it's pretty neat. I can stop it. I can reset it and I can start it So I kind of like this So let's do a little bit more. Let's go back and build some more Okay, let's go back to our palette well, we've got some fun things called sliders let's grab a slider over here And let's make it a little bit bigger maybe

And let's label it. Let's call it, uh temperature So we put our label just fine and let's take this now And move our slider up to our connection panel. And why don't we say, set temperature. And go back to test. And see if it really works. We start. Looks just like before. Let's go ahead and turn up the temperature and see what happens. As we turn up the temperature, the molecule is moving faster and therefore hitting the top more frequently. As we move the temperature down, the molecule is moving slower and hitting the top less frequently.

So we actually see the effect of temperature on our gas molecule and our cylinder. Okay, well, we've kind of got a little taste of it now. Let's go do some more. Let's say we look inside our molecule and we see that it understands a few other things. It understands gravity and mass. So we've kind of got the bug here. Let's go make some more titles. And put mass and gravity. And we might want to stretch those out a little bit. A little more room here. And let's make some more sliders. And let's stretch them out a little bit. All right. And let's go ahead and connect this up to set mass.

And connect this up to set gravity. Can you see all this out there okay? Great. Let's go back to test and see how we work. Okay, here we are. We can turn up the temperature. And we'll notice that the top of the cylinder isn't coming down quite as far. But let's make that thing a little heavier. Oh, boy. And by playing with the gravity and the mass and the temperature, we can see the full effect now of this on the molecule. That's pretty cool. Let's do one last thing. Let's go back to build. And let's grab this other object off the physics palette. This is an object that we have in our library.

It's a strip chart recorder. And we're going to go ahead and record what the path of this thing really is over time. Let's make it a little bit bigger. And let's go ahead and connect it to the output and connect it up to position so that we can get the position. And while we're at it, let's go grab another button from our panel back here. Let's try a switch maybe. And let's label it sound. And we might want to make it a little bit bigger so that everybody can read it. And why don't we go ahead and make it 24 point. Great. So we've got a sound button. And let's go ahead and connect that up to set sound.

Okay. Let's go test it one last time. And now when we say start, we can actually see the path of the piston being plotted over time in our grapher object. You see that okay out there? Faintly. The light valve doesn't quite show it up. But one thing we will be able to do is. Okay.

So I think you start to see the power of the application kit and interface builder for building user interfaces. And what I'd like to do now is I'd like to introduce somebody to you named Dr. Richard Crandall. Dr. Richard Crandall is a Howard Vollum professor of physics at Reed College up in Portland, Oregon. And we have had the privilege of having Richard also be a NEXT education fellow for the last nine months. Where he's come to NEXT and helped guide us to make sure we're really on track in terms of making this computer everything that higher education wants. And Richard has written some demos which I'd like to have him show you now using all the tools in our kit. And he's written them in anywhere from hours to a day or two. Things that would normally take months in a current environment to write.

So I'd like to introduce him to you now. Dr. Richard Crandall. I'm proud to be able to stand before you and show you with my little examples how really powerful this system is for science and education. Two things I think that are very important for this country. I'm going to start with a an elementary statistics laboratory. You can see on the right hand side of the dock that there is an icon for it.

What this program does, my friends, is it ingests data files for breakfast, lunch and dinner. And it takes virtually any set of data file, any data file set. And it counts the number of columns and does other intelligent things. When I open up a file. It processes it preprocesses. It basically does all the elementary statistics you need to do in school, like to take the mean variant standard deviation of temperature values for chemistry laboratory, something like that. This particular file turns out to be 10 years old. This file is part of a model for the Earth's magnetic field and how it wanders.

This was actually taken from a college mainframe. And of course, it loads right into this application. So that represents a 10 year forward compatibility. Now, on the upper left, you see these beautiful radio buttons that are part of the application kit. It is indicating that it is ready to process column zero or column one on either the X or Y axis. What I'm going to do first is plot the data on channel one. Well, that was fast. Hundreds of points. Let's display PostScript for you. I have the option of inverting. The option of putting up a grid and so on.

The other column, column zero looks like so almost the same thing, but not quite. If you take the Fourier spectrum and the next computer is particularly good at this, you can even use the DSP for this. It's frighteningly fast and does long FFTs in milliseconds. Anyway, the spectrum is characteristic of a chaotic process one learns in modern times. This is still not understood, but this is a good example of how to learn chaos. This actually is called a strange attractor, a beautiful name for a physics object. And a strange attractor in this case is two coupled differential equations, if that means anything. And you get these two signals and only when you plot them against each other.

So I select button zero for X and button one for Y and I do a scatter. I get this strange attractor and there's there's no way you can see this. Really, if you imagine two columns of data on your old mainframe. Now, this is this is supposed to be the Earth's magnetic field. So can you see the mouse cursor? Yes. Oh, say it spends. It's up north now and it didn't always be there. So the magnetic field wanders around the North Pole for maybe one hundred thousand years. And then it spends the next hundred thousand years down here in Palo Alto.

If I want to know the latitude and longitude of Palo Alto, I move my cursors in. You could say, let me move the cursors down near my little text fields, which, as Mr. Jobs said, are objects that are part of the application kit and they display numbers readily. You basically write one line, one line of code to get the to get the thing to display its floating point number. Extremely convenient. My second example is called Molecule. In this example, I really tried to make a physics object which had some educational bearing. Let me open a familiar molecule that that of water or H2O.

