Professor Reddy, other faculty, graduates, families and other honored guests: thank you very much for this fine greeting; and I want to bring greetings from my colleague, Bill Joy, who wished he could have been here and fell suddenly ill this week. While I hope to be able to fill Bill's shoes, I find that his hat doesn't fit me! So I'll hope you'll excuse me if I perch it up here. [removes mortarboard from head and places it on lectern]

I love computers and computer science; and I like it because in computers and information science and computer science and all its applications, the rules are constantly changing and that keeps it constantly fresh and interesting. I'd like to begin by reminiscing about old times. My first computer, back in 1968, was an IBM 1130 and it had 8 kilobytes of memory. I wouldn't use that to run my toaster nowadays! But back then that was considered a full-sized scientific computer. It cost probably $100,000 at that time; and it had a full Fortran compiler and could be used for all kinds of interesting, scientific computing work.

By 1975, I was at a university and using a much larger computer, the DEC PDP-10, at that time. It had some nice round figures that make it easy to remember: it ran at 1 million instructions per second, had 1 megabyte of memory, and cost 1 million dollars. And like many people of my age at that time---I would have been, oh, in my early 20s---I didn't dream of having a fast car, or owning a speedboat, or anything like that, but I dreamed of owning a computer like that and having it for myself in my basement (instead of having to share it).

Well, in 1987 I bought one. I walked into my local computer store and I bought a machine that ran at 1 MIP and had actually more than a megabyte of memory, and it cost only half the price of a used car, along with the laser printer that came with it. And now you can buy the same for less than a tenth of the price of a new car.

How is it that computers can keep improving at such a rapid rate? At that same time in 1975, the world's fastest computer was something called the Cray-1. That was 20 years ago and now the world's fastest computer is over a thousand times faster than that. This is consistent with something that is known as "Moore's Law", which is that computers seem to improve at roughly exponential rates and have been doing so for 40 years or more, so that today's fastest computers are over a million times faster than those that first came out in the middle 1950s or so. Moore's Law comes in various versions---roughly, that the number of transistors on a chip will double every two years, or that for a fixed cost its speed will double every two years. (It turns out most of these measures [are] consistent.) But computers have been doubling roughly every two years, by whatever metric you want to choose, and it looks like they are going to keep doing that for at least another 20 years---possibly more---and it's hard to understand. This is a very unusual technological phenomenon, which I think has no comparison anywhere else. Let's try to compare it.

How many of you have ever taken a flight on the Concorde, which is (sort of) the world's fastest airplane? I haven't. It's a pretty pricey ride, actually. But it can get you over to Europe in, what? two, two and a half hours. It's the fastest airplane we have today. Suppose that we could improve the speed of airplane transportation by a million times. You'd be going slightly faster than the speed of light. Professor Einstein told me not to bet the farm on that one!

So things keep improving. What is it that improves? The secret is that in computing we aren't just trying to speed up physical processes. If you ask people about Moore's Law, they'll say, "Oh yes, the technology keeps improving. We make process improvements. The transistors get smaller. The transistors get faster." But those explanations miss the point. The point is that we're not speeding up physical processes, we're speeding up computational processes. We are dealing with information, and information is very special; it's flexible in a very special way. The point of a computer is not to achieve a physical process but to process information ---to compute---and the key thing is that you can change the way the information is represented without changing the computation. You can get the same informational result even if you get it using a very different physical machine.

So, 40 years ago, they were using devices that are hardly used at all today to represent information. Vacuum tubes were used. Information was stored in acoustic form, as acoustic vibrations on the surface of a pool of mercury. Electrostatic TV screens were used and electrostatic charges on the face of the TV screen were used to store information. Nowadays we make nearly everything out of silicon. We put them on teeny, tiny chips--- sometimes called microchips---and information is stored in tiny capacitors and is processed by tiny transistors and that is what, lately, we have been able to make smaller and smaller. But there may be other approaches in the future. We may use optical computing and make a complete switchover from electronics. But the basics of computer science will remain the same because the emphasis is on the information. All these computers, no matter what they are made of, whether using electronics, using acoustical vibrations, optical representations, or whatever, in principle carry out exactly the same computations on the exact same information.

