THE FORUM ON TECHNOLOGY & INNOVATION

+ + + + +

COMPUTER EXPORT CONTROLS:

PROTECTING OUR SECURITY

OR

HARMING OUR GLOBAL COMPETITIVENESS?

+ + + + +

THURSDAY

JUNE 23, 1999

+ + + + +

This transcript was produced from a tape provided by the Council on Competitiveness.

C-O-N-T-E-N-T-S

Opening Remarks and Introductions

Mr. Rooney 3

Senator Frist 4

Senator Rockefeller 7

Featured Speakers

Richard Perle 10

Andrew Grove 19

William Archey 28

William Reinsch 35

Roundtable Discussion and Q & A 46

P-R-O-C-E-E-D-I-N-G-S

(12:30 p.m.)

MR. ROONEY: I just want to tell you quickly about our format today.

Senator Frist is going to briefly outline the topic for you. Senator Rockefeller is going to introduce the speakers. Each of our speakers will come up here and talk to you for about eight to ten minutes telling you their view of the computer export control issue.

Then the two Senators are going to throw the floor open to you for your questions and moderate the ensuing discussion.

We have two floor mics here on the floor. We'd like you to come up and use the mics and address the panel. We ask our guests here to give congressional staff priority at the mics. We're here and the Senators are here for them and we'd like to hear from them.

For you staff members, I ask that you do come up and use the mic. We have at least two members of our panel who are former Senate staffers and they didn't get where they are today by being shy or retiring. So this is your opportunity to step forward.

If you're congenitally shy we do have green question cards that are in our packet. You can fill those out. Please hold them up in the air. That's the only way we know that you've got questions. We have staff who will circulate among the tables, pick those up from you, and the Senators will ask the questions on your behalf.

When you do come to the mic, we give our Congressional staff the privilege of anonymity, but for the rest of you we ask that you identify yourselves. It helps the panel understand where you're coming from and do a better job of addressing your questions.

So I turn it over to Senator Frist.

SENATOR FRIST: Peter, thank you and let me welcome everybody today to what will be I think, just a truly exciting and educational discussion over the next hour-and-a-half. I want to welcome all of our panelists today and as usual, we're going to keep things moving right along, stay on time, get people out on time, and as you can see, we're starting right on time as well.

Very briefly, as for a general overview of our topic today which is "Computer Export Controls", all of us know this is an extremely complex topic which entails striking just the right balance between protecting our nation's ability to compete in the global market and at the same time, preserving our national security.

Most of us have read the Cox Report from several weeks ago which highlights some of the inherent dangers of exporting technology to nations with advanced military capabilities that one day could pose a threat to the United States or its allies.

Specifically, Chapter 3 of the Cox Report documents several instances of possible diversions of U.S. High Performance Computers to the Chinese military. These computers could be used to design more advanced missiles, nuclear weapons, or Stealth aircraft.

Now, the computer industry as we will hear today, would like to see current export regulations revised to reflect the increased processing power of consumer desktop machines. Over the last several weeks   --   and a lot of discussion went on in Washington over the last two weeks and especially last week   --   we all heard the story of the Sony Play Station II which is a video game player that has an Intel processor that is so powerful that it would be classified as a supercomputer for export purposes and thus would be banned from U.S. sale to military end-users in India, in China, and a host of other countries.

Computers are a dual-use technology and there's a natural tension between the security needs of society on the one hand, and the commercial interests of its supplier on the other. Is it really two hands or is it merged together?

Two attributes of computers do make them unique. First is the rapid pace of technological development where the half-life is shorter and shorter it seems, each day; all of which tends to render regulations based on processing power obsolete on roughly a 2-year time scale   --   a time scale that some people say is becoming shorter and shorter.

Moore's Law, the law that says that the power of microprocessors doubles every 18 months has held true as everyone knows, for 30 years and is likely to continue. You can buy an Apple iMAC that today is several times faster than the computer technology in the radar evading F-117 Stealth Fighters. Again, just to demonstrate the speed with which technology is developing.

Computers are also unique because they're part of the information technology revolution; that is, catalyzing, supporting, energizing the change which pervades just about everything in our lives   --   from my own field of medicine and technology and scientific advances down to everyday affairs, whether it's purchasing vitamins over the internet or the latest book purchases in e-commerce.

Computer and networking technology dramatically lower barriers to a free-flow of information   --   something very exciting to all of us as we look forward to the future and future economic growth in this country.

The challenge for Congress, the challenge for the Administration, is to find just the right balance between the critical national security requirements and export competitiveness for one of our most lucrative and important industries.

Today we have a first-class group of speakers and I'm very excited to more formally hear their introductions and dive right into the presentation. Senator Rockefeller.

SENATOR ROCKEFELLER: Thank you, Senator Frist, and once again I would remind you that Bill Frist and I and this forum, take no side, and have no point of view. What we're here to do is encourage debate between the panelists, questions from you, debate between you and the panelist, all of that. But we don't take a point of view and that's I think, part of the reason why you can trust this forum, as I know you do.

I'm going to introduce all four speakers now   --   I don't like doing that, but that's what I'm going to have to do anyway. The first is Richard Perle   --   and I'm going to deal with legends and semi-legends in two categories here.

Richard Perle is in the legend category. In national security circles he's a legend, having served as Assistant Secretary of Defense for International Security Policy during the Reagan Administration. He was the most articulate advocate of a forceful approach towards the Soviet Union.

He currently leads a high-powered group of former senior officials that monitored U.S. foreign and defense policy, including obviously, technology transfer and export policy. So he will be the first speaker.

The second speaker is already listening to us and is with us, although not physically. Actually, I thought he was going to be in London where he was, but it turns out that he's in California. So just like the evolution of technology he moves faster than the speed of light, and that is Andy Grove who also very much fits into the legendary category.

And he was Man of the Year by Time magazine in 1997. Neither Bill Frist nor I have achieved that to this point. He's co-founder, president, CEO, and now chairman of Intel, as you know. He built Intel into the world's largest semiconductor manufacturer and an American powerhouse.

He's authored a lot of books, including widely used textbooks on the physics of semiconductor devices. His latest work which is called, Only the Paranoid Survive, may sound like either the United States Senate or a political science textbook, but it's actually a fascinating treatise on business management in the digital age.

I really want to thank Dr. Grove for being with us via satellite, and even more   --   not only is he here with us as I talk, but also Intel paid for the satellite, and we at the forum love that.

Thirdly, and I now go to the about to be legendary but not quite there yet category, and Bill Archey. He's very close to it. President and CEO of the American Electronics Association which is the largest association of computer and electronic manufacturers, as you know.

Prior to joining AEA he was Senior Vice President of the U.S. Chamber of Commerce. He has also seen the export issue from the other side of the table, having served for a time as a senior export control official in the Reagan Administration.

The fourth near-legend is Bill Reinsch, who I happen to know very well. He's eating at the moment. He's Under Secretary of Commerce for Export Administration where he administers and enforces   --   in other words, he is the person who does this for the U.S. Government   --   this whole question of export control policy.

I am an enormous fan of Bill Reinsch. He worked for many years for John Heinz and then I was very honored to have him work with me where he did trade, economic affairs, foreign policy, defense policy, and a whole lot of other things. So we're very happy that Bill is here.