I can magnify it. And wouldn't it be nice to be able to see around it? In physics, we have these things called rotation matrices and they take three parameters called Euler angles. And of course, the natural symmetry is to invoke from the application kit these three sliders for the Euler angles. You see, there's one. There's another one. You see, with display postscript, you can use transparency. And you can see that little hydrogen guy behind that oxygen guy. Or or here you can see it in front. And yet another angle.

OK, now, you know, in the English language, there are nouns, verbs and adjectives, and those form the clue part of of dialogue. Sure enough. In Objective C, which we use when programming this computer, a noun becomes something I mean, an object in Objective C is something like a noun. What is called a method is something like a verb. And what is called an instance variable is something like an adjective. So I have, say, a molecule object and the method will be rotate. So I'll say my molecule rotate. And that's all I do. And if that object is properly constructed, it'll just motor on like that. The method is invoked by that application kit button. So you see, it's all very simple.

I basically described the whole idea to you just now. And that's the wonder of it. Oh, you know how also when you're working with your older computer that. That you sometimes get eye strain staring at the monitor or a headache. Well, we have a cure for that in this program, which is you load in some aspirin. And if you want to see something about modern biochemistry, you put the labels on. So you've got these nice labels and you can even motor it with those labels. And PostScript is actually putting up all those spheres with transparency and the labels as it rotates.

This guy. Now, I made it come to this position for a reason. On the right, you have this beautiful acetyl group. I think that's what that is. I'm not a chemist. Yes, it's a CO bond here. And sure enough, that interferes with an enzyme called cyclooxygenase that is in all of our bodies. And that's sort of that's the beginning of the pain killing mechanism. The point is, it's structure dependent in chemistry, especially in modern times. You need to see the structure sometimes and you need to be able to forage around the molecule, especially in organic chemistry, which you have to take to go to medical school. You will want to be able to look at stereoisomers.

One of the hardest cases. Now, one of my favorites is this beautiful sixfold symmetry molecule, cyclohexane, wonderful organic compound. I love to watch this one rotate. And let me tell you now something that no other system has that this system has, which is. There's something called Mathematica, which I use to draw a quantum wave function plot for this. And I can have a quantum wave function object. And the method could be to solve the Schrodinger equation, which solves the quantum mechanics. And if I have it online here. Yes. I got this graph from Mathematica. Mathematica is a 10 person year program.

And I just linked to it within my application. It's a piece of cake. So you can see here the quantum wave function with sort of six little mountain peaks. And I've taken advantage of enormously deep software that I didn't have to write. Finally, I have an application here that uses the microphone, the free microphone input. Well, while it's launching, let me say in no uncertain terms that like Mr. Jobs, I am not known for my programming.

I think the software group next, led by Bud Tribble and inspired by Steve Jobs, is one of the most powerful ever assembled, if not the most. And I think it's obvious that that be so. This application kit is what I love to use and how I make an application like this an instrument. I do it actually with ease. I really mean it. This oscilloscope runs like so. It has a real time display that I'll change the color for you in this environment. How about that? You can see my voice. Richard.

I knew this would happen. Now, wait. Can you whistle, Steve? Show us the 3D plot. On the right is a Fourier analyzer, a spectroscope, a spectrum analyzer, making spectrograms rather. And sure enough, I also have a log scale for it. And, you know, this is the kind of machine that can compute those logs all the way across the screen in real time like that. And you can see spikes from frequency. So this conveys the fact one about Fourier transforms to students, which is that the horizontal axis on the right hand guy is frequency as opposed to time.

Now, this machine is really an inspiration for me, and it's like I've had a cast on my mouse arm all these years. And now this cast has been removed and I have a free hand to do science. And to underscore that, I'm going to make you as my last act, a minor piece of art. And I'm going to turn this 3D spectrum option, which displays little Christmas cookies of, you know, cross sections of spectra and it's uses a graph or object that does its own rotation. And anyone can use and I can mail this cross country or anywhere on a campus network and everyone else can use it. And I'm going to make a little piece of art like so I'm going to take the microphone and hit it with my hand.

And I'll tell you what that is. That solves a problem that was first presented to me in college about 10 years ago. That is the sound of one hand clapping. It's people like Richard that we're making this thing for. OK. We have a few more demos to get through here. One of the most exciting things. About this computer is the sound and people have asked, why sound?

Why does the computer need sound? And the answer is because sound is an integral part of how we communicate with each other. We mustn't forget that until recently, AT&T was the largest corporation in the world. Their business had something to do with sound. And so what we try to do is capture a few things today about how sound really does enhance that process of not only simulations, but communication. I'd like to play for you now four soundtracks completely generated by the computer. Next, it's it's great to read things in books, but sometimes hearing them in a person's own voice is far more moving.

Next, it's it's great to read things in books, but sometimes hearing them in a person's own voice is far more moving. I do not believe that any of us would exchange places with any other people or any other generation. The energy. The faith, the devotion, which we bring to this endeavor, will light our country and all who serve it.

And the glow from that fire can truly light the world. And so, my fellow Americans, ask not what your country can do for you, ask what you can do for your country. That big shadow out there, 13 forward, 11 forward, coming down nicely, 200 feet, 4.5 down. That is looking good, down a half. Lights on. Picking up some dust. Big shadow, 4 forward, 4 forward, drifting to the right. Contact light. OK, engine stop. APA at a decent. Out of defences. Mode control, both auto.

