Manhattan Institute for Policy Research.
search  
 
Subscribe   Subscribe   MI on Facebook Find us on Twitter Find us on Instagram      
 
 
   
 
     
 

Wriston Lecture
November 5, 1987


The Technological Vision

First Annual Walter B. Wriston Lecture In Public Policy
(unedited transcript)

Carver Mead

William Hammett—My name is Bill Hammett and I'm the president of the Manhattan Institute. We're very pleased to have everyone here at this occasion tonight. It's a dual occasion, actually—in celebration of two things. One is, of course, the first award of the Walter B. Wriston Lecture Award. The second one, which is coincidental, is that this happens as well to be the tenth anniversary of the Manhattan institute, founded in 1977. It is purely coincidental. Two at the same time, however; one had to celebrate.

What I'd like to do is to take a moment to just introduce the trustees of the Manhattan Institute because they're really the ones that are responsible for an organization like the Manhattan Institute being successful. Because they're people. These are the people who do it without any pay and they put their reputations and energy on the line and they're very important for an organization like ours. Particularly for an organization like ours which is a rather ill-defined operation which strives to work in the area of ideas, developing ideas, communicating them, putting them into circulation, which is a fairly refined pursuit and not greatly appreciated by a number of people.

I think these are very special people that we have on our board. Now I don't know where they're sitting so if I call out the name, I'd appreciate it if they would stand up. Nelson Broms ... Wright Elliott from the Chase Manhattan Bank ... Jim Evans — Jim's the retired chairman of the Union Pacific Corporation and a very advocate of the New York circle ... Peter Flanigan ... Bruce Gelb ... and not to be confused with George Gilder, we have a Richard Gilder.

Our newest trustee is here with us this evening; he's a member of our international contingent. It's Tom Griffin, who runs the GT Management Corporation. And one of our academic trustees, Lowell Harriss. He's professor emeritus of economics at Columbia University. Roger Hertog, another banker. We have three bankers on our board. Ed Hoffman of Citibank. We also have J. P. Humphreys of Joplin, Missouri. Dillard Munford of Atlanta, Georgia. Here is one of our returning members we wish we could see more of but he doesn't want to come up here to the Big Apple. We're very pleased to have you here. Thanks for coming. We have a bullish contingent; now we have the bearish contingent however. But I was told that he covered all of his shorts [laughter], Jim Rogers. And another one of our academic trustees, Ernest van den Haag. We have a Bob Wilson with us tonight, the stock market Bob Wilson, not the Frankenstein of the beach, Bo Wilson. [Laughter] He's also responsible for the choice of wine this evening. There a few additional trustees, one of them I will introduce now and he's a very special person at the Manhattan Institute. As I said earlier, the Institute is ten years old. There's only one person in this room who's been with us that entire ten years and that's the

Chairman of the Board of the Manhattan Institute, whom I've known now for almost eight years and I've never ever heard of anything but positive statements about him, and I've never heard him say anything negative about anybody. He is one of the most delightful people I know and he's been with the Manhattan Institute events and personal times. And I would like to propose a toast, if I may, to the chairman of Manhattan Institute, Charles Brunie.

And with that, I would also like to turn the podium over to Chuck, who will introduce our last unintroduced Trustee. Chuck?

George Gilder — Thank you very much, Bill. Without entrepreneurs of ideas like Bill, all of this would be left ———— and this contribution to recognize and support independent ideas that's been incredible to the success of the institute in which it exemplified once again by this board.

The special genius of Walter Wriston as an innovator did not come merely from his role in building Citibank into one of the world's greatest financial institutions. He is a great banker but all great bankers are not innovators. What made Wriston special was that before his retirement, he had the vision to transform Citibank into one of the world's leading computer companies. It was Walter Wriston, more than any other man, who led American banking into the information age.

In a similar way, the special genius of Carver Mead is not manifested in any one single breakthrough but in an entire career of constant innovation. Unlike many innovators, he did not suffer from what he calls "the hardening of the categories." He remained in the forefront of his industry for over three decades and he is still pioneering today.