So, let's try one more time: Is there some way we can make transportation faster? Well, we might be able to if we could change the rules of the game. So I invite you to consider the transporter on Star Trek. They don't actually say so on the TV show, but maybe it can go at nearly the speed of light. They don't seem to talk too much about the technology on the TV show. But something like the transporter effectively changes the rules of the game and in much the same way that the rules of building computers have changed. Interestingly enough, in the original Star Trek series ---here I'm going back to the mid-1960s again---they seemed to have this assumption: that a transporter just magically takes a physical object and moves it from here to there. By Star Trek: The Next Generation, they seem to have a more sophisticated attitude and there are some indications in the TV shows that the transporter operates informationally rather than purely physically, because there is at least one plot line in which the computer reports, "Gee, while I was transporting this guy, he seemed to be holding a weapon and it was in the middle of discharging. Would you like me to suspend that before I finish the transport?" And in another case, the computer indicated that something was transporting aboard bearing diseased bacteria: "Would you like me to edit those out of the transport?" So there is some indication that the transporter had suddenly been reconceived, within this fictional framework, as something that converts the physical object into information, which can then be modified and edited so that you can reconstruct something that's slightly different on the other end. This is a very interesting shift in the point of view and it shows some of the influence of computer technology and its culture on the thinking of those who are basically weaving the new myths for our modern culture.

Information has another interesting property: it's flexible, in that it can be processed in a large number of ways and, in particular, you can bring many computers to bear on a single problem. Again comparing it to airplanes: Hiring ten airplanes instead of one is not going to get me to London any faster but, while using ten computers to process may not make my computation ten times faster, it might make it five times faster. There is a whole area of the study of computer science that studies this phenomenon: how is it that you can bring more resources to bear to get something done faster.

Let's consider another example of how transforming something into information can change the whole way you do business. How many of you have seen such movies as Terminator II or Jurassic Park or The Lion King? Ah, now I'm beginning to see hands waving in the audience. This is good! Okay, I enjoyed those movies. Computers were involved in the production of all these movies to provide the special effects and I'm sure you have seen them: the strange morphing in Terminator II; the dinosaurs, particularly the thundering herd of dinosaurs in Jurassic Park; and the similar thundering herd of gazelles in The Lion King. (A friend of mine who is an expert in computer graphics commented that morphing was probably the graphic cliche of the early 90s and thundering, animated herds of animals is probably the cliche of the late 90s. You can expect to see those in one or two more films before they will lock onto some other cliche. Part of the reason for that is that as long as things keep moving, the animators don't have to do too good a job. It's actually much easier to draw something running fast than to draw something standing still, where you can admire it. My friend says that when you see one horse standing still for you to examine in the middle of the screen, then they'll really have done something, if they have made you believe that it really is a horse!)

So how is it that computers manage to do all these wonderful things in the movies? Well, they don't do it by making the movie camera run faster. They don't do it by spraying paint onto the animation cels faster or better. Instead it is done in an entirely different way. The problem is transformed into one of information. Instead of making a model of a dinosaur out of papier-mache or latex-rubber or whatever, you make an informational model of a dinosaur. The computer computes what the dinosaur would look like and only at the last minute is it converted into film. And by doing it in this manner you get much more flexibility and---in principle, eventually, perhaps---much more speed, although right now it's still a fairly computing-intensive, and therefore time-intensive, activity.

But this is a standard strategy we see a lot in the application of computers nowadays: Take some problem in the real world, convert it into information, and then throw a supercomputer at it---or maybe even a small computer---and then, at the very end, turn the answer back into something that is applicable in the real world. In areas of industry where they used to used wind tunnels and make physical models of airplanes or of automobiles, now it is all done with computers. The oil industry still depends on geologists to actually go out in the field and look at the rocks and try to decide where the oil is, but more and more they are also relying on computers. They lay out a bunch of microphones on the ground, they set off some charges of explosives, listen carefully to the echoes, record all of these reams and reams of information on magnetic tape. Then they take the tapes back to the laboratory and the computer tries to figure out from those echoes where oil is likely to be. Most of the work is done in the informational domain.