Descent engine command override off. Engine arm off. 413 is in. We copy you down, Eagle. 3-2-1-0. We copy you down. We're out of defences. 3-2-1-0. We copy you down. We're out of defences. We copy you down. We're out of defences. We copy you down. We're out of defences. 3-2-1-0. We copy you down. 3-2-1-0. We copy you down. We're out of defences. 3-2-1-0. We copy you down. We're out of defences. APA at a decent. Auto dec... Mode control, both auto. Descent engine command override off. Engine arm off. 413 is in. We copy you down, Eagle. Houston. Tranquility base here. The Eagle has landed. None of this stuff would be possible without the mainframe-on-a-chip architecture, which allows us to be pulling things off the disk at the same time we're sending sound out at the same time we're doing graphics.

Now one of the other things that we've done is we've... by picking Unix, we've inherited a vast legacy of really great communications. And as you know, there's people across the country and the world tied together on Unix mail. The problem is it's pretty difficult to learn how to use it. And so this is our solution to that problem. It's our mail system. And in addition to doing standard Unix mail in a way that people can use, we've also integrated voice into it. Let me give you an example of a message. It is hot. We can record a message ourselves if we want to. Let's go ahead and send something.

Brings up the send window. And hitting voice brings up this thing called lip service. Hi, this is Steve sending a message on a pretty historic day from Davie Symphony Hall, San Francisco. Let's get a round of applause in it too. And of course, we can edit that if we want to. Hi, this is Steve sending a message on a pretty historic day from Davie Symphony Hall, San Francisco. Let's get a round of applause in it too. So I think you get the idea. We can send this not only within our own buildings, we can send this all over the country and all over the world.

Now the next thing I want to talk about is I think one of the most important breakthroughs we've made. Just like desktop publishing was one of the most important things of the late 80s, we believe that one of the most important things of the first half of the 90s is going to be the concept of the digital library. In other words, using our mass storage and some remarkable new software that we've written it next, we can get the knowledge of civilization at our fingertips. And what we have done is we have made the first real digital books. There has not been an advancement in the state of the art of printed book technology since Gutenberg.

And I think what you're about to witness here is the first advance of that technology. The first digital book we've done is we've licensed Webster's Ninth Collegiate Dictionary and Collegiate Thesaurus. They are online in every system. You can find what you're looking for in usually under a second. In addition to this, we've licensed the Oxford Dictionary of Quotations and the Oxford Complete Works of William Shakespeare. To start off the digital library, these works are online. You can find out anything that Shakespeare wrote having to do with love in a matter of a few seconds, as an example.

So we've given the digital library a start. In addition, all of our own documentation is online. And of course, many of the software developers are going to put their documentation online as well. And with the optical storage technology, all of this is possible, conserving a small portion of the storage available. So let me show you a few things. First thing I'd like to do is show you what we consider to be the first really good digital book, which is Webster's. And I can turn off and on the thesaurus here. I'll leave it on. And a word that's sometimes used to describe me is mercurial. So I decided I'd look it up the other day. And here's what mercurial means, of, relating to, or born under the planet Mercury.

I think the third one is the one they mean, characterized by unpredictable changeableness of mood. If we scroll down to the thesaurus, though, we see that the antonym is satirine. We go, well, what's that? By simply double-clicking on it, we immediately look that up in the dictionary. And here it is, cold and steady in mood, slow to act or change, of a gloomy or surly disposition. I don't think mercurial is so bad after all. Technology actually has a very beautiful definition. Number three, the totality of means employed to provide objects necessary for human sustenance and comfort.

We can take any word in here, like language, and blast around the dictionary in about a second and understand what these words mean. In addition to that, we have inter-application communication between all of our applications so that you can take a word from one application, have it selected and go define, and immediately message the dictionary. This thing will be up in a second or two, giving you the definition of the word. And let me show you an example of that now. Bundled with every next system is a word processor called RightNow. And this is what RightNow looks like on the screen. This particular page of RightNow has got a lot of words which I don't really understand, like anthropods. What's an anthropod, bar terrestrial? What's terrestrial mean? By double-clicking on it and asking it for the definition, there's terrestrial.

Let's go back and ask for another one. What's this mean? It's got a spelling checker in it. It doesn't understand that one. Let's pick this one. Okay. Let me show you a few other neat things having to do with the dictionary. Let me go ahead and bring up something I think you'll like, amphora. I didn't have any idea what this meant before we started this, but we have all of the pictures in the dictionary as well.

This is what an amphora is. Another nice one is cone. And you see, the quality of the megapixel display not only lets us see all of these things at once, but lets us see them in their vivid detail. So this is Webster's. Let's move on. Let's go to the general digital library. And in the general digital library right now, we have it loaded with a few things. We have Shakespeare as an example. So we'll select Shakespeare. And let's go find out there's a beautiful quote in Shakespeare that I had forgotten where it was. And I knew it had books in the running brooks, so I typed in brooks. And it found five of them for me.

And here they are. And the one I was looking for was actually the first one. As you like it, find tongues in trees, books in the running brooks, sermons in stones, and good in everything. And we've just blasted through the entire works of William Shakespeare to find these five examples. We also have the entire works of Bob Dylan online. I'm not sure we can ship this. And let's find out what Dylan has to say about love. Thirty-seven things about love. That's probably pretty good, starting with the wedding song.