In the 1950s, he built the first workable gallium arsenide transistors still used widely in  devices today. In the 1960s, he showed the way to intregrate thousands of transistors on one chip and became the fifth of Intel Corporation. It was not until the 1970s, however, that he achieved his most famous invention. When he launched a new generation of semiconductor companies based on his concept of computeraided design. This breakthrough which he called the silicon compiler aroused the complete design of computer chips on a simple workstation. Today, this invention is the leading force of innovation in the entire industry, increasing the number of designs produced annually from 89,000 to over 100,000. Thanks to Carver Mead, we now have a system of desktop publishing of computer chips. Carver has been much acclaimed for this feat and most innovators were arrested on their

But long before this breakthrough triumph, Carver was off learning neurobiology, a completely new area for him, and pioneering in the field of massively parallel neuralnetwork computers. These concepts are the now the leading edge in developments in computer architecture. Unlike the many others in the field who theorize about neural-networks, Carver has actually built several functioning devices, including an artificial retina that can outperform supercomputers in perceptual efficiency.

Mead's theorizing, on the other hand, is often focused on the economics of innovation. In the late 1970s when many were predicting the demise of entrepreneurship in semiconductors, he prophesied a new era of unprecedented opportunities in the field. And sure enough, hundreds of companies, many led by Carver's studentsover the next few years and revitalized the industry. Collectively, they now have annual sales of close to a billion dollars.

But I shouldn't leave the impression that the life of an innovator is an easy road. Most of Carver's breakthroughs were fiercely resisted by the established powers of the industries. Perhaps the roughest going came outside the laboratory when he nearly lost his life crashing inadvertently through a glass door. He emerged from the hospital to make his biggest contribution to the industry. Perhaps tonight he can offer his grounds for optimism amid the shards of the recent crash downtown. With the greatest pride and enthusiasm, we give you the first Walter Wriston Lecturer, Carver Mead of Caltech.

Carver Mead, California Institute of Technology  Thank you very much, George. It's indeed a great honor to be invited here and share with you some of my thoughts about innovation and the of especially this lecture in honor of the man who has done so much in leading the way to joining forces of the information technology in the financial world. I'm going to talk a little bit about economics and the economics of an industry which has seen so many very revolutionary ideas.

There are certain intrepidations associated with talking about economics with a story about modern economics conventions, there was an academic by the name of Professor Jones talking about his new theory of economics. He drowned on for forty five minutes or so before it was opened for questions. Someone in the back of the room said, "Professor Jones, how does your theory apply to the recent crisis in Silicon Valley?" Jones stretched his sketch a little bit and said, "Well, my theory applies more in general than an unspecific case." I'm going to air on some of the other side today. I've on the electronics industry and that's the thing I know about. I think there are lessons there that have a ity through any industry which is basically driven by large innovative steps. I'm going to share one of those ideas with you but I'm fully aware that what I have to say really applies more in a more specific case than in general.

Electronics … let me just remind you by the way of some pictures, the electronics has a need come a long way since I started practicing it back in the fifties. We used to build electronic circuits, these are all computer circuits, we used to build them with these chips ... these are the things we call active devices they are the things that actually do the computing. But in order to have computing systems, they have to be hooked together, they used to be hooked together by wires, and they can always be in this case the lowcolored elementaries of  it takes all of that stuff and some way to support it all to make a computer system. Now I'm only showing just the computer system. These things were assembled in a and they took up lots of power. Little vacuum tubes … gave up a lot of heat and that was the technology in the computer world when 1'entered the

And there's been a number of in innovation that changed the nature of how we do computation. And by computation, I mean the very, very general sense of information processing. I don't mean any particular kind of computation. The first big step is, you see here, something that doesn't get much. It was very tedious  these wires and resistors together in that technology. The first innovation came by way of putting things more easily. This was a thing called circuit board. It's an insulated piece of fiberglass with a on each side. The copper was patterned to the  process, a simple printing process that was used in printing for so many years. So that the conductors in the conducting that was left was able to connect up with in the proper way. In that way, you always got the right connections. People didn't make any mistakes and came up in the wrong way in the wrong tunnel. And it was much cheaper and more effective. That was a step that happened while we still used vacuum tubes for the computing circuit.