And each of the speakers will speak for about ten minutes and then it's your turn, and I encourage aggressivity on both green cards and all presentations on the microphone. Bill Frist and I will moderate and we start with Richard Perle.

MR. PERLE: Well, thank you Senator. Thank you for that generous introduction. It's not every day that I'm called legendary. It was a good introduction. I was going to say that it was a good introduction but not the best introduction I've ever heard.

The best introduction I've ever heard was an introduction of my old boss at the Pentagon, Cap Weinberger, who was also Secretary of Health, Education, and Welfare in the Nixon Administration. And in that capacity he dedicated the Home for Unwed Mothers in Washington.

And on the occasion of the dedication he was introduced by one of the unwed mothers who wound up an effusive comment on his career by saying, "And so Mr. Secretary, we want to thank you. If it were not for you, none of us would be here". That's my idea of a great introduction.

In 1981, shortly after going to the Department of Defense with some responsibilities for export controls, a young officer walked into my office with a briefcase, dramatically double-locked. He unlocked the briefcase and invited me to read a document which he would then have to take away with him.

It was an extraordinary document. It was the product of a Russian official having risked his life to make it available to us, and it was the Soviet blueprint for the acquisition of Western technology intended to fuel the development of their advanced military forces.

And it identified over 5,000 separate military programs then underway in the Soviet Union that benefitted from Western technology. And it laid out in great detail the sources of the technology, the organization assigned to acquiring it. It was everything you wanted to know about what the Soviets were then up to.

The response of the Reagan Administration was to tighten significantly, controls on the export of advanced technology. With the knowledge that we then had with the use they intended to make of it, it seemed a perfectly natural thing to do. At the time we couldn't talk about why we were motivated as we were, the information was so guarded and so sensitive.

Just a footnote to history: unfortunately the man who supplied the information was ultimately caught and executed. So we put in place in the early '80s a stringent   --   or as stringent as we could make it   --   effort to control the flow of technology useful for Soviet military purposes.

Well, that was then and this is now. The Cold War is over. It ended with a Western victory that none of us could have anticipated; at least not in the timeframe in which it occurred.

I think we can look back at the conduct of the Cold War with some confidence   --   based on commentary coming even today from the former Soviet Union   --   with some confidence that the effort to restrain Soviet access to our technology played an important role in hastening the demise of the Soviet Union.

We still face national security threats and challenges. It's not so easy to identify the precise nature of the threats we face, but if human history has taught us anything it's that it is better to be strong than weak; better to be technologically advanced than technology backward, particularly where national security is involved.

And therefore, it is essential in my view that we continue carefully and in a balanced and measured way, to try to protect against the use of our advanced technology against us by those who wish us ill.

At the same time, our own security and our own economy is increasingly dependent on vigorous technological innovation and technological growth. So we must not attempt to control the flow of technology in a way that inhibits that vital element in our national success and our national security.

And since the days of the Cold War   --   in fact since roughly the middle of the Cold War   --   an important change has taken place. In the 1950s and '60s and even into the '70s, the military establishment, the military services, and military research programs in the Department of Defense, were the leading edge of technology.

That situation has long since been reversed. It is the civilian sector that is now the leading edge of technology. Andy Grove is doing more to promote technology that is vital for our national security than anyone in the Department of Defense, I daresay.

And so the balance of technology has shifted from predominantly military to predominantly civilian and indeed, if we were to do anything of significance to improve our security going forward, it would be to incorporate advanced technology much faster in our military systems than we now do.

The fact is that our civilian systems typically rely upon far more advanced technology than our military systems. It takes so long to procure them that by the time they're in the field the technology that underlies them is frequently antiquated.

The case of technology at the same time, has accelerated and I believe, will continue to accelerate. So what does this mean for the topic of today's discussions which is, how do we deal with the control of computers in relation to our national security?

I came to the conclusion some time ago that attempting to control raw computing power   --   and in particular, raw computing power that is resident in equipment that is distributed on a mass basis   --   is not feasible or effective.

And an effort to control raw computing power, an effort to control the export of PCs   --   even those that rise above the MTOPS levels that are now enshrined in the existing control regime   --   those efforts are self-defeating; will undermine an important market for the United States, and the growth and development in future of one of America's most important industries   --   important for national security and economic well-being as well.

It's the wrong way to try to achieve a result, even though the result which is the restraint of technological development among our potential enemies is a highly desirable result. In my view, we should be looking for new approaches to solving the problem; where the problem is appropriately defined as restraining the acquisition of technologies vital for military purposes by our enemies.

One way to do that is to focus much more closely on what it is that our enemies and potential enemies are attempting to achieve. And this is a far more intensive, manpower-intensive approach to control than a routine process like the granting or not granting of export licenses. I think we do that in part because it's easy. You can set up a regime, you can define a limit, you can assign people reviewing when licenses should and shouldn't be issued.

And if it fails to solve the problem, at least you've done something that's demonstrable. You can collect statistics on the number of licenses that you processed. I think we've reached the point where we're processing licenses often for the sake of processing licenses, and without any significant impact on our national security.

I would much rather see the resources now devoted to that ineffective effort, devoted to the collection of intelligence and the effective mounting of operations against proliferating states and others who are seeking to acquire sophisticated technology in order to fuel weapons programs and programs for the development of weapons of mass destruction.

And that is far better done in my view, by a more robust collection of intelligence and by intelligence operations aimed at disrupting the flow of those critical technologies. And the critical technologies in my view do not include and cannot reasonably be expected to include, things like the kinds of computers that you'll find on my desk or your desk. An effort to control those will ultimately prove unavailing.

There is, even with respect to the more orthodox system of controls now in place, an idea I'd like to throw out   --   and it's the last point I'll make this morning   --   I think there's a strong case to be made for improving the process by which export licenses are issued for those commodities that are truly sensitive and whose acquisition by countries or groups hostile to the United States would prove damaging to our national security.

And what I have in mind is a partial privatization of that process. And it would work something like this. Bill Reinsch has to sign lots of licenses, or his subordinates do, every day. Licensing officers have a modest amount of information with which to work when they make those licensing decisions.

Suppose we permitted a system of technology audits in connection with licenses to be provided by private companies, which companies would be approved on an approved list held, say, by the Department of Commerce or somewhere else. So that when a licensing officer has to make a decision he has available to him a full report containing all of the information of relevance to that licensing decision.

It is beyond the capacity of the Department of Commerce now to assemble that information, in my view. But it is not beyond the capacity of private, auditing companies to assemble that information in a convincing manner.

And where a technology is particularly sensitive in some ongoing supervision of that technology   --   making sure that it's where it's supposed to be and doing   --   where that is entailed, again a private, monitoring function in which the monitors themselves are monitored by the government; much as the SEC monitors auditing firms, but it doesn't do every individual audit.

I think this could vastly improve the efficiency of the system of export controls now in place so that industry would get faster and less capricious decisions implemented more fully and in a manner that better protects our national security.

But that is a suggestion I make, not with respect to computers in everyday use. Those I think, are beyond practical control. That is a suggestion for those sensitive technologies that require serious scrutiny under the Export Administration Act and other legislation.

Thank you all very much.

SENATOR ROCKEFELLER: Thank you, Richard, and now we turn to Dr. Andy Grove. And I would say to Andy Grove that in the audience is probably about 23 to 25 percent of the staff of the entire Congress which is interested in this subject. So that is your audience, sir, and we welcome your comments.