So I think you understand the potential here. There's, oh, a wonderful book, the Oxford Dictionary of Quotations. There's a wonderful quote in this with the word breakfast in it. It's from Lewis Carroll. The Oxford Dictionary of Quotations, it's found 25 of them and we'll just, notice I have by the way smooth scrolling in this whole system. I can just blitz down here, Lewis Carroll, why sometimes I've believed as many as six impossible things before breakfast.

Now the next thing that I'd like to share with you is also a quite remarkable part of this computer system. Using the digital signal processor and using the talents of our music team at Next, they have written some software which runs both in the 030 and the digital signal processor which allows us to synthesize music in real time from pure mathematics. Sounds were never recorded but purely synthesized from mathematics. And what I'd like to do now is play you a few small pieces and what you're about to hear is being synthesized about a tenth of a second before you hear it and takes up only about 10% of the main system throughput and is at a level of quality which is unprecedented for the state of the art today.

The first one I'd like to play is one of the first things we ever heard. The system sprang to life almost overnight and we were able to hear this. I personally don't understand the mathematics behind this but I do understand that it's

going to change quite a few things out there. And the last piece I want to play for you is a multiple pluck string called Bullfight. Now, what we wanted to do, since we know that we're going to have to do a lot of work to What we wanted to do, since we know there are people out there like Richard Crandall waiting to get their hands on this machine, is to take the same approach in software that we did in hardware.

What we want to do is raise the lowest common denominator because that's how we're going to make a revolution. And the next computer system comes with an unprecedented amount of software bundled with every system. I'd like to take you through that. First, we start off with the operating system, Mock. Every system comes complete with PostScript. Every system comes complete with Next Step, development environment for the 90s. Every system comes complete with all of the sound and music software you've heard generating the pieces today. A digital library on your desktop. All of the digital library software and all of the books that we went through to start your digital library.

A computer isn't any more complete without a word processor than it is without a central processor. So there's right now an entry-level word processor that ships with every single system. We need to be able to communicate our thoughts and our ideas. So Mail ships with every single system and it's compatible with Unix Mail around the world. As many of you know, we were approached by Stephen Wolfram last year who showed us his remarkable program called Mathematica. And we decided that Mathematica is going to revolutionize the teaching and the practice of mathematics. And so we were able to work out an arrangement with Stephen and his colleagues to bundle Mathematica with every system. This is about a $1,000 product on a typical workstation.

One piece of the puzzle that was missing though, even still, is that as PostScript lets everyone use the same imaging model, so would a bundled database allow everybody to use the same data structures and data model. And we were able to go out and find the best technology in the world and come to an arrangement whereby we could bundle it with every computer. And the best database technology in the world, bundled with every next computer, is Sybase. And for those people in higher education and beyond who are working in artificial intelligence, we have picked a piece of software to bundle that was the most requested, the most sought after by people doing AI applications.

We're bundling Franz's Common Lisp with every next computer. Now, this is literally thousands and thousands of dollars worth of software. And you've seen the hardware. We decided to go out with a system in its minimum configuration that includes 8 megabytes of memory and a quarter gigabyte optical drive. And in terms of raising the lowest common denominator, we know we are beginning a 10-year architecture. And the worst thing we could do now is to leave out some essential ingredients. So for a 5 MIPS computer with a megapixel display, 8 megabytes of RAM, a quarter gigabyte of mass storage, and all of the bundled software you've just seen, we are going to be charging higher education a single price of $6,500.

Our breakthrough printer that everyone seems to want in their office, we are going to charge, I think, an outstandingly great price, $2,000. Now, in addition to the optical drives, we can also add Winchesters. And as I mentioned before, we have extremely fast SCSI electronics in the box. And so we're going to be offering two Winchester drives. Not so small, 330 megabytes and a 660 megabyte drive. And we're going to offer these at some pretty unprecedented prices, $2,000 for the 330 and $4,000 for the 660.

Now we're starting to roll this system out early in November of this year. And we have a 0.8 software release, which is rolling out to software developers through the rest of this calendar year. Early next year, we will have our 0.9 release, which is for software developers and aggressive end users. And sometime during Q2 next year, we have release 1.0, which should reach perfection and be usable by everybody. Now, to do all this, to accomplish what you've seen here today... We haven't done it ourselves. We've had great partnerships. And partnerships take time, so you can't have very many of them.

And we've been fortunate to have some of the best. These folks have invested their dollars. They've risked their dollars long before we could even show them a prototype. And I'd like to have some of them talk to you in their own words now and tell you what they think of what we've accomplished together. Hi, I'm John Warnock from Adobe Systems. I've actually been working 10 years to get to this day. A computer system where the screen matches the printed output and where all the software interfaces are device independent.

For the last year and a half of those 10 years, our engineers at Adobe and the engineers from Next have been working to integrate Display PostScript and the Next software into this truly remarkable system. We at Adobe believe that what you are seeing is an early look at what computing will be like in the 1990s. Congratulations go to you, Steve, and to your team. Hi, I'm Bryant Wilder from Motorola. We at Motorola are extremely excited that Next has chosen the 68030 as the central processor for the Next computer system. But what has got me really fired up is that Next has designed Motorola's digital signal processor, the 56001, into every Next computer to give it compact disc quality sound, facts, modem, and other applications limited really only by our imaginations.

When Next came to us with this vision, we were convinced the 56001's incredible sound capabilities in their computer would change the way people and computers work with one another forever. We knew the Next computer system would end the age of the mute computer. The Next computer is the starting gun to begin the age of the sound computer, the humanized computer. Next is on the right track, in my opinion, and Next's sight and sound is the way computing is going to be from now on. Mr. Jobs, everyone at Next, I would like to thank you from the bottom of my heart for the innovative product announcement based on a new concept.