The next step was an even larger step that happened in 1947 when the invention was transistor and for vacuum tubes. And you probably noticed that the characteristic of any conventional sort of technology that a new step is used first for the thing that people already know how to do and the way that they already know how to do it. So you know there's ' a surprising similarity between this vacuum tube circuit and this transistor circuit. They're put together in the same way and used in the same technology support. That's why all this is true and we can't think of more than one idea at a time. That's not smart enough.

Well, transistors began to dominate all electronics. Transistor radios to the largest computers do so today. And there's another enormous step shown here that you can't see because I have things on the wrong scale. If you look inside at one of the of the transistor, you see a transistor which is now I've increased the scale here ... so the distance across this little piece of silicon is about the size of a match. So we'll magnify everything else so we can see it. I'll keep doing that for a while so that we'll be able to see down inside of the technology that we're talking about. And this is the way we used to make transistors. We would make a whole bunch of them, it's like printing stamps. We'd go through the processing steps. In those days, they were about two inches in diameter and flattened like a piece of cardboard. And we'd make hundreds of transistors on one piece of silicon and then we'd beautifully cut them up into these little squares the size of a match and we'd glue them down on those little cans that had three little feet coming out and then glue those cans down on these  circuit boards. They all seem to be a little interesting. It's a little like the progress … since they discovered gold in California and took a hundred years to the  the gold in California and carry it back to Fort Knox. That kind of progress was going on. A friend of mine by the name of Rob Rice noticed that it was a little silly because what we were doing... they were putting the transistors down on this layer of insulating material that had on it then it was carved into stripes of a shape that would connect up the transistors properly. Well, when you looked at the way the transistors were made, they were made on an insulated piece of stuff and they had  on it. This white stuff is  He asked a very, very obvious question. He said, "Why couldn't we use the aluminum that we already have there to hook up the transistors we already have there and I have to cut them all apart and then put them all back down and put them on another piece of stuff?" It's such a stupid question that it was one of the great inventions of the decade called the innovated circuit. And, in fact, it has changed in a very, very fundamental way the whole nature of what's going on in today's world and it's the basic invention of the information age. And it came from a very, very stupid question about something we were doing that was even more stupid.

Well, then what happened is year by year, shown on this slide this is the missing link this is the first integrated circuit in the of 1959. It was a very simple launching circuit equivalent to what would take a couple of those vacuum chips. This is a much bigger than a match, a little bit bigger. Of course, it's a little... giving technology like this is like giving up the whole Eskimos a little bit in forest  and people starting to . And first of all, they decided to do a small circuit than do a bigger circuit and maybe a bigger circuit. And these are about oneyear intervals. These slides are  about the of Intel Corporation, good friends of mine. It's really nice to have a feel for all the people of the original are all ing to each other still and moving fast enough, if that's still possible. It's really a nice kind of a feel that way.

And you noticed an interesting thing has happened in here just by looking at it without going in deal whether what circuits are and what they do. You can see that people are learning to make more complex things to use a little more silicon to do it. But mostly, we're getting better at putting more stuff and packing them better on the silicon hand   here a small semiconductor memory, a very early one, is not a whole lot bigger than the individual transistor that came ten years or eight years before. So this is the beginning of a revolution that none of us saw when it started. And its characteristic of these inventions that you don't know. I was just working in the new kinds of transistors that  And when the integrated circuit came along, and Bob Rice told me about it, I said, "That's interesting. All he's done is just connect together some transistors, a reasonably trivial thing." None of these inventions at the time they were made were obviously going to transform the world. When the transistors were invented and introduced to the marketplace, I remember on many engineers walls there was a cartoon that showed that if any of you can get your hands on one, you just got to have one  But what's the along on big long legs is chasing us, little tiny transistors trying to scurry away from it. And the caption said, "Help Stamp Out Transistors." That was the engineers' reception to the biggest advance of our time in the electronics business. So none of these things were recognize universally and instantly as the great inventions they turned out to be. And it was in their evolution of the real world that the true values have

