From Infoware to Infowar
Joseph F. Traub
At its first meeting on May 7 and 8, 1986, the Computer Science and Technology Board (CSTB) identified six critical national issues. I will tell you how those issues looked in 1986 and how they look today from my 1996 perspective. Then I will turn to a seventh issue that I did not want to raise publicly in 1986.
From its inception, CSTB deemed it important to look at both technical and policy issues. In 1986, it was fairly unusual for National Research Council (NRC) boards to consider policy issues. Therefore, at that first meeting we invited senior federal officials to give their views of the critical national issues. The officials included Congressman Donald Fuqua, chairman, U.S. House of Representatives Committee on Science and Technology; John McTague, acting science adviser to President Reagan; Gordon Bell, assistant director, computer and information science and engineering, National Science Foundation; Robert C. Duncan, director, Defense Advanced Research Projects Agency (DARPA); and Alvin Trivelpiece, director of research, U.S. Department of Energy.
I will discuss the six critical areas identified in May 1986. When I speak about them from my 1996 vantage point, I am giving only my own view; I have not had the benefit of discussing these opinions with the Board.
In May 1986, the Board's number one concern was competitiveness. How can the United States best ensure the continued leadership of its computer science and technology enterprise in the face of intensified global competition?
Competitiveness was a big issue in 1986. Indeed, the opening sentences of the NRC's press release announcing formation of the Board were these: ''U.S. leadership in computer research and manufacturing has been seriously eroded. The NRC has established a computer science and technology board to advise federal agencies and private firms on ways to strengthen U.S. international competitiveness in this field and to ensure that the full promise of this area is realized.'' We were deeply concerned about our competitive position, particularly vis à vis the Japanese, in areas such as computer chips, artificial intelligence, and software. For example, we had heard that software from Japanese factories had unusually few errors.
Today, we must remain vigilant regarding our competitive position. The economic and national security stakes are higher than ever, but our worst fears of 1986 were not realized. For example, according to William Spencer, president of SEMATECH, in the mid-1980s we were losing 3 to 4 percent of market share per year. Today, the United States and Japan have about 40 percent of market share each, while the rest of the world has 20 percent of the $150 billion semiconductor market. Business is booming for American companies in software, semiconductors, Internet-related infoware, and above all, in producing content. We must maintain and leverage these positions in an economy that is increasingly international.
The second item on CSTB's list was talent. How can the gap be closed between the small number of U.S. citizens graduating with Ph.D. degrees in computer science and computer engineering and the large demand for graduates with these skills? How can high school graduates be taught to deal with the computers and high technology they must use on the job in fields that range from the military to banking?
I will divide this issue into two parts: Ph.D. production and K-12 education. We have succeeded in building Ph.D. production from some 200 a year to about 1,000 a year in computer science and engineering. In the 1990s, there has been concern about overproduction of Ph.Ds in some specialties as universities and research laboratories downsize and as the field matures. I am not convinced the country needs 1,000 Ph.D.s a year in computer science and engineering. I have seen the pain in physics and mathematics since about 1970; can we learn from history? Incidentally, after a quarter of a century, and due to a number of circumstances, the crunch in physics and mathematics seems worse than ever. Since most of these Ph. Ds will not be appointed to faculty positions in the top research universities, questions have also been raised about possible changes in the Ph.D. program to produce people better suited to positions in industry and colleges.
Problems in K-12 education seem more serious and overwhelming than ever. There is a widespread feeling, which I certainly share, that Americans do not get the education required for an informed populace in a democracy. They do not get an education that will enable them to fill many service sector jobs or to function in our high-technology armed forces. Students are not adequately prepared in analytic and writing skills to do well in universities. If we believe that educated people will be the key resource in the twenty-first century, we have much to be concerned about. Bruce Alberts, president of the National Academy of Sciences (NAS), has identified K-12 education as the most important problem of his presidency.
The third item on the 1986 list was scope and support. What will be the nature of computer science and technology in the 1990s? How can its health and vitality be sustained during a period of uncertainty and stringency in federal research and development budgets ?
I will discuss this issue in two parts also, beginning with nature and scope. It is typical of NRC boards to do studies on the nature of their fields. Some have become classics, such as the Bromley report for physics1 and the Pimentel report for chemistry.2 These studies set the standard for other studies. Since we were the "new kids on the block," CSTB decided that beginning with a report on the nature of the field would be self-serving. We wanted first to build a record of reports dealing with critical national issues. Computing the Future , CSTB's study of the scope and direction of computer science and technology, was published in 1992. The committee was chaired by Juris Hartmanis.
The second part of this item asked, How can the field's health and vitality be sustained during a period of uncertainty and stringency in federal research and development budgets?
Oh, my prophetic soul! Since there are people in this audience who have been grappling with this issue during a period of uncertainty and stringency, I will not pursue it here.