In this system, based on Canon's latest technology, a laser printer has been adopted as an additional output device for the light-sensitive memory of the main unit. With Next and Canon's innovative technology, I am very pleased to be able to contribute to society by cooperating with both parties through the Education Committee. I sincerely pray for the success of Next. Hi, I'm Stephen Wolfram. This is really a great day for computers and mathematics. The combination of Mathematica and your computer system, I think, is going to bring computer mathematics to a lot of people.

I think we're finally going to be able to show the world that personal computers can do more than just things like word processing. They can actually do stuff like mathematics that involves computation. Having Mathematica bundled on your machine is going to be particularly exciting. Just a little while ago, nobody really thought that one could do real mathematics on a personal computer. When we brought out Mathematica, we showed them that wasn't true. Now, by having Mathematica bundled on your machine, we're going to let them take computer mathematics for granted. My guess is that a lot of very good science and mathematics will come out as a result, and we're going to see a lot of new, exciting ways for people to teach those subjects.

Good morning. My name is Bob Epstein. I'm Executive Vice President at Sybase, and it's our pleasure to participate in the announcement of the Next computer. The Sybase SQL Server is an integrated part of each Next machine, providing the standard means to store and retrieve data for a new variety of application programs. Next, by incorporating all the critical software elements for a complete system, represents a new standard for desktop computing. Good morning. My name is Fritz Kunze, and I'm from Franz, a company that develops software tools for scientists and engineers.

I really am very happy to be here as a partner with Next, making our Allegro Common LISP available on Next's revolutionary new machine. LISP has made possible many software innovations in the past, including artificial intelligence, windows, and pull-down menus. But until now, all serious machines that ran LISP were well beyond the reach of all but the biggest budgets. We really believe the Next machine is the first computer that can bring all the benefits of the LISP technology to everyone. By working with Next, we will make LISP the dominant programming environment of the 90s. Congratulations to Next on developing a machine to take us there.

Excuse me, I'm going to have to cut this short so we can finish up. We're running a little late today, and there's this other group called the Symphony that thinks they need to use the hall this afternoon. There's one last partner that we didn't have a chance to get to. Oh, hi, John. There's one last relationship that we've been very fortunate to form over the last year, and I need to tell you about it today. It's true. Let me brief you on what we've done. You remember our architecture and that the four center parts we gave a name to, which was Next Step.

What we've done is we have licensed Next Step to IBM, and they have put it on top of their hardware and on top of their Unix operating system, which is called AIX. And they have licensed PostScript from Adobe as well. And the entire Next Step platform is running on top of this with the same applications running on the top. And that is the key. The key to this relationship and what makes it so great for both companies is that the same application software, which is written for one machine with a recompilation, will run on the other.

Same user interface, same documentation. And this is going to be a tremendous boon to software developers and to our market as we move into the 90s. What do you think? One of my heroes has always been Dr. Edwin Land, the founder of Polaroid. And he said that he wanted Polaroid to stand at the intersection of art and science. And we feel the same thing about Next. And of all the things we've experienced together here today, I think the one that strikes closest to the soul is the music.

And so we've decided to end our presentation here today in Davies Symphony Hall with a concert. We're going to have a duet, Bach's A Minor Violin Concerto, performed by one great human artist and one Next computer. The computer is going to be synthesizing its part, again, from pure mathematics about one-tenth of a second before you hear it. And now I'd like to introduce Dan Colby-Alka, the principal second violinist with the San Francisco Symphony Orchestra. Thank you. Good luck. Thank you.

Thank you. ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪

♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪

♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ Thank you. Thank you.

This is great. We didn't have time to program in an encore today, but all of us at NEXT are really confident that as we move this technology out into your hands over the next few years, you'll make your own encores.

Thank you very much. applause ♪♪♪ ♪♪♪ ♪♪♪ ♪♪♪ applause applause

Wow. Board of Director, member of NEXT, one of the people that really urged us to start this company in the first place, and the president of Georgia Institute of Technology. I was supposed to have one on here, actually. And, uh, is this one working? Oh, I didn't turn it on. How's that? Thank you. Technology, can't live with it, can't live without it. Ross Perot, member of our Board of Directors. Dr. Bud Tribble.

Computer. And I think that's about it, so why don't you guys come on over here. applause We'll deep-six this. Thank you. Okay. So, what should we do? Is there color in the room? Oh, excuse me. Color. Why don't we explain about color? We had a... What is that strange noise? We had a fundamental decision to make about color, and that was that we decided that megapixel was more important than anything else.

In other words, once you put out a small screen, you've got to live with the results forever during the entire computer architecture's life. So we knew if we came out with a small color screen, baby color we call it, that we would be forced to live with software needing to run on that small screen forever, and all the application developers would be constrained. You would never see programs like Dr. Richard Crandall's Stat Lab today, because they wouldn't fit on the small baby color screen. And to do megapixel color today would have doubled the price of the machine. So what we decided to do was do megapixel black and white at a level of clarity and definition that I don't think the industry's ever seen before, and you will see some very fine stuff from us in color next year.