Today we've gone all beyond the sixties kind of  on chips. And if you look down on one of the things in your personal computer, you'll find memory chips like this that have millions bit of information stored on it. You'll find microprocessors like this that have a whole  of computers built on one chip on half of the transistors. You can't  the individual transistors on a circuit like I said, there's close to a million of them here. You look down close at the technology it's a very beautiful technology. It's mesmerizing for those You notice that it's a multilayered technology. The metal that Bob Rice noticed was there already to be used to hook up the transistors ... see here the shiny stuff. The little bundle looking things are places where the metal goes down through the insulators and makes contact to the transistors underneath. The transistors themselves are formed by one of these lescent green areas processed all the way by the shady stuff underneath. So-called  transistors that are the most common technology today.

In today's world, the smallest dimensions of a feature on one of these things are  millions of them here. Hundreds of these take up a human hair  visible light depending on how you like to measure things. One of the enormous driving forces behind the old information revolution is in the evolution of this technology, the fact that the technologists have learned to make the same basic technology better. That's been a straight evolutionary trend. The dimensions have gotten smaller by a factor of 100 since we started doing it and they're still going. We still have quite a way to go before we get to the ultimately small device where if you make it any smaller  ng. What you do with one of these chips is, of course, they don't look like much if you're not used to looking at them. There's one right there. They're put in these things called packages at the box, these ceramic putties there, the little caterpillars that crawl on in 16 bytes. These things are called packages and chips are put in one of the packages, the little wire to them, the actual connections on the chip to the outside world. Then you take one of these devices here, that's a computer that has a lot more capability than the mainframes that we saw on centers that are still existing in large like the ones in These things are all completely obsolete a few years, not even  for ers.

I'm starting to make these  of the complexity of the devices that we are manufacturing  time. I was very taken by this. There were these trends  that Bill has started doing this shortly after he invented or had a hand in the transistor, and started doing it. And I was the young  and very abstract by these  and they started out with such an obvious trend. But that's part of the complexity and you'll notice that's going up in many, many orders of making  This curve shows an evolution over a period of years to a factor of a million in the complexity of a long time.

Basically, the cost of a chip today is about the same as the cost of one of those individual transistors when they were being sold in the  on the circuit boards that I showed you. So the cost for chips has really not changed all through the years in real dollars. The capability represented by each of those chips has gone up by a factor of more than a million since the 1950s. And this is one of the obviously the most remarkable evolution in the history of mankind. Just to  you, this is the factor of a million.

The whole industrial Revolution which substituted fossil fuel for human power and animal power and gave us  and urban waste and all the other good things about modern society. In any of the important capabilities like getting from here to the West Coast or pretty close or any of those things, it is about a factor of 100. We've already gone over a factor of a million and the informationbased technology and the end isn't in sight yet. This is a revolution of the many, many much, much larger magnitudes than the industrial Revolution itself  happened as revolutions go overnight, this sort of exponential growth has come to be known as  the fact that the capability of the technology that double every year or else would go down a little bit. This slide is one of the slides and it shows a number of ... that one is more of the technology every year. The other one is called Its one space in the future lasts scarcely passed the last .

