TCS Daily

Exporting the Supercomputer?

By James K. Glassman - October 30, 2000 12:00 AM

Jim Glassman: U.S. corporations say that they are being undercut in export markets because the government is increasingly willing to use export controls for other foreign policy objective. When you were at the Defense Department, one of your areas was export controls and sensitive technology. Do you think that such controls make sense in a global economy with technology changing so rapidly?

John Hamre: Well, I think the government has a legitimate need to manage the export of crucial technologies for security reasons. But I don`t believe that our current system of export controls is working at all.

Glassman: What`s wrong with the current system, and what would you like to see replace it?

Hamre: First of all, the current system was designed back in 1950s and 1960s, back at a time when the mindset was that all the important good technologies were in the United States, all the bad people were outside the United States, they all wanted our stuff and we got to keep it from them. Well, that isn`t the way it is today.

Glassman: How is it different now?

Hamre: We have entered into a world where there is just as much challenging and interesting technology overseas, when American companies have gone transnational in their operations. We have transnational, transcontinental, just-in-time manufacturing systems. We have software that`s written in five places around the globe so that the software team hands off its work as the sun moves around the globe.

I mean, this is a radically different world, and to have an export control regime that still is based on parochial manufacturing, where things are all manufactured in one place that you can put can put a licensing process around to modulate and moderate and control everything that goes in and out of that place, is obsolete. Now, we still need, in my view, security procedures, and export controls are a potential form of security, but not the way we`re doing it now.

Glassman: Give me an example of the kind of technology that we really should not have to worry about being exported?

Hamre: Well, again, let`s draw the broader picture. Of course, we have two different structures of technology control for export. One is under military items, and there is a defined list of things called the munitions list. And nothing can be exported that is on the munitions list unless there is a license for that item granted by the State Department.

The other form of export control is so-called dual-use items. And virtually everything in the world is on the dual-use list. If it`s not on the munitions list, it is on the dual-use list and nothing can go out unless there`s a license for that that is granted by the Department of Commerce.

Now, the process for dual-use items is simplified through an elaboration where there are some things that require approval each and every time that it goes out, and there are some cases where it requires a one-time license review. And then there are times when the government just said we`re not interested in this; you don`t have to worry about it.

So to answer your question, I need to talk about each one individually. But let me talk about munitions list. We still require that you get an export license if you want to export a 4-5 ton truck from the United States. Now, in all candor, there are lot of ways to get a 5-ton truck. But that`s on the munitions list item. If it`s an item in the U.S. military inventory, we won`t let them export it unless they get a license for it.

Glassman: Where do these exports go? We are not allowed to export just anywhere, or just to -

Hamre: Oh, anywhere without a license. Now, there is one modest exception to that and that is currently between Canada and us. But only for the manufacturing process between Canada and us, not, it can`t leave Canadian shores without an export license from the United States. So, for all practical purposes, nothing can go out.

Now, one of the big problems with the current system is that we do not differentiate at all between sensitive technology and insensitive technology. I mean, a 5-ton truck requires a license, as does a reconnaissance satellite.

Glassman: What about dual use --

Hamre: Let me just finish just for a second before dual-use. Another way in which we do not discriminate is we do not discriminate over the age of the technology. We require something that is 30-year-old technology to go through a license just the same as something that is five-year-old technology. Now, that`s obviously not - that does not make sense.

A third area where it does not make sense is that we require export licenses whether it`s going to the United Kingdom, one of our best friends, or whether it`s going to Syria. In short, we do not discriminate on the basis of the risk of the recipient of the technology. So we have some fundamental changes that we have to make to make this system more sensible.

Now, if I may say one last thing before I talk about dual use. I am not an advocate for relaxing export controls because that starts with the premise that the current system is really protecting us. I don`t think our current system is protecting us adequately, and it sure is irritating a lot of people. I think we need a system that is much more effective than the one we have now, but we should not have to waste our time regulating things that don`t matter and regulating them to countries that are our best friends.

Glassman: Right.

Hamre: Now, there are dramatic ways that we can make the current system better. And I am an advocate for stronger export controls where they really count and where it really matters.