The display here is not only very high-definition monochrome, but it's also two bits per pixel, so we get some grayscale involved. And I think because our unit... expandable you'll see some stuff from us next year that you can add on to it if, you want to go into new dimensions I think we'll be talking about a lot of, stuff next year in that regard do we expect software developers I shall do, we expect software developers to distribute their software on optical, disk cartridges the answer is yes the question is if the cartridges cost $50 is that going to mean that software costs more for packages that cost in the, multi hundreds of dollars of course not one thing we also need remember is that, the most expensive part of any software package today is often it's, documentation because of the digital library capabilities of the machine we, can put the documentation on the disk itself thereby saving all the production, costs and in addition to that since we can make as a software developer can make, the disk a day before they ship it to a customer unlike manuals which they have, to print two months in advance they cannot have inventory of products that, aren't selling and things like that so there's some tremendous advantages to, to actually using that technology we're selling the machine to the higher, education market we started off saying that that was the market we wanted to, serve and and and it's really true and we're going to be selling the machine to, higher education for $6,500 as we outlined in the presentation today it,

will be available to people who work and study at educational higher institutions, higher education currently just in the United States turns out that the higher, education market is is very large and it's going to probably take us a while, before we we can even begin to meet the demand of higher education in America so, that's what's going to come first for us and after we get to a point where we can, do that then then we can start thinking about other things yes next is building, a direct sales force and with the help of higher education to guide us and steer us, right and we're really happy with how that's coming we'll be selling directly to higher, education I think we're going to be selling to higher education real real, simple folks more more companies tend to die of indigestion and starvation and so, we're going to stay really focused but we're most of you at the presentation, today I assume all of you were what did you think all in all did you like it yeah, yeah, question was are we making any money when we sell one at $6,500 are we losing, money on everyone and making it up in volume no we are we are making money we, like to employ engineers and come up with new breakthroughs and to do that you, have to make a profit and and we are what is our optical disc technology yes,

we are using magneto optic technology to most of you how that works it's it's, really great well the way it works is on a disc drive like a Winchester you can only, make you the whole point is to make the magnetic spot size very small so you can, fit a lot of bits on the surface and to do that you have to fly you have to put, the head closer and closer to the media and it's been described as an air 747, flying at 500 miles an hour six inches off the ground that's what a head of a, Winchester looks like so you can understand what a Winchester crash is, like when you lose all your data the optical disc is very different what it, does is it uses a beam of light to heat up a material to its curie point in which, case it changes its magnetic orientation and it works very well the only thing, that ever gets close to that media is a beam of light and we're finding that the, optical media is far more reliable than then the low-end Winchester's that you, find in most computers how fast the optical disc actually spins that I, believe 3,000 RPMs giving us giving us a transfer rate of about a megabyte per, second access time on a disc the usable access time is I think around 60 to 70, milliseconds and I can sorry what's that, oh it's really easy you pop it out you take it to the library where there's a, to drive machine and you copy it or you use the network and copy it on your, friends in the next office because these things are removable you don't have the, same problems you have with Winchester's you can pop your whole world out and, throw it in your backpack.

I believe Intel has Intel excuse me IBM has plans to bring this out on their, risk-based platforms and on their Intel based platforms I think it's gonna be, pretty exciting I think the 386 platforms as well but I think probably, it's best to ask them they decided this probably should be next product, introduction but so I think we've there's some press materials. The question is will we be incorporating object-oriented kits to help build, animations and how far can we go we already have that's part of what's in, the system and you'll be seeing some things I think over the next six months, roll out that are they're pretty exciting we currently don't have NTSC, video output on the machine although that's entirely possible we will be, selling to universities who will in turn be reselling to faculty students and, stuff in the the prices they charge for those things are up to them do we have, any contracts with any universities yet yes we do several I have absolutely no, idea some people have said a lot about this computer haven't they and some are,

right and some are wrong I get those are the kinds of things we just can't, discuss publicly we can make a lot but probably not enough Daniel Garrick. European universities as soon as we can come even close to supplying the demand, of higher education in America we would love to have this be a worldwide, standard and we intend to do that but first we've got to focus on higher, education in America I don't know it's a pretty big market. We could but but we'd rather not question was asked about side base one of the, great things about this computer is that it's not just a hot hunk of iron it's a, very very sophisticated software system and we realized that the more we put in, this box the higher level that application software developers start at, which means the neater programs they can write in the same amount of time or they, can write programs of current complexity in far less time and realizing that, databases are very important we were able to reach an agreement with, Sybase to bundle their entire SQL server and the database libraries in this thing, so that people are going to be able to take advantage of what we feel is the, best relational database technology around in every computer they buy and it,

what that means is that when all these people write all these applications, without ever talking to each other they can exchange data because they'll all be, in the same database well application programmers have access to you mean how, will they duplicate their their disks oh yes they can just get a system and, duplicate them themselves yeah our relationship with Lotus is that we've, been working with Lotus for some time now to put their software on this, platform and I think you'll see some pretty neat things unfortunately we, can't announce products for our application software developers they have, to who makes our disk drive we were really lucky when we started next in, that we got calls from the presidents of Sony Motorola Canon and others and they, said hey when you guys were in your former lifetime you used to romp through, our technology laboratories and you used to spot technologies and help us form, them into marketable products and we really like that because both of us won, so we'd like you to keep doing that and we'd like you to keep doing that because, a maybe you'll be the next big computer company and be because you're the, cheapest R&D we could ever do so we had access to I think probably more of the, in-depth technical research going on at some of these companies than most other, people have the good fortune to have and we spotted something in the laboratories, of Canon.