So my first point is that information is very flexible and this allows the underlying computer engines to change. Therefore, coming around, for at least the first time, to the title of this talk, "The Computing Landscape in the 21st Century": I don't know what the supercomputers of 20 years from now, in the year 2016, are going to be made of. They may be electronic, they may be optical, they may be something else. About all I feel confident in stating is they'll probably be about 1,000 times faster than today's supercomputers; and the personal computers of 40 years from now will be about that fast. Again, I don't know what they'll be made of. The only constants are that things keep changing and things keep getting faster.

So we are going to have all these ridiculously fast supercomputers in the 21st century, I project. What are we going to use them for? Well, that's hard to project but I'm going to try. Back in the 1950s, when the first computers were built, they tried to project how much computing power would be needed for the next 50 years or so. And their estimate was that 20 ENIAC computers would probably carry the United States through to the end of the century. I don't know about you, but I've got more computing power than that in my own home right now---and I'm sure it adds up to a tremendous amount more than that throughout the country! About the worst thing you can do is underestimate the uses to which we can put computers.

So what are we going to use them for? Well, one much ballyhooed concept, perhaps rightly so nowadays, is virtual reality. I'd like to suggest that we can construe the term "virtual reality" in two different ways. One is the image that you often see in popular culture nowadays, virtual reality as simulations of physical impressions, making you think or feel as if you were in a different physical environment. You've seen people in the movies, on television, wearing goggles and perhaps wearing earphones and sort of wandering around---like this---and looking ridiculous to everyone else, but they're in another world, in some sense. And it's actually not all that crazy. It's an important thing to do and probably has very practical applications as well as obvious entertainment applications.

But I'd like to suggest that virtual reality can be construed in another way. I believe that our notion of reality, at least among the computer propeller-heads---among which I count myself!---our notion of reality is changing because our modes of human interaction are changing because of the computer. Electronic mail, the World Wide Web, online systems such as CompuServe and America On-Line, online chatrooms, are providing new ways for us to interact with other people. And because these carry different information, some of the social cues that we are used to seeing are eliminated and news ones are introduced. So when I'm exchanging email with someone online or chatting with someone, I have to ask myself: Is this in fact an expert I'm talking with, or perhaps just an articulate 14-year-old? (Actually, maybe that 14-year-old knows more than the expert does about something!) Is the person I am talking to a woman or a man? And additionally, for what purposes does that distinction really matter when you are online? Soon we may ask the question: Is that a person I am talking to or a computer? Is that really an incredibly good simulation, on the other hand, of what I think is a person that I am exchanging letters with?

So this raises a very interesting question, which is certainly not new with me---it's been around for quite a while, 40 or 50 years--- could a computer pass itself off as human? Well, it might be very difficult in the physical-presence sense, but in the sense of exchanging electronic messages through a network, I think that's entirely plausible, perhaps within the next ten to twenty years. This raises such questions as: Does a good simulation of a person necessarily have to be intelligent? Does it have to be conscious, for that matter? Or does it merely suffice that it appears to be to the recipient? Is it possible to fake a simulation well enough that it appears to be intelligent or conscious? And of course this raises all kinds of philosophical questions about what does it really mean to be intelligent or be conscious and I certainly don't propose to answer those here---I'm going to entirely finesse that!

But I'll just point out that these questions have been raised before. The great computer pioneer Alan Turing raised it quite some time ago, proposed this as an empirical test of intelligence. He proposed an experiment in which you put a man and woman in one room communicating with a third party by teletype---which feels very much like email---with the man trying very hard to convince the interrogator that he is the woman and the woman trying to convince the interrogator that she is the woman. (I'm sure he chose the sexes fairly arbitrarily for this purpose.) And then, he suggests, suppose you replace one of the participants with a computer. And if the computer can fool the interrogator, then maybe the computer shows some semblance of intelligence. I'll leave it to you to decide whether you believe that argument or not, but I think that we have ample evidence already that computers can fool some of the people some of the time. There have actually been competitions held to see whether computers can fool a panel who have been brought in, knowing they are trying to tell the computers from the people, and not everyone has guessed right.

I think that in the 1960s, the discipline of artificial intelligence was a bit too optimistic, thinking they could solve this problem of building an artificial intelligence within ten or twenty years. But I think that surely within the 21st century, and perhaps within even twenty years, we are going to face this problem squarely and head-on of how do we know when it's a person we are talking to and how do we know when we are talking to a computer. And again I raise the question: for what purposes does that distinction matter?