Item 4 on the 1986 list was supercomputers. How can the power of supercomputers be exploited to promote scientific and technological advances, and how can U.S. leadership in this area be maintained?
Supercomputers continue to have economic and symbolic importance. At the time that I write this paper, a major battle is under way to determine whether the National Center for Atmospheric Research will purchase an American or a Japanese supercomputer. The contract is valued at $13 million to $35 million; the amount of interest being generated must be due to more than just the amount of money involved. A teraflop computer will soon be installed at Sandia National Laboratories. Faster, more powerful machines are needed for aircraft design to ensure the safety and effectiveness of nuclear weapons with zero testing, for molecular dynamics in biology, and for cosmological computations.
The federal government's interest in supercomputers has evolved since 1986. A program plan for high-performance computing was published by the Office of Science and Technology Policy (OSTP) in 1989. The High Performance Computing Act of 1991 authorized a five-year program known as the High Performance
Computing and Communications Initiative (HPCCI). In 1995, a CSTB study, chaired by Frederick Brooks and Ivan Sutherland, responded to a congressional request for an assessment of HPCCI with a study titled Evolving the High-Performance Computing and Communications Initiative to Support the Nation's Information Infrastructure. Although high performance continues to be extremely important, there is less focus on supercomputers since the parallel processing paradigm has been accepted and higher performance has become pervasive.
Issue 5 was software. What can be done to promote the economical production of reliable software, which represents a major portion of the cost and effort in the design and use of new computer systems?
I was looking through my CSTB files and found the following note from 1987: "Sam Fuller says Bill Gates has really thought about software; he would be a good person to talk to the Board." Quite so, Sam.
In April 1988, CSTB met at Stanford University rather than in Washington, D.C. Edward Feigenbaum arranged for John Young, president and chief executive officer of Hewlett-Packard, to host a breakfast at which leaders of Silicon Valley companies could tell us what national issues most affected them. Attending the meeting was the chairman of Hambrecht and Quist, one William Perry. I was struck by the consensus among these industrial executives on the two most important issues. They identified K-12 education because they were concerned about finding good employees, and they identified software because they recognized that the economical production of reliable software was a crucial problem for their businesses. This meeting underscored the Board's concerns about software.
The final item on the initial list was infrastructure. What are the important underlying capabilities, such as national networks and electronic libraries, that are needed to support the healthy evolution of computing? How can they be provided in a timely and effective fashion and integrated into daily activities to enhance national productivity ?
The first two major studies the Board published were Toward a National Research Network in July 1988 and The National Challenge in Computer Science and Technology in September 1988. The National Challenge was unique in that the study was done by the Board rather than by a committee specifically appointed for the purpose. Since it was a Board report, we were all equally involved. However, Michael Dertouzos was far more equally involved than the rest of us! The National Challenge presented two recommendations. Here is the first in its entirety:
Enhanced, nationwide computer networking should be seen as essential to maximizing the benefits in productivity and competitiveness that are created by computers. Networking will facilitate the application and delivery of diverse advances in computer science and technology to the benefit of all segments of society. The board envisions an enhanced national information networking capability, and it has already begun to examine a host of related questions about how physically to improve data networking infrastructure; associated costs, impacts, and benefits; and the roles of industry, government, and other interested parties.
Seems rather prescient from the vantage point of 1996.
During the past 10 years, infrastructure certainly became prominent in the Board's portfolio. Indeed, CSTB was initially the abbreviation for Computer Science and Technology Board. After the Board was asked by Frank Press, then the president of the NAS and chairman of the NRC, to add telecommunications to its responsibilities, Marjory Blumenthal and I decided to rename it the Computer Science and Telecommunications Board to preserve the abbreviation.
The Board's list of critical issues in 1996 can be ascertained by looking at the list of current and future initiatives listed in CSTB's new brochure. You might want to compose your own list of the six most critical national issues in computing and telecommunications in 1996.
In 1986, there was also a seventh topic on my mind. In 1985, I began to notice various ways in which the information infrastructure was vulnerable to electronic or physical attack. I imagined myself to be a terrorist or an enemy country, and targeted aspects of what today we would call "the national information infrastructure" (NII). I did not go public with this concern because I feared it would do more harm than good. There is now considerable attention being paid to our vulnerability, especially by the Department of Defense and the media. However, I have strong concerns, and I feel now is the time to express them. I will focus on the civilian infrastructure, although it is sometimes difficult to separate military from civilian in this domain.
It is just because the United States is the most advanced country in the world when it comes to the use of information technologies that we are also the most vulnerable to attack. I will give you one illustration. Financial security is obviously very important to us as a people. We have two primary types of assets: real estate and what I will call virtual estate. I will confine my comments to virtual estate, which consists of bank accounts, equities, certificates of deposit, pension accounts, and so on. As you all know, this is "virtual" estate because it is recorded in electrons. If you were a terrorist and wanted to do a great deal of damage to American institutions and individuals, a natural target would be the virtual estate. Although there are electronic backups and paper trails, I am not convinced that virtual estate is secure.