And what it was was it was an early prototype of this optical technology. Now Canon as you know is very very good at coding materials like lenses, like, laser printer toner cartridges, like optical cartridges and they came up with, the best media in the world and we work together with Canon for the last two, two and a half years and it turns out on our board I don't think I have one right, here do I know on our board those two big mainframe on a chip chips it turns out, that three-fourths of one of them is an optical disk drive controller on a chip, it start off with three boards this big in the laboratory and we shrunk it down, to three-fourths of one of those chips that's all proprietary to next and it's, what's enabled us to get the performance up and bring the cost down the other, thing is it turns out hooking one of these up to a software system is really, difficult and we put in a tremendous amount of work to make this real in a, computer system as the main primary storage device so we've done this in, partnership with Canon and it's turned out really fantastic of network computers, can look like one large file system and secondly we've put front ends on Unix, mail as you saw today during the demonstration so that mere mortals can, use these types of things but the base underlying technology we inherited a lot, of the minute we decided to go with Unix the the other thing that I might mention, is we have a course full very high-speed Ethernet built in every computer and the, thing of course supports the TCP IP protocols which are the standards of, higher education first question is are we sending any of this to OSF we have no, plans to do that at the present time and secondly why didn't we use a risk,

processor it turns out that the 68030 in this box running a 25 megahertz is, running a real solid five MIPS that's about seven MIPS from some other, companies but a real solid five MIPS and that is approximately half the, performance that you get out of a forty to fifty thousand dollar risk-based, workstation so what our game plan is is real simple we may not always build the, fastest computer you may be able to go pay fifty thousand bucks and buy one is, faster but given a certain amount of time later we want to bring the same, thing on a desktop for this class of price and that's that's what we're, planning on doing I think five MIPS puts it about ten times over where we were, with the Macintosh the day it was introduced it is a Canon engine inside, the printer yes it is actually we worked with Canon on this engine as well and, what we did was we took a standard SX engine and we made it a lot smaller so, it uses the same exact cartridge as all the other popular laser printers in the, industry but it's about sixty percent of the size it's got a totally straight, paper path which means it's a lot less likely to jam in these curvy things and, it has to plug directly into the wall but it connects to the computer with one, three meter cable sec and the there are no switches and no lights on this, printer it is turned off and on totally by the computer when a paint when you, get a paper jam or you're out of paper you don't get some little LEDs telling, you the big screen tells you matter of fact it can even tell you as in a voice, message you can say you're out of paper so sorry we do our imaging inside the,

workstation if you look at a seven thousand dollar laser writer you've got, a 68020 processor and you've got about two megabytes of RAM we've got a 68030, and eight megabytes of RAM and a disk to hold all the fonts so we image like, crazy and we turn on one of our DMA channels and blast the bits out of our, over a cable we have a custom VLSI chip in here that inherits those bits and, controls the entire printer in one chip the printers designed to work with a, next computer system only the thing that software developers want is when, they develop an application for one computer they want to in a reasonably, short amount of time with a reasonably small amount of investment be able to, generate versions of that applications that they can sell on other computers, they don't need the same disk to work in these various computers but within a week, or two or a small number of weeks worth of work if they can take the result of a, three-year development effort and all of a sudden in three weeks have a version, for a new computer by simply recompiling it to run on top of the next step running, on another hardware platform this is a revolution and that's what we're trying, to achieve as an example if somebody writes a program for this next computer, here as a perfect example is frame we looked out there and we asked who's got, the best technical publication system out there running on workstations and, our answer was frame and we started working with frame frame was the first, application software developer I believe we started to work with we've been,

working with them for a long time their software's up I was going to try to, demonstrate it today but we were running out of time and it's quite remarkable, and they will be able to take this when they're done and theoretically recompile, it for next step running on the IBM platforms and be able to generate a, product for the IBM platforms in a fraction of the time it took them to, write the original product for this oh I mean everything's this is not a problem, yeah do we excuse me what was that do we allow a bridge, no this machine doesn't run Macintosh software just as the Macintosh didn't, run Apple to software sorry what's that, um sure we um it actually happened there was a little snippet in Forbes, magazine I don't if you saw it it it actually started when both John, Akers and I happen to be attending the 70th birthday party of Catherine Graham, in Washington we were introduced and we got to talking and he invited us out for, lunch in Armonk and this was an interesting experience what's really, what's really turned out to be great is that we're focusing on higher education, and IBM has a much broader market than we are focusing on ourselves and they, also have the ability to incentivize every single software developer to write,

for the mainstream platforms that they adopt so informing this this partnership, with IBM we've pretty much guaranteed that software is going to be developed, for the next step platform at a previously impossible pace, um yeah I think the team that built this computer works at next, and the team that did next step works at next and there are some things about, this computer of course which we haven't licensed anybody obviously not just the, hardware but the entire digital libraries the digital library stuff has, not been licensed to anybody nor is any of the sound and music nor have any of, the bundled application software so I think that taken together we have a, pretty strong competitive offering but most importantly the world has now got, two companies who are going to both be working like crazy to make the best, next step platforms in the world and everybody wins when that happens so stereo.