 Fortunately, the Japanese came along and helped pop up this curve up there so they've been otherwise it might be drooping even more than that. It's a marvelous phenomenon, unprecedented in any kind of technological evolution. The only place you see exponential explosions like this are in things like rat population. This is really remarkable. I've stood in all of this for many years, being in and around the industry. And  many people for many years for how neat this One of the people I was ... You know that not all is well that about and you've probably heard that we heard that at lunch today about people in Washington  One thing led to another. And it seems  that an industry that had offered the most amazing revolution in the history was at the doorsteps of the bureaucrats the minute they were going down . Didn't seem quite right. All of that sort of cried out for some kind of model of what is going on. What is it that takes us from an industry that's conquering the worlds to an industry that's at the doorstep at Washington in a couple of decades. There's something funny about all that that we don't understand. One of the people I talked to about all of the things that were going on was Tom Perkinson of Venture Capital out on the West Coast. Tom is a Harvard man, and he said, "Oh, Carver, you think very remarkable. That's just a  curve." George Gilder tells me Bruce Henderson of the Boston Consulting Group had really championed this idea of the experience curve the fact that in many industry or service industries of that matter people get better at what they're doing. And if you keep doing it longer, you get better at it and then in general if you plot the cost of doing whatever it is you're doing as a function of an unlogarithm of how many times you've done it. You get something like a straight line this particular straight line is something like a cuberoot kind of log variable after the slope they always look kind of like that. Say you're looking at it just for another learning purpose. I looked at this idea for a while and somehow I did. There's a problem on Mr. Henderson's experience curve and the decrease in cost for production of goods or services with the accumulated experience. Somehow, I didn't seem to be enraptured of all what was going on in this. And it's sort of easy to get lost on this curve if you don't know quite very  Feature looks landscaped. It didn't seem to quite capture for me, an industry which was punctuated by really very strong landmarks, the transistor and the integrated circuit and the microprocessor and so forth. In fact, with surprising regularity, we've had very surprising innovation in this industry. It hasn't stopped, it hasn't slowed down.

Once again, this is a phenomenon that needs some break in some way to conceptualize what it is that's going on. And Henderson's  didn't help very much. Morris Law really only speaks to the evolution of a manufacturing technology. Doesn't really talk about innovation and where it comes from and how it affects the economics. I have always been a real fan of C. Northcote Parkinson, who wrote his wonderful book called PARKINSON'S LAW, which has a lot to say on large part about institutions. You probably all know Parkinson's Law. His first law says that work expands to fill the time available under a penny. Highly recommended book. The thing that I most appreciated about Parkinson was that he used the article I called constructive oversimplification that by oversimplifying to the point where the basic ideas was clear, it can say something that it was really harder to say if you really wanted to put all the is, ands and buts. I decided I kept a at trying to summarize what was going on in innovative industry in a set of little  that I largely called Mead's Laws. So I would like to share it with you. I think about these and once again these are in the spirit of Parkinson's Law, they're acts of what I hope to be constructive oversimplification of what's actually the extremely complicated thing that involves human errors, greed, generosity, and ideals and the whole way society is put together and all that which obviously can't be captured in a set of simple statements. It'll be very oversimplifying instead of .

Let me start by saying what happens when there's a great invention. What happens when the transistors, what happens if the innovated circuit isn't  There's always an older way of doing things. Populations used to be done by people pushing numbers around a piece of paper with pencils. There's some value associated with doing a particular operation. As you can do it in this case electronically, it has some value. Initially, when you learn how to do something a new way and the new way is much more effective than the old one, the cost of doing it the new way is not related to the price you can do it the new way. If the cost of doing it the old way was up here somewhere, that sets a value on doing the operation. And if the learning curve, the Henderson Law, down here has to do with the cost of doing it that way, now you have some LEEWAY (spelling). And what's a great  is that things are priced in a  according to their value. And there is a considerable profit to be made by charging more than it costs to make things. That's the principle of capitalism. Eventually, of course, competition comes in and everybody learns how to do it the new way and pretty soon you have everyone doing it the new way. And you can price things only slightly above cost in order to get some return on your capital investments. So out here, this is business as usual, the view about economics that come from a steady state period, say, you're sliding down some   curve, and everybody's learning how to do the thing by producing more stuff. The cost goes down, therefore, the price goes down. And that's certainly true in the steady state if nobody makes a new pension. But every time there's one of these new inventions, you have this opportunity to recoup something for all the energy and time  you put in to making the invention and causing it to  So this is what I call the return on innovation. Walter Wriston has written a wonderful passage in his book, RISK AND OTHER FOURLETTER WORDS, where he talks about intellectual capital formation. You can think of this area as a shaded area between the two curves here as a return on intellectual capital. out here we're return only money capital associated with production. In a statement, the opportunity lies in the tran down on the citystate. And when things are sliding along perfectly, predictably, there's not a lot of opportunity. You can go in and you can perfectly well and U.S. companies can  the best of the foreign companies in a well-defined and predictable industry. But the big opportunities come with trans and there's a whole new way of doing something. And the others haven't figured it out yet. So that's the first observation.