Then there are the electrons in the foreign exchange markets. According to the Bank for International Settlement, turnover in this market is $1.2 trillion a day. This, of course, overwhelmingly dominates the value of goods moved around the world. Furthermore, an amount of money that equals the annual gross national product of the United States moves through the foreign exchange markets in not much more than a week—and it all consists of electrons. These markets are international, but a successful attack could also have major domestic consequences; the line between an international and a domestic attack has become blurred. This fading of distinctions is characteristic; we will see more examples later.
Our virtual estate is just one example of a potential target. Others include the power grid, the air traffic control system, and the communications system. I see the protection of information assets as a national security issue, although this view is not universally shared. In November 1993, I was one of seven civilians who participated in a seminar convened by Andrew Marshall, director, Net Assessment, in the Office of the Secretary of Defense. Marshall is a highly respected and influential pioneer in what is sometimes called Revolution in Military Affairs (RMA). Mr. Marshall organized the meeting because he believed that the world stands on the threshold of new information technologies that would have a profound effect on the U.S. military establishment. Participating on the military side was a group of general officers from all the services.
I argued that there was a history of the military defending transportation assets—such as rail lines, harbors, rivers, and airports—during wartime. Should the armed services consider protecting the national information infrastructure? Such a mission poses many questions and presents its own risks. It is further complicated by the blurring of wartime and peacetime activities that has already begun and will likely increase.
There was a surprising split in views among the participants. A number of civilians, myself included, felt that the protection of civilian information assets might fall within the parameters of national security, whereas the military believed that responsibility for protecting these assets belongs in the realm of private industry and the police. Military uses of infowar (IW), both defensive and offensive, are clearly the responsibility of the Department of Defense (DOD). In January 1995, the Secretary of Defense established the IW Executive Board to facilitate "the development and achievement of national information warfare goals."3
I want to focus here on protection of civilian assets, which, unfortunately, can be difficult to separate from military assets. An enemy might attack the United States exactly by attacking the civilian infrastructure. This is just an example of a more general tendency. In his book The Transformation of War, Martin van Creveld argues that the low-intensity conflicts that have become the norm since World War II will become far more prevalent and will spread to developed countries, instead of being confined primarily to the Third World. He writes: "As the spread of low-intensity conflict causes trinitarian structures to come tumbling down, strategy will focus on obliterating the existing line between those who fight and those who watch, pay, and suffer." By trinitarian structures, van Creveld means the division among the state, the military, and the people.4
Here are two specific examples of the difficulty in separating military assets from civilian assets. Approximately 95 percent of all military communications are routed through commercial lines. We buy most of the chips used in military systems from commercial vendors, many of whom are located in foreign countries. Why might
infowar be the weapon of choice of foreign or domestic terrorists, and of foreign countries, small and large? Here is a partial list of reasons:
- The United States is the only current superpower. As the Gulf War showed, it is foolish to challenge us in conventional war.
- Since we have the most advanced NII, it is the most vulnerable to attack. (Of course, this would give us an advantage in offensive IW.)
- The price of entering into IW is low.
- The learning curve is very steep.
- It might be difficult to detect who attacked us or even whether there was malicious intent.
Who should be in charge of protecting the civilian infrastructure against attack? I believe that there needs to be strong government leadership and that it should be located in the executive branch. Should the lead role be played by an existing entity, a combination of existing entities properly coordinated, or a government structure created for this purpose? This is an exceptionally complex and important question that I will not pursue here.
It has been argued that the NII will acquire at least partial immunity due to repeated attacks, analogous to a biological organism. It has also been argued that the problem can be left to the private sector to solve. Although I believe that the private sector has a very important role to play, I am not convinced it can do this on its own. Coordination between the government and the private sector will be another difficult and important area to address.
Protecting ourselves against infowar may require the careful balancing of our desire for liberty and privacy with our wish for security. As the Clipper Chip illustrates, the conflicting demands of privacy, commerce, and security can generate strong tensions. It is a particularly difficult issue for democracies, far more difficult than for countries that do not place a high value on privacy and liberty.
How imminent is a serious IW attack? How much time do we have to prepare? Unlike most conventional warfare, in which there is a visible buildup of forces, an IW attack may come without warning. Although we should think carefully about how to meet the threat, I believe that time is of the essence. I suggest that no issue is more important to the nation than the defense of our national information infrastructure.
JOHN MAJOR: Do you see a dichotomy between the military's desire, on the one hand, to restrict the level of security that private industry has access to, and on the other, to pass on to private industry the responsibility for protecting its information from international attack?