I think you need to ask Bill Gates what's that, we'd love to have Microsoft software in our machine I think you've got a, we can't announce products for these guys so I think you have to go go ask, Bill yourself we did it all internally to next and we can make chips at a variety, of places we've used several people actually yeah it doesn't really matter, we did the first set of chips with Fujitsu we could do them a variety of, places what's that oh this is a this is a very sophisticated machine and VLSI, design is not easy nor is writing software and so if you drove by next at, one or two in the morning you know over the last nine months you might have seen, the lights burning in a few offices but we managed to finish it and it's come out, great and I think a few people have said it's a little late I think most people, who have seen it say it's about five years ahead of its time, what impact will this computer have on the computer industry we think more in, terms of what impact is this computer going to have on the people that use it, and so we aren't you know much more about the industry than we do at this,

point we were focusing on the people that are going to use this thing and how, to bring the most far-reaching capabilities to them and so that's that's, where our focus is right now, what do we tell people that want to buy, one but aren't in higher education enroll, By the way, let me take a moment to introduce a few other people that showed up. This is Rich Page. Rich is one of the founders of NEXT and has run the entire digital electronics engineering effort. Those two mainframe-on-a-chips were the result of Rich and his group working extremely hard over a long period of time.

I'm amazed at it. Randy Hefner. Randy comes to us where he used to be at Hewlett-Packard. And Randy's been on the NEXT team for about a year and a half now, two years. And Randy runs the entire automated factory at NEXT. All of manufacturing works for Randy. Susan Kelly-Barnes. Susan is our CFO and is responsible for all of the finance, all of the information systems, and a sundry of other things at NEXT. Gary Moore. Gary Moore is our general counsel and also a member of the senior staff at NEXT. And George Crow. George is one of the founders of NEXT and runs all of the analog engineering and did all of the engineering for the megapixel display, all this crisp video you're seeing we thought could never be done on a big display.

And George came out of the back room one day and said, you guys got to see this. So I think it's very easy to get up and demonstrate a product when you have such great people running the company and designing it. I am exceptionally proud of the 175 people that work at NEXT because they started without, in some cases, when we started the company, of course, we just started it with a fresh sheet of paper. And as people have joined us over the months and the years, they have joined based on not a product, not even a description, they've joined on the vision.

And I think this is a really great day for all of us to be able to talk about it because we haven't talked about it for so long and get some feedback. Boss? We don't have a role. It's been a fascinating day for me. I'm one of the old people in the computer industry. Go back to the vacuum tube machines, the 650, the 704, 705. I can't tell you how many hardware announcements I've attended in my life. And this is the most exciting one and the best presented one I've ever attended. I don't think Steve ever left the computer industry.

You've got to remember, Steve is a young man, is a very young man, changed forever the computer industry. He did things when he was in his 20s that most of us don't do until we're in their 50s. And while I've spent my life in advanced technology, my fascination has always been with the people and not with the chips and the machines and the lasers because the people do all of that. I have been around many, many high-talent teams. I have been around environments to produce excellence that I thought were perfect or near perfect. The thing that captured my mind and heart and pocketbook was Steve Jobs, who at 32 has to be at least 85 in terms of business experience in the computer industry, and this incredible team that he assembled with him.

And if you could be there as I have been and see them work and see the excitement and the electricity and the candor and the commitment to being the best in the world, then you wouldn't be too surprised at what happened today. Now in closing, I've got to mention this, it's just been, Steve's been very sensitive. See, Susan's a Texan. And from the time I became an investor, she got a, and Susan and Bud are married. Now you don't find that much intellectual horsepower typically in one household. So Steve and I have a clear understanding, I'm never going to try to lure Bud or Susan away from next. I get first option on the children. Thank you.

We've got time for, I think, one or two more questions. Cheryl. University what? A degree? Oh, I don't, you mean the fact that I don't have, I'm a dropout. I have a pretty fond spot for higher ed in my heart. What happened was I went to Reed College, just like Dr. Richard Crandall. And I ran out of money after about six months, so I dropped out. But they let me drop in for about a year and a half after that. And Stanford let me drop in for a year. And it's kind of something that's always stuck with me. I didn't have any money to pay and they said, fine, just take the classes and learn.

And that to me is what it's been all about. So hopefully, I think all of us feel really great about being able to put something like this back into that community. And again, I can't stress enough, this thing wouldn't be here today if it were not for our advisory board. They have led us into their research labs. They have let us hire their best grad students. And they have kicked us in the pants when we wanted to compromise, and more than once, matter of fact. If it weren't for them, this machine would be a very different machine. I think all of us at Next feel an incredible debt to them for hanging in there with us and making sure this thing came out great. I don't, I mean, I'm, I don't think any of us have a better crystal ball than you do.

Who knows what's going to happen? We made a, actually a very important decision at Next. We said, why should a student at Reed College, a small liberal arts college with a thousand students, or a professor or a researcher at Reed, have to pay more for their computer than someone at a very large institution, let's say the University of California at Berkeley? And the answer is, they shouldn't. And so we decided, what's the best price we're going to give to anybody? And we said, can we architect our company in a way that it doesn't cost us any more to sell one computer than it does to sell a thousand? Obviously we made some distance there, and we decided to offer our best price to everybody. So everybody gets the best price, and that's how we decided the price.

It's very democratic, and I think it's in the spirit of what we're trying to do with higher education. One last... How did we decide on the cube design? It turns out after examining several, it is by far the most efficient in terms of packaging stuff in here. As we said earlier, we've got one board in here, we have room for three additional boards. We can even plug three of our processor boards in here, in addition to the one that's in there. We're going to get up to 1.3 gigabytes in here, and up to three next year. So what you can fit in a one-foot cube, which is smaller footprint than a PC even, is quite remarkable. Thank you very much for coming today. We really appreciate it a lot.

Thank you.