Now on dual-use exports, the Department of Commerce puts more resources into it, so that frankly there isn`t the same level of irritation in the business community on them as there as on munitions list exports. The great problem here is that we don`t have a transparent way for the government to work with industry to decide what is truly sensitive and what is commercially available.

Glassman: Meaning that we don`t know why they are making decisions?

Hamre: It`s that the government does not have an explicit process by which it says why it has decided it would or would not be allowed for export. There needs to be more transparency. Sometimes when the government has very good reasons for it, but it ought to be able to explain that and frankly it has an obligation, I think, to explain that rather than just simply saying, oh, we don`t want to do it.

Glassman: Now, we have heard horror stories about computers, which are basically off the shelf used by consumers that are not allowed to be exported. I mean, for example, Apple has been advertising its latest computer, little 8-inch cube as being a super computer. Now, can something like that be exported at will?

Hamre: Let me say first that there is a unique and special problem with computer products. About five years ago Congress passed legislation that defined a super computer in law as any machine that has in excess of 2,000 million theoretical operations per second, in shorthand, 2,000 M tops. Back in the early 1990s, a 200 M top machine was a supercomputer. The problem is that the Sony Play Station meets the supercomputer definition today. So, the difficulty with putting a definition for computer speed into law is that the law does not change nearly as fast as the technology changes.

Glassman: Right.

Hamre: And so, if we did not evolve a rising of speed proceeding standard, virtually every laptop that is going to be produced next year would have to go through an individual licensing process to sell any one of them. That obviously does not make sense. So a year and a half ago when I was in the government, we developed an alternative method so that we advance the supercomputer speed threshold to stay ahead of what we call the commodity rate. That way we could control supercomputers when they were really supercomputers, but we didn`t try to regulate something that had become a commodity because of the advance of technology. That process is in place today. I think it is an adequate process, but unfortunately we have to convince Congress that they need to change the law.

The real issue is supercomputers; regulating speed is not a meaningful matrix for export controls. And let me give you an example. The supercomputer that was used to design the F1-17 Stealth fighter, the stealthiest airplane that flies in the world today, was only 100 M tops. A supercomputer today is in the 10,000 M top category. In other words, it will be a hundred to a thousand times more powerful in terms of speed than the one that designed the stealth fighter.

So to say that controlling a computer by speed is controlling sensitive technology is wrong. It`s really the application that runs on the machine. The only thing is, you just run it for longer, the computer just runs for 10 hours rather than one hour. It does not mean keeping laptops out of the hands of bad guys to keep them from developing stealth fighters. You have to find other ways to keep them from developing stealth fighters, such as the use of applications.

Glassman: What about the argument that even if we deny the potential bad guys these high-technology tools, someone else is going to provide it?

Hamre: That goes to the issue of commercial availability and I must say I am slightly ambivalent on this. I absolutely do agree, in general terms, that it is pointless to take and try to limit the export of a good if a comparable product is widely available internationally. And therefore, all you are doing is you`re just diverting the market to some body else. In general, I agree with that.

However, I really think we should not let people export key components that are required to build nuclear weapons, for example. And that is something that we should take the lead working with the international community to design meaningful, multilateral controls so that nobody exports the key technologies required to build nuclear weapons. There are sets of things -- nuclear weapons, biological weapons, chemical weapons, long-range missile technology -- that the international community ought to jointly find ways to control. In that case, even if the technology is available elsewhere, we ought to limit its export where it really matters, but we should not extend or use that logic as a reason to block 5-ton trucks.

Glassman: Let me just switch to a corollary subject - cyber terrorism. How vulnerable are U.S. information systems - government, defense and private -- today to outside interference?

Hamre: I think they are quite vulnerable. But I also believe that the protection environment is evolving very quickly and in positive directions. I mean, I think we`re still vulnerable as a country. But let me use an analogy. I think that cyberspace today is very much the way Chicago was the way before it burned down, with the fire. And that it was designed without having any fireproof materials, it was designed without fireproof codes, it was designed without having water mains out in the neighborhoods so that it could quickly put fires out. It did not have organized fire departments, etc., etc. That`s where we are in cyberspace today. We have a fire prone cyberspace today and we need to fireproof it. Now, this is going to take time.