I would compare this to the situation with chess, which thirty years ago was considered a problem in artificial intelligence. Now I can buy a computer off the shelf for $30 that can beat me at chess routinely. (Yes, I am a life member of the U. S. Chess Federation, but that doesn't mean I'm that good a player! It just means that I ponied up the dollars to buy the life membership. I figured on the magazine alone I'd break even in twenty years. I've reached that point now, so I guess it's gravy from here.)

But the situation in chess now is that the best computers can beat 99.999999---[pause] give me a couple more 9's---percent of the people in the world. Perhaps you read recently in Newsweek or Time about the computer Deep Blue facing off against the world champion, Garry Kasparov. And he beat the computer, but out of six games he lost one. And so with chess we are in a situation where computers can beat most of the people and only the very expert can beat the computer. I think, in the situation of conversation with computers, trying to decide whether that's a computer you are talking to or human, very soon computers will do well enough that they will fool all but the experts who know exactly what to look for. I think that is something that should give us pause. I think it's an exciting possibility, but it may also cause some interesting social problems.

Okay, I would like to get away from philosophy now a little bit. I'd like to talk a little bit about something more practical. Let's talk about the World Wide Web. An amazing phenomenon! I told you earlier that computer technology seems to grow exponentially, at the rate of about doubling every two years. Well, interestingly enough, the Web has been exploding even faster than that. Electronic mail, from its first, moderately widespread use, to becoming a common thing that is used by people outside the specialty, took about 15 to 20 years to spread. With the Internet already in place, it only took about a year or two for the World Wide Web to suddenly spread and be used very widely throughout the network. And one reason it was able to spread so rapidly is that it changed the rules. Suddenly the concept of a Web browser made it much easier to get at information in the Internet. Because information was easy to get at, there was more incentive for people to go to the trouble of creating the information and to publish it, which in turn led to even more incentive to make better browsers. And the thing fed on itself. Right now, I believe the size of the Web is doubling roughly every six months. There was a period a year ago when it was doubling every six weeks. I think it has slowed down a little bit, but not that much.

A year ago the projection was that the Web was growing so rapidly that by the year 2000 every person on earth would have a Web page. Of course this is crazy. Whenever you have any kind of exponential growth, you know it has to top out somewhere and projections like that are usually an indication that either the projection won't come true or that something is going to change.

There was a prediction earlier in this century, perhaps in the 1930s or 1940s, that by about the year 1980, at the rate of growth of use of telephones, everyone in the country would have to be a telephone operator in order to put all the calls through, just because there is so much work in plugging and unplugging the jacks and making the connections. Well, guess what? That prediction came true. Today, every one of you is a telephone operator. You make the connection yourself instead of asking somebody else to switch the jacks for you---and the secret is that a computer got into the picture. The problem was changed from the physical one of connecting jacks on a switchboard to the informational one of telling it what's the phone number and letting a computer make the connections at the last second. Again, an example of how shifting the problem into the informational domain can make things much more efficient, much more cost effective, and allow people to do more things for themselves.

By the way, it's interesting to note that the number of telephones is growing so rapidly that we are beginning to run out of telephone numbers. I don't know about you, but I've already undergone the introduction of a new area code where I live in Massachusetts and new area codes are popping up all over the country and we're beginning to run out. This raises the question of whether we're going to run out of Internet addresses---that is, those short, binary numbers known as "IP addresses." They've had to be expanded. When the Internet---then called Arpanet---was first designed, they said, "Well, let's plan for a really big computer network: we'll have up to 256 computers." Well, of course, the Internet is now at millions and still growing. And it was raised to a 32-bit address. We may run out of that---four billion addresses may not be enough. That's already not enough to number all the people in the world. Of course, not everyone has a computer yet.