The second observation is on caution. It looks like we'll be making creative inventions everyday. I'm reminded of the cartoon in Chief Executive that has the scientist on the  in front of his desk and saying, "Well, Dobson, how long has it been since you made your scientific invention?" Those don't happen because somebody decides they're going to happen this week. They happen because there's been a great insight and we never know where this is going to be or where it's from. There are many insights that turn out not to pay off enough in the cost of common

The second thing I have to say about innovation is that it doesn't always work. I've noticed investment people talk to me all the time. They always have some scheme that someone's come to them with for revolutionizing this side or the others. Many of those schemes.... Well, in fact, they may be technically sound. They don't have what I call "head room." if you're going to do things a new way, it better have a learning curve that's well below the learning curve for doing things the old way. Otherwise, you're going to get behind the par curve. You're never going to be able to produce enough of them to get down the learning curve nearly as far as the people I  the old way. It's like trying to replace the internal combustion engine. It's the reason trouble memories ever made.

There's a whole lot of examples and technology in something that might in theory have been better. But, in fact, you can never get nearly as far enough down the learning curve to cross over and get below the old way of doing it. Companies doing it the old way spend their money a lot better by evolving the old way into gold than take on a risk on new ways of doing things. So if you're going to do something by way of innovation, it better have some head room. It better really make a difference and my rule of thumb is it better be a factor of  if some things are a factor of two better and at the same point the learning curve, ... you're probably never going to get there because you'll never get accumulated  into the old way of doing it. So that's the The third thing I have to say is a lot more exciting and that is an industry like ours who are ... there are unpredictable, but surprisingly regular major inventions. The transistors, the integrated circuit, the microprocessor, so forth. There's a thing called the composite learning curve. Any one of these learning curves will be coming on here on a slope like this. And companies that are stuck on one of those learning curves will be reducing their costs according to their learning curves. But going from technology to the next one, and following the innovations that come, this curve is much, much deeper. And I've drawn these curves roughly to scale, being the accumulated volume that elements are built and the cost per logic element is not quite as  because of what we mean by  But it's not too bad. And you'll notice this is really an extraordinary important curve and it's much deeper exponentially deeper than any of the individual manufacturing margin curves. So this is really the bottom line in terms of public policy that ... There's the old saying, "We don't believe in miracles, we depend on them." I think that's true, certainly true in electronics industry. And in my view, it's to a large extent true of the economy that we depend on miracles. We depend on the innovations of the citizens of a free economy to keep ahead of the bureaucrats and the people who make a living on control and planning and those kinds of things. It's the element of surprise that gives us in the long term our over much more controlled times of You can see it in your action in the electronics industry  I think the same thing would be true in any of the industries who are driven by the intellectual insights that make whole new ways of doing things possible. This is the basis about my optimism about the future, that we are in an economy that is driven almost totally by the very large kinds of innovation made by individuals and brought to fruition by the dedication of the entrepreneurial spirit and the people who bring these things up and cause them to be real.