DR. GROVE: Good morning. Can you hear me okay?

SENATOR ROCKEFELLER: Very well.

DR. GROVE: Excellent. Well first, let me start with a bit of a technical correction. I believe Senator Rockefeller described it as a satellite connection, and satellites have bad names these days and computers have good names, so let me correct this.

This is a computer connection. I'm talking through a computer: first a computer whose brother very soon is going to be called supercomputer by fiat, and that is connected to a personal computer that you are looking at through ordinary telephone lines.

So it is not a satellite connection; it is a connected, computing connection. That may not be a very important distinction to you but for those of us that have worked at making this possible, it's very important.

With that, and picking up on Senator Frist's introduction, I would like to talk a little bit about Moore's Law, so your chart should be showing what Moore's Law is. This is a semi-log chart. On the vertical axis we show number of transistors, and each division represents a tenfold increase of the number of transistors that are in mass production at the time shown on the horizontal axis.

A straight line on a semi-log chart represents a constant improvement per year, or per 18 months in this case, corresponding to approximate doubling of the transistor density that's practical in mass production every 18 months. You can see that the state of affairs has been operational in the industry since the early '70s.

The first point on the chart is the first microprocessors that were introduced in 1971. The last one on that chart is the Pentium III; that is the last microprocessor that was introduced earlier this year. And if I can show you what a Pentium III looks like, I glued it on a 3 X 5 card.

That is it. That little black dot that you see in the middle of the card contains 10 million transistors and its operational speed today that you can buy in any retail store is measured at the operational frequency of 500 megahertz.

This is not a static affair. It has never been a static affair in our industry and I think that is the key message that I would like all of you in the audience to take away. If I can move to the next chart, let me give you a bit of a picture of how this industry operates.

Vertically, from top to bottom, I'm showing different generations of microprocessor architectures going all the way back to the early part of the decade of the 486; ending down in the bottom with the device that I just showed you, the Pentium III.

On the horizontal axis on the top I show the characteristics of the silicon technology that we produce these devices with, measured in the finest dimension of the finest feature on the silicon surface. Understand what those little dots are in the context of what we operate, this is a silicon wafer. This is what we produce in our factories.

One of these silicon wafers has 200, 300, or maybe even as few as 150   --   but some number of hundreds of those little squares that you see on there. The way we improve the quality and the performance and the cost of these microprocessors is moving in both the vertical direction downward and in the horizontal direction to the right.

We improve the architecture of the microprocessors and we improve the silicon technology by making it more and more fine. The consequence of that as we move diagonally from the top left to the bottom right, is two things: constantly increasing performance and constantly declining cost of computation.

This chart is the heart of the PC industry and this chart frankly, is the heart of the internet because the internet is built on silicon chips like this on both ends of the wire when you connect to the internet.

But just to illustrate what I'm talking about and bring it home to the subject that we are talking about today, let me show you on the next chart the cost per million instructions per second; again on a semi-log chart where one major division corresponds to a tenfold increase in performance   --   actually 100-fold increase in performance in this case.

And what you see is Moore's Law in reverse in a way. It's a straight line in a semi-log chart corresponding to a constant rate of improvement in costs per year, and at over 100   --   almost 1000-fold improvement in a 15-year period of time in the cost efficiency of computation. This is why competition has become ubiquitous; because it has become so inexpensive while at the same time it has become so much more powerful.

Concentrating on the power and going into a more near-term picture, let's take a look at the next chart that shows the current Intel devices against the quarters, calendar quarters of the current year and the next couple of years.

The red lines on this chart, the solid line represents the current level of acceptable of microprocessor performance measured in MTOPS, millions of theoretical operations per second, which is the parameter that was constructed for the purposes of export administration.

And what you see in that is that the devices that we ship, starting July 1st in the upcoming quarter, will be exceeding the current operational MTOPS level imposed. And by definition, by regulatory ruling, those devices, those little chips that look just like this, will be rendered to the class of supercomputers.

There has been a great deal of discussion between ourselves and the Commerce Department and ourselves and the Defense Department on giving temporary relief by raising the MTOPS level that is applicable to chips from the 1200 level where it is at now, the solid line, to 1900 MTOPS level, which corresponds to the dotted line. And this would give us relief   --   assuming we perform on the expected rate that I'm showing to here, the little dot   --   would give us relief for about a year.

But it is clearly not the right structural solution. It is what people in the medical field call "palliative solution". It makes the patient feel better without curing him.

We have a more fundamental problem which is that as the commodity computer that I described   --   the ubiquitous computers that come out of microprocessors that are incorporating personal computers like the ones you and I are talking on, the ones you can buy at the retail store   --   as their performance increases, supercomputers in a similar, parallel fashion increase in performance also.

So that the distinction between what is a real supercomputer and the commodity computer continues to be very wide. And this is illustrated in the next chart where I'm showing in this big bar   --   I'm actually showing two bars on this chart, believe it or not.

On the left, the big one, is the IBM, deep blue, whose performance is measured in MTOPS. The bar that you can't see because on this scale it is basically unseeable, is the performance of a 8-way computer; eight Pentium III Xeon processors are put together operating at the top frequency today   --   at 500 megahertz.

That is the difference between the highest end of what is considered a commodity computer   --   and I will show you in a moment what that looks like   --   and a real supercomputer. I've shown you what the chips look like. The chips then get put into this little cartridge, and this little cartridge is put into a computer board that looks like this.

This as you can see, has two Pentium II processors on it. Four of these put into a chassis, into a rack, make up one of those 8-way computers that you cannot see   --   whose performance you cannot see in comparison to supercomputer.

The assembly of one of these computers takes perhaps a half-an-hour. Anyone with an 8th grade education can put it together. All the subsystems are available from international sources from all over the world. They are all standard; none of them are customized. Quite clearly, this is very much a commodity computer in characteristics.

Let me move on to the next chart, please. There are something like five million multiprocessor systems who incorporate anywhere between two and eight processors that are estimated to be sold in the year 2000. Sixty percent of them, like 60 percent of anything we make, Intel makes or this industry makes, are sold outside of the United States.

A good fraction of them   --   in this case estimated to be 22 percent, and that assumes no change in export laws   --   are expected to be manufactured by foreign companies. There are tens of thousands of distributors and dealers who supply the subsystems and the systems that I showed you, and systems built out of those systems that I showed you, worldwide.

And as I mentioned, the know-how of how to build these things is widely distributed. It is known by hundreds of thousands of individuals all over the world. And the fact that the design is modular like this, lends easier assembly.

So what we are facing with, what we are flirting with, is trying to control that silicon chip that I'm showing comparative to a postage stamp, and it is clearly an uphill battle. And not only is it an uphill battle, it is a counter-productive battle as Mr. Perle indicated a few minutes ago.

For export control to be effective   --   if I can go to the next chart   --   the key is the need to really choose what we want to control. We need to focus, we need to be selective, and in the moment that which we try to control becomes a commodity, it becomes uncontrollable for the purposes of export control.

If we accept the principle of selectivity, I think effective export controls can be implemented. But we have to really accept the nature of computing having become a commodity in the last decade. The ubiquity of computing, the ubiquity of computer technology, and the wide proliferation of expertise around the world are the realities in which we tackle this problem.