We can ask the question: In 20 or 50 or 100 years, how many IP addresses, how many Internet computer addresses, will there be per household? ---just as we now can reasonably ask how many telephones per household, instead of assuming there's at most one and maybe you share one on a party line. How many IP addresses will there be per person? Of course, here I'm speaking of averages. Unfortunately, the way the world is at this point, some people will probably have more than others. It's a statistic that, currently, today in the world, approximately 1/3 of the people will never make a telephone call in their lives. They don't have access to that technology or [in some cases] perhaps they do and don't want it to some extent. It is well for us to remember that everyone here in this room is among the technologically privileged and you have access to technology and ideas that are not currently widespread throughout the world.

This still leaves the question, under this assumption: How many Internet addresses, how many computing devices, will you routinely carry on your person? is an interesting question to ask. This is currently a subject of research; in fact, I know there is research going on here [at Carnegie Mellon University] in the School of Computer Science, into wearable computers: heads-up displays with little eye-goggle things you can put in front of your eye, so you can see what the computer is displaying while walking around or carrying tools in your hands, and microphones and earphones and so forth. That's in the research stage now. When is that going to become routine? As routine as, well, eyeglasses or wristwatches are now, technological aids to our personal well-being as we walk around?

I think we need to consider, for the 21st century, what are the kinds of functionality in the way of computing technology that we are going to want to carry around with us, and what is its relationship going to be to our person and to our bodies? I got into a discussion with someone recently on this point and we concluded that you'd probably have at least five computer devices on your person--- maybe three, maybe six. The ones we counted off were [ticking off points on his fingers] (1) the ones that are embedded in your body, such as perhaps your pacemaker. You certainly want to be able to upload and download information between what's going on in there and the hospital. Probably something convenient on your wrist (2) that you can glance at unobtrusively to see information without people really being sure of whether you are paying attention to them or not. You might want (3) something a little bulkier that you can punch in information on, receive information from, so the model for that is the cell phone---maybe something like your laptop or notebook. But think about the different ways in which the sizes and the shapes of these things relate to their physical convenience and the ease of carrying them, as well as their functionality; I think {this} is going to be an important question.

I'd like to draw an analogy to the development of the technology of wristwatches, which goes back several hundred years by now. The first timepieces were, in fact, not portable. They were fairly large things. It was difficult to make small things with the technology of the time. Then, finally, very clever clock makers managed to make something, perhaps the size of your fist, that you could conveniently carry in a pocket. So they became portable. Then they got even smaller and you could carry them on your wrist, which is the customary place to carry them nowadays for many people, although not all. And once they got on your wrist, the first ones were still fairly expensive. So you had fine wristwatches. Then they got cheap. Then they got very cheap! I remember 15 years ago, when I was teaching here at Carnegie Mellon; the batteries in my watch died. I went to a jeweler to get new batteries and discovered that I could buy a new watch for cheaper than I could buy the replacement batteries for my old watch. So I bought a new watch. And now it has reached the point where watches are fun. Look at the success of the Swatch company. You can buy watches in 50 different colors, you know, neon plastic colors, with various logos, athletic logos on them, whatever. Now watches are fun as well as being functional.

I think we are going to see the same evolution with computing devices, personal computing devices you are going to carry around with you, that will assist you in whatever you do. We're already beginning to see that with the Nintendo Game Boy. It used to come in one color, this sort of off-grey. Now you can buy them in transparent plastic or neon yellow or whatever you want. Suddenly it has become not only a computing device but a "fun" device. We have reached the point where we can afford that because we know how to make them cheap.

As computers become so cheap that they can become ubiquitous, I think we need to think about how they are going to be used. In this connection, I would like to relate a story about my colleague Danny Hillis, with whom I worked at Thinking Machines.

Sometime, I guess it was 15 to 20 years ago, he went to New York and gave a talk in which he went on and on at length about how wonderful computers were going to be when they were cheap, and they would be everywhere, and they'd be small and they'd be inexpensive---and a heckler shouted out from the audience, "What in the world are you going to do with all those computers? It's not as if you want one in every doorknob!" Danny returned to New York City two years ago to give another talk---by coincidence at the same hotel!---and he noticed, when he walked into his room, that there was in fact a computer in the doorknob. And there was a computer in every doorknob in that hotel!

So what will happen when computers get cheap? Why would you want to carry computers around? Well, I think one reason is that you might want to talk to them. And this is one more theme that I think will become important in the 21st century. I think we are very close to having a practical solution to the problem of speech recognition by computers. You speak the words and the computers can, in effect, transcribe them. It could make a typescript for you if you wanted to. It knows what words you spoke and therefore can react to them.