I've written down just a few statements about these things. I should point that there's a fourth thing I have to say. Does breakthrough technology come from a direction not foreseen by the existing industry as predicted by the  ? Of course, this will be seen by a lot of people in this room as an amateur taking a  for professional. That's not the way it's meant. What I mean by that is the breakthrough technology is by its very definition not part of the existing culture that's established around the companies. Otherwise, it would be part of their learning curves. Those learning curves, after all, may not pop innovations of their own; otherwise, they wouldn't keep going on. It is one of those innovations that would be part of the Henderson Curve. So, in fact, by the very definitions these curves come out left field. And I can tell you from having lived through them how they look. The integrated circuit was a really good example where all of us in the gain at that time were trying to invent new kinds of transistors because that was the name of the game, right. And all  was just to figure out how to hook them up. Very interesting. They only revolutionized the whole industry, that's all. And we couldn't see it. I stood right there and looked at it. It was not interesting. I've been through enough of those to know that I can look right out of my hand absolutely and not see it. And I'm sure that some of you in this room are capable of people behavior. It's just hard to see when it takes a different way of looking.

There's a correlant to that last thing which new technology are adopted last by the companies that needed the most. That's because they're not part of the culture that  the companies. Therefore, they're not part of the Henderson Curve. Therefore, they won't be seen because they're contrary to the way of thinking that's been successful in the past. For example, you probably noticed, if you've watched, that the microprocessor… the last kind of company to use the microprocessor was a computer company. In fact, there had to be a whole new industry grow up around what we now call personal computers. It was not generated in computer companies. Then the computer company said, "Oh gee, that's interesting. Maybe there is something I want to do something about it." They were the last ones. And before they even got there, microprocessors were built into electric drills and microwave ovens and every other thing in the world. They weren't used by computer companies. And I can tell you that in today's world the last companies to use the silicon conductors are the semiconductor companies. They needed desperately but it's contrary to their culture. So in fact, they're not doing it or only doing it very late. And that's not because they're bad people. It's because when you built the culture and you get successful in a culture, it's very, very hard to see things in a new way. Fortunately, we have entrepreneurial systems. And the way we get new things to happen in our economy is to let people start from scratch. And by the time they're successful, they contributed a whole lot more to our economy than they ever get back for it. And they often lead the way for larger companies to come along and clean up afterward.

There's another one which I just have to mention here because it is a financial community after all. And that is that you're going to see the founding of lots and lots of startups and most of them are what I call  Now a lot of those will make money and it will be okay. But having been through a number of people to start out on really new ideas, I can tell you those are the hardest ones to get any way to  I was fortunate to stand next to somebody today who said, "I can only put money into things where there were intellectual properties were there." That's a really unlikely view that's actually valued intellectual capital And perhaps  obvious. Now what's a hint? This is an exciting game  The industry is now affecting the lives of everyone, whereas twenty years ago, he said electronics didn't mean much to most people. In today's world, we really are in the information age. We should look ahead and see what's happening.

We just sort of  the big steps that have happened in the last.... There's sort of, I say twelve years but actually it's about thirteen but it didn't seem like a good number to put there so I lied, we're right here now. The microprocessor has had its now in the personal computer business. Let's just review a little on what happened there, what was the value that microprocessors contributed? The most obvious kind of incarnation is in the personal computer and there are people that are in business making personal computers. In fact, that's what George Gilder said at lunch of United States holds the lion's share of the hardware market in personal computers. It seems a little strange as they are being clumped all over the world. We're doing perfectly well.

But in terms of the benefit to society, I think that can be seen as a minor player. The big benefit to society has been its  of common by which applications can be worked. For the PC, the personal computer has really given us as it unleashed its enormous way of innovation in software   the individual applications into a machine where it can get worked, the game we now call software. The PC really was just an enabling technology to allow this way of innovation which gave us this attention of a lot of bright, innovative people on individual application. That is immeasurable effect on the economy. It doesn't show up in the that people have but in terms of productivity, it's a  to do things that we just couldn't

I can tell you what the result was. At the time we were doing some simulations on our backs, which is sort of a standard mediumsized computer. It took about 600 times real times. Took about ten minutes to compute one segment of sound on perhaps the whole computer to  In a chip that was  in architecture that was crafted specifically for this task, we could cause about one instrument to one chip at real time. So it's about 600  per chip. For this application, what's the general purpose This illustrates that even using a digital meta a specialized one  very tasks, you get a very easily ... the of course has a very large number of chips in them so we were getting something like 10,000 or 100,000fold improvement in terms of production of useful computation per chip over the general purpose. There's not at all usual and it's not particularly related to this application. This is why I just happen to know.