Thank you.

SENATOR ROCKEFELLER: Thank you very much, Dr Grove. Now, William Archey.

MR. ARCHEY: Thank you, Senator. I was going to change my prepared remarks and just get up and say I agree with Richard Perle and sit down. But I'm going to try to show some things that are a little bit of a take-off from what Andy has done   --   in fact, there's one chart that's the same; the rest are a little bit different   --   but give a little different perspective.

One of the things I would like to say about the days when Richard and I were doing export controls, be under no illusion: Richard was the senior export control officer of the United States Executive Branch. But there are two big changes since; one he alluded to which is that, even Richard, into the '80s there was still an awful lot of the leading edge research that was going under direct sponsorship of the Defense Department.

As opposed to today where more and more of applications to defense systems or to security purposes is coming first out of civilian sector, out of civilian R&D and then is being applied to security and defense applications.

And then the second one has to do with something that's interesting which is   --   we only had a few of these a year   --   but the CRAY supercomputer of about 1983/84 which was about $20 million, is today's $1500 PC in terms of computing power. And I just want to quickly take you through some of that.

This is a different slant on what Andy Grove had to say. I just want you to take a look at this because this is an important point. I may have to get away from the microphone but this is the advantage of being the ninth of 11th kids; you don't need a microphone.

This is when the first liberalization took place of export control. This was the one that caused a lot of the controversy in 1996. The point that this chart   --   this chart makes a lot of points   --   but between 1996 and 1999, again taking the notion of Moore's Law, we're looking at a level of MTOPS now that is five times greater in three years from when the first, major liberalization of computer controls took place.

Second is, that in 1999 we now have a family of three Pentium III chips that starts out at 1167 MTOPS and is going to go up to about 1700 MTOPS. Here's the 2000 level that either requires a notification to the government or at least   --   or a license to about 58 countries in the world.

Then in the year 2000 Apple Computer is going to have a 2500 MTOPS microprocessor, Intel is going to have a 32-bit microprocessor that's going to be about 2600 MTOPS, and then the Merced chip way up here, the year 2000   --   probably the last half of the year 2000   --   is well over 5000 MTOPS.

So you are talking about change; in fact, Moore's Law squared to some degree in terms of the space in which it's occurring. This is also, to take a different perspective from Andy Grove, which shows you what this means in terms of a server or a PC that uses two or more of an Intel microprocessor or anybody else's microprocessor.

The key in this is that   --   and I'll give you some of the metrics of this   --   but the key in this is to know that this year, before this year is out, a 2-microprocessor server is already over what the United States Government says is a supercomputer.

And one of the things that you should note that this was made   --   and by the way, I'd like to offer my thanks to the people from the Computer Coalition for Responsible Exports, CCRE, both members and staff, for a lot of help in all of this.

But one of the points that was made by one of those members a couple of days ago is, you will never see the industry next year going back to making boards or making processors or servers that are going to be less than two times   --   than two servers. It's over. In fact, it's as ancient as when the Defense Department in 1984 decontrolled 8-bit computers. Those aren't going to exist anymore.

The other point is that when you get to next year with a Pentium III 750 megahertz chip and you have eight of them   --   eight microprocessors on a board   --   you are looking at 12,225 MTOPS.

What we are saying is that given the ubiquity of these   --   and by the way, Andy's already alluded to it   --   worldwide unit sales of multiprocessor systems is going to be about five million next year; about 170,000 of those are going to be 8-processor servers or computers.

They have in fact, become a commodity, and in fact, given their ubiquity, the distribution, the ease of getting access to parts, etceteras   --   these are going to be, with all practical purposes, no matter how much we would like to think to the contrary, uncontrollable.

And that gets me to another point, and this is Andy's chart from a little different perspective   --   well, same perspective but it just shows it a little differently   --   that's an SGI in the Pacific Blue IBM, true supercomputer all the way at the top. There is the level for the Pentium 500 megahertz, 5-processor system.

And here I think to maybe graphically portray this, what the industry would like to have is 13,200 MTOPS as the bottom line for no licenses, because you're going to be able to get them. But here is all we're saying if you look at this from a different perspective. What we're saying is, here's the current level. We want to be able to go to about here. That's less than one percent of what a true supercomputer is.

We're not talking about, let's go way up here. We're talking about right about here to be the level for being able to ship a computer without notification or without a license. We don't think that that's excessive. And by the way, I will just show you something; the point that Andy made. We've got one right here along the lines   --   Andy's is probably better.

But this is the Mother board available all over the world   --   particularly in places like Taiwan. And all you do is you basically take the chip and you plug it in like that. That's one microprocessor.

You take two and what you've got is a 2-microprocessor system, which I've already noted with the 500 megahertz chip is going to be above what the government says you can do without a license or without notification.

Somebody may want to just pass this around to   --   I'll first of all give it to the panel and then let them look at it. It literally   --   if I can do it it's as easy as what Andy suggested.

And then lastly, I just want to show you one other aspect about this because this always comes up. These are the major producers of multiprocessor systems that are over 2000 MTOPS. You can see that the foreign competition is increasing and it's going to increase even more. This is where the industry's going, and it's very clear that the foreign manufacturer do not want to if they can help it, cede the industry if you will, to the United States' companies.

I think the bottom line for this industry is that we need two things. We need to establish what I'm suggesting: a new floor if you will, on the ability to export both chips and the computers and servers that they're in.

And the second thing is, we need to do what Andy Grove has talked about: we need to start looking at microprocessors and the computers they're in in terms of being a commodity with enormous distribution, ease of being able to get them, ability to order them off the internet, ability to get service without having to go to the original manufacturer   --   all kinds of different criteria.

We are heartened by the fact that the Senate bill is in fact, got a provision in for commodities and we'd like to see some further work done on it. This is the Enzi Bill. But it does recognize the concept of commodities in high technology. We think that's a major step forward.

And then I'd like to end on just one other comment coming from a circle of somebody that Richard knows very well, and I'll end on that note, which is a comment by Frank Carlucci in front of the Senate Committee last week.

We need also to acknowledge that many categories of products and technologies are becoming less and less susceptible to control. What this suggests is that we should focus our export control efforts on those categories that have important military uses and which we can actually control with some effectiveness.

But we should do that only which has a true effect; not that which simply just makes us feel good. Many technologies are uncontrollable given the access on the internet and other ways; others can and will be supplied by our competitors.

That's the former Secretary of Defense and National Security Advisor in the Reagan Administration. I think he's absolutely right. Thank you very much.

(End of Tape 1, Side 1.)

(Start Tape 1, Side 2.)

UNDER SECRETARY REINSCH: Well, this isn't any fun at all. I was looking forward to an opportunity to disagree with Richard because I usually do   --   just wait. Well, I'm going to try to give you one or two.

I think I can agree with most of the things that most people have said so far, and I do want to make clear that with respect to the pending issue of computers, the Administration is reviewing exactly the information that's presented to you, exactly the issues that have been raised, met with the industry including Mr. Grove   --   I think the week before last   --   and is determined to address this problem and has told the industry that we're going to address this problem by the end of the month. And as near as I can tell, we are on target to do that. And I think that's all I'm going to be able to say about what we're trying to do.