It is also something where the greatest asset is coming to the rescue, the profit motive. We now have companies that are making money trying to sell security products. It`s going to take a lot of work. There are a lot of problems and it`s especially difficult because cyberspace was designed to be unusually accessible and open. And the entire structure of fielded software is frankly fire prone not, fire proof. So there is still an enormous amount of work to be done. The country is vulnerable, but fortunately we are making great strides in fixing it. It will take years, however.

Glassman: Everybody these days seems to be talking about the need to increase defense spending, and one area in which defense spending has not been as extensive is research and development, especially in missile defense system. Will such a system work and are we spending enough money on it?

Hamre: Well, let me take the broader question about spending on R&D and then the narrower question about national missile defense. The broader question about R&D is, as a country during the period of budget scarcity in the second half of the 80s and almost completely through the 90s, we decided that we have to cut back on research and development throughout the federal government, not just the Department of Defense, but throughout the federal government.

We are now at a period where that has turned around and we now have budget surpluses and clearly there is an important and positive role for federally funded research and development. I believe that needs to be a new consensus on the role for the federal government to spend money to support research and development. Some very, very exciting things are going on in genome research, for example. That`s building the infrastructure of knowledge for the next 15 years. It`s enormously powerful.

And that`s the sort of thing that the federal government should be funding. On the specific issue of national missile defense, I am a supporter of national missile defense and believe that we need to have a program. Why? Because there is no reason why North Korea or Iraq would need to build a missile that could fly 5,000 nautical miles for its own defense. The only reason why they want to build such a devise is to politically intimidate the United States. And we cannot be in a position where we are vulnerable to that. So, we need to have a program, we need to have a response.

The difficulty is that building it is really tough. Hitting a missile with a bullet in space is really tough. And where we have been evolving is towards a program that probably does it the hardest way, which is on the intercept at the final stages of the attack, rather than at launch.

Glassman: Why not at launch?

Hamre: The difficulty with launch is that you have to be in a position where you can chase a missile that is flying away at three or four times the speed of sound. There are very few good opportunities to do that. You can do that for a missile from North Korea, but only if you have a ship positioned to the north of the launch site. It`s very hard to figure out how you do that for Iran or Iraq.

So the notion of intercept at launch or boost phase intercept, which is the term of art, is highly constrained. Now, there are some that say we ought to go to a space base for this. If you think it is hard to do national missile defense from the ground, where we`ve been at it for 40 years figuring out how to do it, it is dramatically harder to do it from space. And this is years off. We have a program, but my sense is it is more constrained by the technology and by the science than it is by the money just now, although people would argue with me about that.

I personally think that we need to have a near-term response or nearer-term response than you would get from a space-based system. I think the one that was proposed by the administration, I personally favor it, should be fielded when it proves it works. Unfortunately, the tests that we`ve had so far have not been successful. Two of them failed, one of them succeeded. We should not be pushing technology that does not work. I personally favor pushing a technology solution here that does work. And we should not be driven by an artificial deadline to push something in the field that is not going to work.

Glassman: Is it one that you think that`s funded well enough?

Hamre: I think from my understanding, the development program is fully funded. I think where it`s not is in the deployment stage. But the deployment stage is on hold now until we can demonstrate that it is really working. To my knowledge, the R&D is fully funded. I think that people would argue that there ought to be more money put in for testing and it is a very, very aggressive test schedule for the national missile defense program. So if there is additional money that ought to go in, it probably should go to more rigorous sets of tests to make sure it works and especially works in a countermeasure environment.

There are those who argue very serious and very credible scientists who argue that we are not adequately testing the ways in which an opponent would confound a national missile defense system, and that we really ought to be testing in a more stressful aspect. My sense is that we are doing some testing in that area, but probably not sufficient in the long run. And if we were to add more money, it ought to be in that area.

Glassman: Well, thank you very much, John Hamre.

Hamre: Is it helpful?

Glassman: Very helpful and very fascinating subject. I wish we had more time. Maybe we`ll come back to it.

Hamre: Sure, goodbye.

TCS Daily Archives