We also need a little bit more work on speech generation. Right now, it's not unusual to run into computers that can talk and perhaps even feed back to you information that you just punched in on the telephone, whether by canned words [recordings of humans speaking] or by computer-synthesized voices. Right now, they tend to sound [speaking with a nasal flatness] a little bit like this, you know, because they have strange intonation. We haven't quite solved the problem of producing natural inflection and intonation. But I think that will come. For some purposes, it is good to have it "sound like a computer," so you can tell it's a computer you are talking to! But that's going to get better and better and I think we're going to be able to talk to our computers as well as having them talk to us.

What are the consequences of this?

I project that the need for the written word is going to decline in the 21st century. I believe that keyboarding skills will decline and become more of a specialty. I think that we will become a more oral society than in the past. You know, a large part of our society already runs orally. We do rely to a large extent on the written word, on libraries in written form, and we spend a lot of emphasis on reading and writing in the schools. And so I ask, will that change if computers get to the point where they can recognize and emit and process spoken text much more effectively? When you think about it, written text is actually a very strange encoding of information that is normally thought of as spoken, in which we carry most of our discourse, even as I am talking to you right now, rather than holding up words or projecting them on a screen.

Text does have some advantages. One is concise storage. And until a hundred years ago, when Tom Edison got going with his phonograph, there wasn't any good way of storing live sound. Now, with computers, we've got that; and with dense information storage, storing speech may be just as convenient as storing raw text. Or you can store it in textual form and reconstruct the speech on the fly, again with the advantage of computers.

Another advantage of text is for things such as diagrams for mathematics, for showing other things that have complex and nonlinear relationships. Music comes to mind. I think that specialized written notations for those purposes will still continue to be more convenient than audio forms of that information.

The third principal advantage of text is that it is easy to do random access. You can flip through a book. You can skim through the text. You can jump your eyes around from page to page. With computer-assisted searching and storage and automatic indexing, that advantage may decrease. So I think that with conversion of voice to text by computer, automated search of audio will be possible and we may simply do more of our business with the spoken word.

As further evidence of this, I would like to cite Star Trek one more time. I have studied the episodes diligently and I have found no particular evidence that Captain Kirk can read. Think about it! He talks to the computer. He asks the computer to speak information to him. When he wants anything really complicated, he asks Mr. Spock to get it, who then peers into his mysterious screen that we never really quite get to see. So he knows something that Captain Kirk doesn't. But it may be that in the kind of culture that we mythically project in that television series, the written word is relatively less important. I'm not saying that it will disappear, but it may; as a general tendency throughout the next century, we may become a more audiovisual society and more of a gestural society. And we will speak to the computer and make gestures to it. But we may become a less literate society in the strict sense of depending upon letters, upon the written word, the written form of speech.

I also note that already penmanship is in decline. My nine-year-old son is being taught penmanship in school, but they're not spending nearly as much time on it as they did with me, and certainly not as much as they did with my parents. Penmanship is in decline. Gregg Shorthand used to be widely taught but is not much used anymore. Dictation machines largely replaced the use of shorthand. Again, recording it in audio form is more convenient, and then it can be transcribed at leisure once it has been put onto a tape.

So, maybe just as now many persons carry cellular phones, perhaps in the future we will wear devices that we talk to and perhaps simply capture in recorded form everything that goes on around us. It's not all that crazy. Right now, I keep all my email. I don't delete it. Fortunately, the prices of disks are dropping faster than the amount of email I receive keeps increasing; and so, in fact, I have been able to afford to keep all of my email going back to about 1980 or so. It's currently sitting at about 100 megabytes and, fortunately, I have a good search engine to help me find that weird thing in 1987 that Bill Gosper wrote to me---and occasionally it comes in handy.

So, suppose I had a device on my person that was sort of like a glorified tape recorder, perhaps with video capture, too. And it were just sort of on all the time and sort of noticed what was going on around me and made records---perhaps beamed it by infrared link or cellular radio to a server computer that would then store my lifetime's informational transactions. Well, I've done a rough calculation, and if you assume a bit rate of about 100 KB a second---that's enough for a 16 bit audio at 44 KHz, which is good CD quality---I'm assuming I'm just trying to capture the audio; I think we can compress video down to that point.