I want to make instead I think, two points; having reflected my basic agreement with everybody else which makes this less of a debate than we might have had; although maybe it's true that everybody out there agrees we're all one voice   --   in which case these will be easy decisions to make. But my suspicion is they're not going to be easy decisions to make and I want to make two comments about that.

One is to pick up on a couple of things that Richard said in particular, with respect to what's happening in the global economy. And the Administration takes as its starting point in thinking about these issues, the statement the President made last year at the 50th anniversary of the GATT when he said globalization is not a policy option, it's a fact.

And that is the environment in which we live. And I think what Richard talked about very effectively was, the futility if you will, of export controls in some situations, in mass market situations, in that kind of globalized economy. And that is certainly something that we're trying to deal with and respond to.

There is another issue though too, that we've began to think about and that is how the very ubiquity of that technology and the globalization economy affects our security in some positive ways.

In that sense, what we have also witnessed in the last 15 years has been first of all, the blurring of the line between civilian and military technologies, the evolution of the civilian sector   --   particularly in information technologies   --   becoming the technology leader in the United States, and the migration of the Pentagon increasingly toward commercial off-the-shelf procurement rather than military prime contractor procurement, for all the reasons that my colleagues cited.

That's a sea change in the way we think about things, because if you look at it that way, the Pentagon and the Defense Department, not to mention our Intelligence establishment, need high performance computing power. They need it very badly. It's a very important element. And if any of you have watched closely what's been going on in Kosovo you can see why we need it; for a whole host of functions.

The reality at the same time, is that the military establishment in this country doesn't buy enough of these things to keep anybody in business. These companies' sales to the Defense Department and to   --   at large I suppose, the Intelligence community as well   --   represents single digits of their sales.

What keeps our companies alive, what Mr. Grove referred to, is exports. Computer companies, more than 50 percent of their sales in each case, come from exports. So the equation that we have tried to construct, which is a little bit different from previous equations is, strong exports mean healthy, hi-tech companies, in turn means a healthy defense and strong national security.

And that is the equation that we've been trying to articulate up here. Meaning that exports and the ability of our companies to compete in the global market is integrally related to our national security and the health of our defense infrastructure. Because we need these companies healthy, we need them making money, we need them investing their money in R&D on next generation products, we need them ready, willing and able to work with the Department of Defense to develop some of the specialized products that they need.

Those things are going to happen in a globalized environment, they're going to happen in an economy in which our companies are selling successfully and capture market-share. They are also going to compete in that economy in an environment in which foreign availability   --   which is something that many of you are going to be thinking about as the Export Administration Act makes its way through the Congress   --   is thought about in a different way, in a far more dynamic way than it has been before.

When you think about computers, we think it is not appropriate to think about availability in terms of who makes the box today. First of all the issue   --   as I think the two preceding speakers have pointed out   --   isn't the box of chips. And the ease of putting the chips into the box is established and widespread; as in fact are manufacture, sale, and distribution of the chips.

So we have a ubiquitous technology here. There are a lot of boxes that people don't make. And the high end of what Bill Archey was talking about is not I would say, available in the traditional sense of the word, but it is not unavailable because we've monopolized the technology. It's unavailable because we monopolize the economics; because people cannot compete with companies that are as efficient as Intel or HP or IBM or some of the others.

If you are going to take those companies out of the marketplace   --   whether it's a particular country or a particular region   --   then I can guarantee you in ten months, 12 months, our friends in Japan, in Singapore, in Taiwan and Korea and in Europe, will be making the same products that we are no longer selling there.

So we are trying also to look at availability from a more dynamic standpoint of thinking not only who competes against us today, but what are the consequences of our expert control system and who is going to compete against us tomorrow if we opt out of the market? And if they do that, what is that going to mean for our ability to meet our own needs, and particularly our own national security and our intelligence needs?

So that's a slightly different way of looking at I think, the same points that have been made. The other point that I want to make that's just why these decisions, if you will, are more complicated is that there is a mythology that builds up, particularly in the Congress, frankly, about this process, where it becomes very easy to make generalizations as we all, including me, made   --   and have everybody nod their heads and agree.

I just came from a hearing on the Export Administration Act where we all made the same generalizations and everybody nodded their heads. But when one gets down to specific cases there is enormous opportunity here to come out in a different place and to make some different points. And I want to cite two examples of the mythology: one of them I think is amusing and one of them which is computer-specific, I think is not.

The amusing one is that several years ago I had the good fortune to attend a hearing in the Senate with Secretary Brown who was Secretary of Commerce at the time, and we had a United States Senator turn to Secretary Brown during his question period and said, why are you selling the Chinese an aircraft carrier?

Well Brown turns around and looks at me, and I have no clue, so I turn around and look at people who work for me; they have no clue. So we say, well we'll look into it and get back to you. Well, we looked into it and we discover that yes indeed, there's an aircraft carrier. It was built in 1940, it was decommissioned in 1970   --   built in '42, decommissioned in 1970.

Both the Navy and the Defense Logistics Agency certified in writing that it was fit only for scrap and that all the weapons systems on it had either been removed or cut into little pieces and deposited at the bottom of the ship. It was inspected by a member of Congress before it left port in the United States, and it wasn't sold to the Chinese, it was sold to the Indians for scrap. Aside from that, the Senator had his facts straight.

The point is though, that episode has entered the mythology of export controls and now there is a body of people here who are convinced and in the media who are convinced, that the Department of Commerce sold an aircraft carrier to the People's Republic of China.

And we deal with this every day and we deal with it in every sector. In the computer case that I would cite, we merrily went along our way doing the decisions that we made in 1995 to increase control parameters for high performance computers   --   which we were criticized for by some members of Congress at the time   --   which we are now being told of course, they're too low, and they are too low.

But what happened then? What happened then is a couple of computers were sold to nuclear weapons laboratories in Russia. Now, what does that mean? We had a big crisis. Should they have been sold? No. We had rules that would have precluded those, and clearly something went wrong and that's a matter of investigation.

In fact, in one case it's a case of judicial outcome already and a substantial fine and I don't want to get into all those details. But at the same time the fact is that computers went to bad places. Now, what were those computers? In one case they actually were 16 PCs, and there was concern that they could be linked together, or networked   --   which was a valid concern.

In another case it was a computer that operated around 4,000 MTOPS; maybe 4300 MTOPS. This was late '96, early '97. Arguably, based on today's discussion, the right answer would be: it doesn't matter; the technology is out there. Certainly for the PCs.

We're not happy about it because we don't want those facilities to be doing some of the things they do and we don't want them to do it with American equipment, but here we are. Unfortunately, that was not the reaction; that was not the reaction in Congress.

The reaction in Congress was, the system has failed, and now we have 2000 MTOPS as a statutory limitation on computer controls. And the President can raise that limitation but he can only do it with a 6-month waiting period. So the decision the President is going to make between now and the end of the month, which is next week, is not going to be effective until after Christmas.

Regardless of anything you've heard here, regardless of the speed with which these are moving, this decision is going to be effective on December   --   I don't know, 30th, 31st, whenever six months expires   --   unless the Congress decides that it wants to change the legislation.

The Congress has responded in these cases, as has the media, to some of the mythology that goes along with these things. And the dilemma that we face in trying to make these decisions is trying to deal, frankly, with some of the mythology. The science is clear. The technology is clear. The ubiquity of the technology and the speed with which these products have moved is clear.