There are areas which are even more demanding than this one. Just remind you that even the stupidest animals can recognize things and an image in real time with neurons that compute a millisecond in a time range. And there are most powerful supercomputers take an hour to analyze a scene and don't do nearly as well. So if you calculate that back, it's a factor of at least to It's a ten billionfold more computingly need than our biggest supercomputers to do the simplest task, the simplest visual systemaI of any vertebrate animal can do. It must be doing something right with all of those slippy, gooey neurons. It's simply not true the law that surrounds that there's something horribly slow and efficient about the nervous system. Please don't people say that ask them to wash out their mouths with soap.

It's really not true at all. The nervous system is an awesome computational instrument. The eye is only the most familiar example of human beings being primarily visual animals tend to see things that way. We talk about people being visionary, about having insight into certain things. There really is a visual metaphoric hearing commonly for our deepest forms of thoughts, If you look at the cross section of the eye, the  up here in the lens and so forth we know about those  cameras all the time, we don't need to do that. The business, then, is down here in the retina. It wraps around the inside of the eyeball. It has a lot of neurons in it to do a lot of computation right up front, right where the light falls on the If you take a cross section through that very, very thin slice and blow it up in an electron microscope, you see something like this. A of receptors that take the light in and convert it into electrical signals. And then there's electrical processing elements here, these neurons, the little junctions are sophisticated transistorlike devices. The little junctions here are even more sophisticated transistorlike devices and the neurons themselves are like wires and contain the metabolic machinery which we would call  for building electronic stuff.

One of the things that strikes you as you study the nervous system is that the principles upon which it works are completely different from many thinly known individual computers. And the more you study it, the more you realize that there's a computational metaphor here that's really very, very deep. Every time the figured out how something was done and why it's done that way, we get an awesome insight in the way we do computation. And then leaves an open question: why don't we learn enough that we can actually make things that compute this way? This is a very, very fascinating endeavor. I personally believe that building silicon chips that compute in the way that the nervous system computes is going to be the next really fundamental step in electronics. I can show you a very, very simple, very rudimentary, baby step in that direction that's happened over the last couple of years. if you look down on any nervous system through the retina or the cortex, you see a mass of knots, the little trees of wires that run out or have little transistor-like junctions on the ends. You can by carefully studying the very simple structures of the nervous system, you can get a hint of just how they work. Then you can go try to embed those principles in a silicon circuit. And this is the first one we did that way when we tried to build the very processing step that goes on in for this is just taken off the screen. The workstation that we used to design this chip, these are the photoreceptors. They function very much like the photoreceptors in the nervous system. These are some of the neurons that horizontally in that picture I showed you and these are some of the things that correspond with the neurons that connect critically with the system. The chip that does this is built with a standard silicon technology. It looks an awful lot like any chip at this scale. You can't why chips are workstations. In fact, this chip does a fantastic amount of computation. As George said earlier, it does computation at the level that can't be done by a supercomputer. I think it's only the beginning of a whole new way that people are going to do innovation and computation. And once again, it's from a direction that's completely different than we would have expected following the digital paradigm that's been so successful so far. Now none of us can see ahead very far because it's really hard and the gets very thick as you go out very far. But I think I can report to you that, hanging out in Silicon Valley there's a lot of innovation going on in the electronics industry. We're not going to need the federal government to come in and bail out all of our electronics. We're going to do just fine, thank you. Its much more innovation creativity in this business as I have ever seen and there's a lot of place to go in the future. Thank you. (Applause!!!)

 

 
     
 
 

The Manhattan Institute, a 501(c)(3), is a think tank whose mission is to develop and disseminate new ideas
that foster greater economic choice and individual responsibility.

Copyright © 2014 Manhattan Institute for Policy Research, Inc. All rights reserved.

52 Vanderbilt Avenue, New York, N.Y. 10017
phone (212) 599-7000 / fax (212) 599-3494