So run through the calculation with me: 100 KB/second and there are about pi x 10^7 seconds in a year. (That is a good constant to remember: pi x 10^7. It's a coincidence, but it's a useful mnemonic.) And let's assume a generous lifetime, given today's standards, of 100 years. Multiply that all out and you get roughly pi x 10^5 gigabytes. Assuming you use today's compact disk format to store all that information, you'd get a fairly large stack of CDs, but it would still fit in a cube only ten feet on a side. So assuming only modest condensation of that information, using denser information formats, I project that in the future you could store your entire lifetime's informational transactions in your tombstone. That would be about the right size. And so we might literally be able to save everyones' lives---informationally speaking!

Now, is that useful? Once we have captured all this data, what do we do with it? Well, we're pretty good at capturing and computing data. We can store data. We can store data densely....

Oh, I think I forgot to show you this. [shows disk cartridge about 18" in diameter] This is a disk from that old IBM 1130 of mine that I used 30 years ago. It's a hard disk---you can see the platter in there. It's an ex-hard disk; it's received some rough treatment! This is a megabyte. And [displays a 3.5-inch floppy disk] today this is a megabyte, and it costs only about 50 cents instead of tens of dollars. Actually, in this same sized package, if you were willing to spend $15.00, you can get 100 megabytes [in the form of a Zip disk]. It won't be too long before it's a gigabyte, and so forth and so on. A small stack of these would fit in your tombstone.

But could anyone get at it? We're good at capturing and computing the data. We know how to store data. The problem now is access, and I think that's going to be the biggest problem in the 21st century in computing. We are now awash in data and it's hard to figure out what is important. Actually, you can say that of newspapers and television, too. They are presenting more information. You can just subscribe to your local cable and get a hundred times more television than you have time to watch. All this information coming at you. What's important? Or if you can decide what's important, how do you find it?

There are also related problems of authentication. Once you've located what purports to be real data, how do you decide whether it is any good? How do you decide whether it is reliable? How do you know if it came from the person you thought it came from? And there are also issues of privacy. How do you prevent people from having access to things that other people think they shouldn't?

This whole problem of access is going to be staring us in the face "real soon now," with the advent of the World Wide Web. This is a problem for the 21st century. It's also a problem for the next four years! And this problem can be divided into two parts. One is bandwidth. Just having located the data, how do you deliver it to the person who needs it? That's the easy part! I think the hard part is search, locating what is important to begin with.

Someone once told me that computing is the art of throwing away information. This may seem strange at first. When you realize that, if your problem is you have too much information, and you are trying to sift what is important out of it, your principal task is to throw away the part that is unimportant, condense what is important, and deliver that as the answer. Again, I go back to the example of oil exploration. The oil exploration company goes out there, gathers hundreds of magnetic tapes full of data, and they would like to distill all that to answer a simple question. Where's the oil? And that information can be represented on a small piece of paper, on a 3x5 card. Their problem is throwing away information. They've got too much and they need to distill it to what is important.

I was rather astonished when I first ran across this aphorism, that computing is the art of throwing away information, because it reminded me, irresistibly, of something that my father had told me rather early in my life. He said, "Son, life is the art of throwing away opportunities." And that seemed strange. Why would you throw away opportunities? Well, some people have lots of opportunities before them. You graduates have many opportunities before you and you are going to have to make choices about what you do and which opportunities you take. And if you take some opportunities, you are inevitably throwing away others---in fact, throwing away most of them.

Now this should not make you feel sad or put you into a panic. But just realize that this is what you are doing and it is better to make careful and conscious and thoughtful choices about your opportunities, rather than grabbing for whatever comes along.

Which opportunities will you take? Well, in this connection, I would like to tell you about my mother. Mom is great! She's a wife, homemaker, mother of two---I have one brother---grandmother of three so far. She is also a private voice teacher and I learned most of the works of Rogers & Hammerstein and of West Side Story, musicals like that, simply by being around the house while she had her students in our home, practicing those wonderful numbers from the musicals. And that's partly what led me to get into singing myself, which was constantly having that around the house. She also produced and directed dozens of amateur theatrical productions. When I had the title role in L'il Abner [a fact mentioned in the introduction to this speech], that was a bit of nepotism there! She's the one who cast me in that role and I certainly enjoyed being in it.