At the end of the day though, you still have the fact that no matter what line you draw   --   whether it's 2000 or 6000 or 8000 or the 12,300 the industry has told us we should draw for the next year or so   --   somebody somewhere is going to ship a computer below that line to some bad guy, and the fact is that all of these machine have military applications.

I just came from a hearing, listened to Deputy Secretary of Defense Hamre point out that the Stealth Fighter was designed on a 100 MTOPS computer. So it is hard to say these machines don't have military applications. They do have military applications. The right answer is to say, at some level we simply can't do anything about it and we're sorry about that and that's too bad.

But we should move on and focus our attention more in the direction of what Richard was talking about or what some other people are talking about; things that are choke points, things without which weapons systems can't be built, and narrow our focus.

The reality nevertheless is, that every single time we go down that road somebody sells something to a bad guy and then we get involved in a major crisis up here saying, the system has failed, we need to fix the system so that won't happen again.

And if you want me to tell you frankly or honestly why these decisions are difficult, why they take a long time, and why they don't always come out the way that the industry wants them to come out, it is primarily because we struggle with the mythology every day.

And if we can get past that and focus on the technology and focus clearly on our national security and what our national security interests are, then we will be able to make wiser decisions. Thank you.

SENATOR ROCKEFELLER: Thank you. I know that Andy Grove isn't going to be able to be with us for this entire session; at least I've been told that. So Andy Grove, there may be some responses, or comments, or thoughts that you might have.

Can you hear me?

DR. GROVE: Are you asking me?

SENATOR ROCKEFELLER: Yes. I was saying, since I'm told that there's a possibility you may not be able to be with us for the entire time, if there are any   --  

DR. GROVE: I'll stick around, Senator.

SENATOR ROCKEFELLER: All right. Well, we have our first question, and this is I guess to anybody in the panel. "What are your thoughts on vendors outsourcing government computer systems work overseas?" Well, we'll just leave it at that. "And to foreigners within IT companies?"

Any of the panel members want to respond to that? And again, we'll encourage people to come up to the microphones as well.

UNDER SECRETARY REINSCH: Well, we of course, prefer that they source to American companies and create American jobs in the process, but I don't think that's what they wanted me to comment on.

Anybody else want to?

MR. ARCHEY: You handled it very well, Bill.

SENATOR FRIST: Okay, we'll move. Anybody to the microphones?

SENATOR ROCKEFELLER: We encourage questions.

SENATOR FRIST: Bill   --   Dr. Reinsch, please provide the Administration's views on the draft legislation by the Senate Banking Committee on reauthorization of the Export Administration Act. Any concerns with the provisions dealing with mass-market products?

UNDER SECRETARY REINSCH: Well, I'll spare you my testimony, which was about ten minutes. We admire the effort. We're glad it's been made. We've had only since last Thursday to review the Bill. We think that it has some serious problems.

We also think that it has a significant number of drafting errors or unintentional problems that need to be worked through   --   which is normal, I hasten to add, in this case. But we think that it takes some time to do it.

I was very gratified that at the conclusion of the hearing today the chairman announced that he was going to defer the mark-up of the Bill from next Tuesday until after the July 4th recess, which I think will give us time to address the problems.

The Bill has a lot of promise. It's got some significant policy problems from our point of view. On the mass-market provision I need to speak personally at this point because we haven't finished discussing it internally.

I think it's a good idea, it's a good concept. It is certainly something that we wrestle with anyway in the concept of foreign availability, and it is an authority that I welcome. And listening to Dr. Hamre it certainly sounded as though it was something that he would be very interested in as well.

I can say the way the Bill is drafted the process that would be employed to obtain such a determination, which would essential require a consensus of the Defense, State, Energy, and Commerce Departments, is a consensus that would be unlikely to be achieved the way the Bill is drafted. But the basic concept I think, is one that is worth pursuing and worth looking at.

UNIDENTIFIED PANEL MEMBER: If I might, I haven't had much of a chance to look at the details of it. I think the concern of some of our companies on the mass-market commodity aspect of it is that it's only retrospective. We think that we can put in place in the law and in the process and the Administration, a process by which we can look out prospectively in terms of mass market determinations, because we'll always be playing catch-up, as Bill well knows, if we just have something that looks backwards.

UNDER SECRETARY REINSCH: Of course, we did that on computers in 1995 and then were heavily criticized for going to a level that was not at that point, available. I think we did the right thing. I think Bill Archey is right.

SENATOR FRIST: We've got another question from the Congressional Office. "We've heard the argument that unbreakable encryption is a tool for terrorists and organized crime. Would you give us examples of how prevalent this is? Does anyone know?"

UNDER SECRETARY REINSCH: Well, I think that one was directed at me, but it's really directed at Louis Free and he's not here, so I can't cite specific   --   well, I don't want to cite specific examples.

What I will say is that there have been a number of hearings up here in the Senate Commerce Committee on this side and on the House side innumerable times, in which the Department of Justice and the FBI have gone into considerable detail about how removing controls on encryption would handicap their ability to deal with terrorists and drug dealers, among other things.

There are cases on record in which encryption played a role, and they take this matter very seriously. I wouldn't take the time right now to go into the particular cases, but I can tell you that I've heard those stories and they're good stories.

I mean, one of the frustrations of the encryption issue   --   and I share these frustrations deeply. There's no issue that's been more intellectually challenging for me, frankly, then trying to deal with encryption because there's a lot of merit to the mass market argument there.

At the same time I have to tell you that I think the argument of a national security and law enforcement downside to de-control is an unassailable argument. The evidence is clearly there. And if you want to de-control there's going to be a cost in the law enforcement and national security area and we have to recognize that.

SENATOR ROCKEFELLER: At the microphone.

AUDIENCE MEMBER: Hi. I just want to say it wasn't very easy to get here. I just wanted to ask Mr. Perle, in your statement you said you wanted to   --   we should focus on intelligence operations for disrupting the flow of critical technologies. I'm wondering what's your view then, of an export control structure, or do you think we should just scrap it?

MR. PERLE: I think we put too much emphasis on licensing and not enough on intelligence-type operations. It's unlikely to be the case that terrorists who are rogue states   --   who are potential proliferators about which we're concerned   --   it's unlikely to be the case that they will base their effort to acquire technology on the normal license application, and therefore that the normal license application will frustrate their efforts.

We would be far better off in my view to put the manpower and the effort and the imagination and ingenuity into understanding who is trying to acquire what, and then operating intelligence operations against them.

For example, sting operations are effectively used by police departments around the world. It seems to me that in the illicit acquisition of sensitive technologies, sting operations should have a prominent role to play. The reason why you don't hear about them is that we've not been doing it.

I think we should, and I think we should look to those   --   to law enforcement techniques rather than bureaucratic procedures in order to interfere precisely with the programs that are of greatest difficulty. If you look at thousands of licenses in the hope that you will stop an illicit transfer, you may or may not do that.

But I think if you take the equivalent number of people and you put them onto the task of highly tailored intelligence operations you'll get a better result.

SENATOR FRIST: Again, I want to give everybody the opportunity, the other panelists to respond on each other's comments and Mr. Grove, feel free to comment at any time on any of the comments made to get some discussions going back and forth.

Do we have any other comments on that? We have another question from the audience.

AUDIENCE MEMBER: I have a question about the MTOPS standard, specifically for Dr. Grove. Given the pleas for selectivity in the standard, in the evaluation of these licenses; and two speakers, Dr. Grove and William Archey, made much of the comparison between Deep Blue and multiprocessing using just standard Intel 500 megahertz chips. But is this really an accurate comparison?

In my understanding of the way Deep Blue worked, it had a fairly standard series of processors, very advanced architecture and very advanced algorithms that optimized its use for chess decision-making. But for a nuclear simulation, say using legacy codes that were supposedly or may have been stolen from Los Alamos, would it not be just as good to run these codes from a generic 8-processor Intel board as compared to what you're advertising as a supercomputer?

I guess more generally, if MTOPS are at the core of the question of what level you set the MTOPS restrictions, is there not a preferable way to define which computers to be exported? Maybe link more to the algorithms in the architecture in simple processing speed.

SENATOR FRIST: Dr. Grove, if you were able to hear that, do you have a response to that?

DR. GROVE: I heard most of it and yes, I do have a response to that. A couple of points. First of all, a supercomputer may be in fact, be built out of many microprocessors hooked together, but it is not a straightforward extension like what I showed you on the screen here   --   taking the same board and adding them multiple times into the same chassis.

The technology that allowed the power of the multiple, large number of microprocessors hooked together to operate as a single computer, gets to be well beyond the assembly of standard components that we are talking about here.

We had the opportunity to supply the Energy Department with a machine a couple of years ago where we hooked together 9,000 earlier microprocessors. And it took us an extra group of people, large group of people, two/two-and-a-half years to build that machine. So the physical technology, the hardware technology implicit in building these large parallel machines is not the same as the physical technology used in building commodity machines.

The second aspect is that the optimized algorithms that the gentleman who has the question refers to, that's software. And nothing in the current doctrine of export control deals with different software.

And I have a hard time imagining they would say that this machine can only be used with lousy software, not with good software. You know, if we have enforcement problems today that would get those enforcement problems into incredible, unimaginable levels.

But the third point that I want to get back to is, I don't think the right question   --   framing this issue in terms of the question, what is the desirable MTOPS level; is an appropriate one? We heard one of the panelists comment that the Stealth Fighter was designed on a 100 MTOPS machine.

Today you wouldn't use a 100 MTOPS machine, you would use a much higher performance machine and it would do a much better job. But if you didn't have that you would spend a lot of time, a lot of iterations, on a slower machine. Everything that can be done on a slower machine can be done on a faster machine; much faster, much more accurately, and much more responsive due to the physical constraints.

But given that situation, faster machines are better for good stuff, faster machines are better for bad stuff. You have to fall back on what is practical to control. What is practical to control when machines that are reasonably fast are available by the millions all over the world?

Do you want to feel good   --   again to quote one of my fellow panelists   --   or do you want to be effective? If you want to be effective, exclude out of the control regime those things that are available in mass volume through large number of distributors; whether or not those machines can be put to bad use. It is simply a waste of effort and exercising, chasing mythology, to quote Mr. Reinsch, and not effective export control.

SENATOR ROCKEFELLER: To follow directly on your question, there's a question before me which says what you have basically said and then said, "Should we therefore"   --   the question assumes MTOPS is no longer a fair or a proper measurement, and the question ends by saying, "Should we therefore move to something called a 'generally available commercial machine'? Should that be the definition?"

DR. GROVE: I would probably phrase it just a little bit differently, but the basic answer is "yes". I would replace the word "generally available"   --   I would replace that phrase with "mass availability", "mass market availability". You know in a colloquial fashion, if you can buy it in your neighborhood CompUSA, it's broadly commercially available. That's one way to do that.

Since legislation is rarely put in the colloquial terms I would probably substitute "volume level"; something of which there is 100,000 of them are available in the world is mass available. And I would exempt those things, whatever MTOPS they represent, because chasing with MTOPS is like driving looking with your eyes firmly locked on the rearview mirror.

UNDER SECRETARY REINSCH: I think at the end of the day, this ends up being partly semantic in the sense that one has to draw a line. No one has said   --   I don't think anyone believes   --   that computers are undifferentiated products and all of them are mass market.

Some of them are widely available, some of them are not, and the appropriate approach I think, is to draw a line that tries to exclude the ones that are and not exclude the ones that we care about; which presumably are the ones that are not widely available.

Drawing that line can be done multiple ways. We have chosen to do it by MTOPS. You can't find anybody in the Administration that will tell you that that's a wonderful way to do it. It's just a way that we ended up with that was, you know, like they say about democracy, better than all the other ways.

The industry has come in and had a variation of Mr. Grove's comment and said, well you ought to do it by the number of processors, and that 2-, 4-, and 8-processor machines are widely available for all the reasons that have already been discussed, but there's kind of a line   --   in fact, I think Mr. Grove made this point in the White House meeting two weeks ago   --   that machines above 8-processors take some extra technology, some extra skills to produce, and they're not mass market machines.

That's a fair point. And all the 12,300 MTOPS line that the industries advocate is, is taking the appropriate chip they think is going to be widely available next year and multiply them by eight because they're using and 8-processor standard. So you can do it that way, you can do it his way. You can say, you know, in volume. You can do it that way.

But no matter how you do it you're still down to some quantitative limit of deciding essentially, what's above the line and what's below the line and trying to make sure that when you draw that line, however you draw it, what we are about from a national security perspective is above it and what we don't care about is below it.

I think we're flexible about innovative approaches, but the odd thing about them is, they all tend to be roughly in the same place once you crank through what all those formulas mean.

SENATOR FRIST: We have a number of questions, cards on this topic, so we'll spend a little bit longer on the topic itself because of the real interest.

MR. ARCHEY: Senator, if I might   --   on just that last point to the questioner. There is an interesting chapter in the Cox Report that's called "Changing High Performance Computer Technology is Making Export Control More Difficult". And it's actually a chapter on it called the "Technical Afterward".

But it's a very interesting discussion of the various kinds of processing and how to distinguish between a true, supercomputer versus a sexy, commercial computer that happens to be quite quick. And it talks about vector architecture, it talks about parallel processing, massive parallel processing, symmetrical parallel processing, and then basically an array of commercially available computers.

So I'd just recommend you look at that. It's a bit ambivalent because I think the writers of that chapter themselves had some difficulty distinguishing some important variables. But it's well worth understanding how this question is ultimately going to have to get answered.

SENATOR FRIST: Mr. Perle.

MR. PERLE: I don't mean to minimize the difficulty of defining criteria that can be consistently applied, but there's a rule of common sense here. There are computers that exist in relatively small numbers. High percentages of them are to be found in places like the Los Alamos and Livermore Nuclear Laboratories and at Sandia and the National Security Agency and the CIA, and other places intimately involved in national security. They're often special purpose built.

Obviously, you want to be careful before you export a machine of that character to say, the Beijing Institute of Aeronautical Engineering, or to Saddam Hussein. I think the effort to draw broad category distinctions while necessary, may miss the point.

It depends in large part on the context, on who the buyer is, on the institution to which the computer is going to be sold, if you have confidence that you can be sure that it stays where it's sold in the first place.

And the problem has, in my view, always been with the very high-end machines, most of which are to be found in national security applications because they don't make a lot of sense in most other applications. And if you look at it from that point of view it should be possible to distinguish the relatively small number of problem cases and deal with those in context.