About the time that I was working on my undergraduate degree, she decided to go back to school and get her master's degree, after which she taught in the Brookline Public Schools. She was a music teacher for 25 years and is just retiring this year. Earlier in her life she sang with the Opera Company of Boston for 25 years; she was one of their most reliable chorus people. And she was the head of her union for several years, the American Guild of Musical Artists, and had to negotiate with the producers of the opera company. She went on tour in the opera. She got to sing at the White House, got to meet Lyndon Johnson and Hubert Humphrey twice. And, when I was very young, she had her own television show in Missouri and I can remember watching her on the TV set. "Where's mommy?" "Oh, she's on the TV over there." Sometimes she's in the kitchen and sometimes she's on the TV, you know, it didn't matter much to me.

And, recently, looking back on her life, she told me, "You know, actually, you can have it all. You just can't have it all at once!"

And so that leads me to the concluding portion of my talk, in which I would like to give the graduates advice. I would like to give you advice for having a good career in computer science or information science or applications of computers or whatever it is you eventually choose to go into.

Number one: I advise you to take the long view. Try to have some kind of plan for your life; but on the other hand, be ready for new opportunities. You'll constantly be choosing among them. Don't be frustrated if you can't get it all at once. There is, we hope, time enough to do all kinds of interesting things. Don't get frustrated if you can't do it all at once!

Number two: I would like to advise you to put your career second. I'll say it again: Put your career second! If you are married, if you have children, put them first. [applause, mostly from parents of the graduates] Maybe I don't need to hammer this home, but I've got it in my notes so I'm going to do it anyway! Friends, family, religious commitments, community: put them first. And this is not to say that you should slack off, or that you shouldn't enjoy and pursue your career, or that you shouldn't do well by your employer. But the principle is that you'll find that if you put people first, they'll support you. My wife knows with a certainty that if I ever encountered a situation where I had to make a choice between my job, or my career, and maintaining our relationship, that there wouldn't be any question. I'd quit. I'd quit the job. And because she is so confident of that, she can confidently support me in everything I try to do. Because she knows she is not competing with my career, she's working with it rather than against it--- and that's a good situation for me to be in. And, by the way, that cuts both ways. I married a woman very much like my mother; she's busy doing all kinds of things and I try to support her in that.

i find, as I look back on my life, I don't find myself saying, "You know, in 1986 I wrote a really good paper." And I don't say, "In 1975 I wrote a really great piece of microcode." (Well, actually I do! That was a really great piece of microcode! It took me three months to write 100 lines, but it was great.) But I also remember 1986 as the year my third child was born, and 1975 as the year I was reunited with a friend I had not seen in 18 years. And overall, those are more meaningful to me. The career means a lot to me, too, but people really count. So that's my advice: Put your career second.

And third: I advise you to be charitable. Try to give something of yourself to other people. Donate, volunteer. Whether it's time that you give, or money, or of your skill, or a piece of the limelight, or your just giving someone the benefit of the doubt: try to be charitable. And this is not only for the benefit of others, but also for yourself. You'll find that if you hoard all your resources to yourself, if you think you need everything you have, there will always be a little doubt in your mind as to whether it was enough, and you'll feel poor. But if you can manage to give something of yourself away---however much, however little---if you can do it cheerfully and voluntarily, then you'll realize that you really had more than enough, and you'll feel rich. And that makes a big difference in life.

At the end of this ceremony, you are going to go out into the world and you're going to have many opportunities before you. I think you're going to find it exciting. Computing is a fast moving field and I think the 21st century is going to be even more amazing than the 20th. Today, in this talk, I'm only guessing at the kinds of things that might happen. But you are the people who are going to make it happen. I like to think that I have made a few contributions to computing in my life and in other areas. I'm very much looking forward to seeing the fruits of your efforts and I'm hoping that you are going to blow me away with some of the things you do. It's going to be a lot of fun. It's going to be exciting. So I wish you congratulations, Godspeed, and have a good life!

Thank you.

SCS Alumni Home Page