I have one question for Bill Reinsch which is, he was lamenting the fact that when the administration takes action as I assume it will, to at least raise the MTOPS level as a palliative, it has to wait six months, and where were you guys six months ago? Why did you let it get to the point where you now have a problem that can't be solved except by legislation?

UNDER SECRETARY REINSCH: Well actually, we were prepared to do this last year, May of '98. The industry came to us and said don't do it now; the time's not right, politics isn't right. So we deferred to their judgment. They came in in November and said, the time is right, and we said well, what do you think is the right answer? And in April or May they told us what they thought was the right answer and we've been spending our time since then dealing with this.

I have to say, I mean, it would have been nice to have done this in January. I would have preferred to have done it as I said, 13 months ago. I'm not sure, in terms of the conditions that I was talking about in my remarks I'm not sure conditions now are any better than they were 13 months ago, or six months ago.

MR. PERLE: I hope you don't   --   the Administration doesn't allow itself to be intimidated by the Chinese fiasco, which would not have in any way, been mitigated by a different set of export regulations. And if we're going to allow espionage to take place in our nuclear laboratories, controlling the export of equipment is not going to solve that problem.

UNDER SECRETARY REINSCH: Well, I just came from   --   I agree with you. You're exactly right and I think it's a very important point to make. I just came from a two-and-a-half hour hearing where probably the majority of the Senators there missed that point and lumped the two together. So there's a lot of teaching to do here.

SENATOR FRIST: At the microphone.

AUDIENCE MEMBER: Hello. This question is probably directed to Richard Perle. This conversation today that we've been having has mostly focused on the national security issue of raw computing power. And yet when I think about the ways in which the military uses computing, many of the great advances or the great advantages are gained from much smaller but distributed networked computer power, sensors and computers on board vehicles and armament communicating with each other gathering information.

And do we have any sort of analysis or understanding, as the world evolves into one where everybody has this network computing availability, what kinds of national security needs are going to be addressed by export controls that are not going to be addressed by export controls because everybody will have that set of capabilities available to them?

MR. PERLE: It's really a very good question. In my view the single most important factor affecting the future effectiveness of military forces will be success in integrating sensors and data processing capabilities. Because if you can find a target   --   and sensor technology permits us now to find targets that could not have been found in the past, and from a distance   --   if you can find the targets you can almost certainly destroy that target.

And therefore, the military force that has available to it, high quality, multiple sensing capabilities, coupled with data processing, will prevail every time over an adversary with an inferior capability in that area.

And of course none of that is affected by restricting on desktop machines, whose only contribution to military capability is whatever utility is made of the computations done in the design or manufacture of weapons systems.

So you've identified an area that it seems to me, is far more important than raw computing power and for that   --   because there's also a wide diffusion of the microprocessors and sensor technology on which that depends   --   for that I keep getting driven back to the idea of intelligence operations focused on the organizations and countries in which there is a much closer correspondence between precise, military objectives on their part, and operations to disrupt the acquisition of technology for those purposes on our part. I don't know how else to do it.

SENATOR FRIST: We've got a question for Dr. Grove. Before we ask that question, in everybody's packet are some evaluation forms, and as Senator Rockefeller and I continue to evolve and evaluate and improve this conference, we'd appreciate very much that each of you drop that off.

We've got a couple more questions. We start on time, we finish on time. We've got about another four minutes or so. Jay?

SENATOR ROCKEFELLER: This is to Dr. Grove. The question is, "How do computer companies monitor their customers (i.e., the bad guys)?".

DR. GROVE: Our accounts payable department is charged with monitoring them.

SENATOR ROCKEFELLER: Well, I think it is a legitimate question. In other words, in response to Richard Perle's question, it's not so much what we sell but sometimes to whom we sell. And if government can't make all   --   shouldn't make all those decisions then industry   --   the implication is that industry has to be aware of to whom they're selling.

DR. GROVE: Well, let's look at the reality of this. Let me talk about Intel which is the one I'm familiar with.

Intel sells   --   let's say a round number   --   100 million microprocessors that go into computing applications a year; 100 million. They go to computer manufacturers who number 2,000 who distribute their machines through dealers who number 50,000, to end customers who number like a million. We have no idea what happens through that chain, and if we try to find out, aside of the practicality, we would be told to mind our own business.

SENATOR ROCKEFELLER: Understood, understood. William Archey.

MR. ARCHEY: I'd like to take it from a different perspective because I think Andy's talking about   --   Intel's a company with 50,000 distributors. And Bill and Richard both may weigh in on this.

When I was in the Commerce Department running export controls, without question the best intelligence that Richard's been talking about we got was from companies. And I'll give you one example in particular. And this notion that companies don't care about national security which sometimes get bandied about up here on the Hill which is just alarming to us and something we get really, really bent out of shape about.

Because I can remember one case   --   and by the way, this was not a particularly aberrational case. A company in Connecticut that came to us and came to me with an order for $60 million. And the reason they came to us was, was that they never saw the customer before. They'd never heard from this customer and they said   --   this was some pretty good technology   --   and they just said we don't think this is right.

And so we went through our own various and sundry abilities and found out that in fact, it was a storefront outside of Vienna, Austria, getting ready to send all that stuff to the Soviet Union. But companies have set up some extraordinary internal control programs   --   and by the way, cost them millions and millions of dollars to do that.

They take it quite seriously, and I would be willing to bet   --   and Bill, you know better than me   --   that things haven't changed a great deal. That it's still the best source of intelligence about stuff that shouldn't be going somewhere is coming from the companies.

SENATOR FRIST: One final question   --  

DR. GROVE: May I butt in?

SENATOR FRIST: Absolutely.

DR. GROVE: Bill, what you described is appropriate and right and doable and is done for computers that fit in with that   --   someplace on that long bar; of which thousands maybe are sold a year. It is not doable by a personal computer company that manufactures and sells ten million personal computers of different ilk a year.

MR. ARCHEY: No, Andy I agree with you and I was making that point. Because you're dealing with 50,000 distributors. This by the way, was not computers; this was very sexy testing equipment. And they don't sell through distributors.

SENATOR FRIST: One last question. "Examples of choke point technology". To any of the panelists. What does choke point technology represent? Or three examples.

UNDER SECRETARY REINSCH: Well, I think Mr. Perle has offered essentially I think kind of one, in one respect which was his comments about sensors, and I don't want to comment on that right now. I think the two that are agreed upon by everybody are fissile material and Stealth technology.

I think what we've discovered is beyond that. Consensus breaks down pretty fast in the sense of anonymity. But where we focus I think our tightest controls would be on what I would generally refer to as production equipment; that is, things that you use to make other things.

We control the chips the way that Mr. Grove described; we control the equipment that he uses to make the chips much more tightly than we do the chips themselves, and that's generally true across the board.

SENATOR FRIST: Thank you Mr. Perle, Dr. Grove, Secretary Reinsch, Mr. Archey. Thank you all for being with us. Next month's briefing, "The Hi-Tech Workforce: How Will America Meet the Rising Demand for Skilled Workers?", will be held 216 Hart, Thursday, July 29th, 12:15 to 2 p.m. Our announcements will be going out very soon.

Again, thank all of our panelists. Give them a round of applause.

(End of Tape 1, Side 2.)