JOSEPH TRAUB: It is a minefield of difficulties. Some of the major companies, such as AT&T and MCI, for example, say, "Do not worry, we can take care of it." You know how difficult it has been to conduct the Board's study on encryption. I have no special wisdom on this subject, but I am sure there are people in the room who do.
HOWARD FRANK: I would like to react to the original question because it contains an assumption that is not necessarily true—that the military has a desire to reduce or restrict the amount of protection afforded to the civilian economy and infrastructure. I do not think this is true. There has been an ongoing debate, which is certainly not uniform throughout DOD, let alone among the major political leaders, about certain export policies and so on. In general, there is no firm policy that says that the civilian sector should have less security or protection than DOD.
In the first place, I think the problem has been one of technology not responding to the needs because they are basically invisible. We are more aware of this now. Maybe you should have said something in 1986 or 1987, because it appears to me that this came out of the blue. For instance, if you look at the research community, it is devoid of original ideas on this aspect of the problem. It is only now that people are beginning to think about what we might do as a nation in terms of a research agenda.
Second, the issue of privacy is one of the great perplexities in law enforcement. Resolving this issue is going to take quite a while, but it seems to be moving in the right direction across the government.
Third, the civilian sector, industry by industry, varies greatly in terms of what it believes the vulnerabilities
are. In fact, I participated in an NRC study in 1989-1990 that looked at the security preparedness of the telephone system. When we began that study, there was vast disagreement about whether there were any vulnerabilities whatsoever. Then, in the middle of the study, there was a major service disruption in Hinsdale, Illinois, and a failure in Bedford, Massachusetts, followed by another failure. So the reaction was, "Oh, well, maybe we really made a mistake. Maybe we are really vulnerable."
Other parts of the infrastructure, like the power industry, have their heads buried in the sand; there is virtually no security at all. So this is a very complex question, and it is not a dichotomy. It is a continuum of issues, many of which are economic. For instance, given the fact that we do not have a nationwide server for infrastructure, let us say telecommunications, no major trade can now be made that says this is in the national interest and therefore you must do it—the way we could do in providing universal telephone service 50 years ago. This complicates the problem a lot.
SIDNEY KARIN: I want to thank you for raising the information warfare issue. I think it is absolutely critical. You said that determining who should be in charge is very complex. I think that answering the question of whether anybody should be in charge is down the road. We have not gotten there yet.
I think the threat is real. I think in some nontrivial way we are under attack today. There is lots of evidence that security breaches have been taking place for the past several years. The major problem is that we, as a society, have not recognized that there is a threat, a real danger, and that the consequences could be quite serious.
So I submit that the issue is not just a quibbling over civilian versus military security. Although I have a very strong position on this issue, I will not get into it at the moment. The issue is that, as a society, we have not agreed that security is a serious problem. Until we agree on this, there will not be any solutions—no matter how hard anybody works at trying to implement anything, no matter what structures are imposed by anybody to deal with it.
The first thing that needs to be done is to raise everybody's consciousness that something bad could happen, and that there are people trying to make it happen for various reasons. I commend you for raising the issue.
BUTLER LAMPSON: The basic fact about security is that it is expensive; it is a pain in the neck; and people are going to implement it only in response to convincing evidence that there is a problem. The only way you are going to get that evidence is when someone you think is in a position similar to yours gets badly hurt.
In my view, all this discussion about what we ought to do and how much we ought to worry about the threat is entirely beside the point because nobody is actually going to do anything until there is some serious damage done. The fact is that no serious damage has been done. It is fine for us to think about how we might respond once the motivation is there to respond, but I think trying to raise the level of motivation is completely pointless.
KARIN: What would constitute evidence in your judgment that somebody has been badly hurt? What would it take for your organization to decide that one of its competitors was badly hurt? What is a hypothetical incident that would convince you?
LAMPSON: Something that happens that costs you a lot of money. Bankruptcy would do it. That would definitely have an impact.
KARIN: How large a company would have to go bankrupt before your company would take notice, recognize there is a serious threat, and decide to change its behavior?
LAMPSON: How large a company would have to go bankrupt? I do not know. It is an interesting question. Since no companies have gone bankrupt yet, right now the question is academic. If you believe that some companies have gone bankrupt as a result of information warfare and you want to promote action, then I strongly urge you to find out who those companies are and to publicize the situation clearly. This will have far more effect than anything else that you or Joe or anyone else might say.
ROBERT BONOMETTI: I would like to make the suggestion that the problem actually is more extensive than just a deliberate, determined attack against the infrastructure. As society becomes more dependent on this infrastructure, natural disasters also become of great concern because the social fabric will be ripped apart and become dysfunctional in the aftermath of a disaster.
There are examples we can look at, such as the aftermath of Hurricane Andrew in Florida, that provide some lessons learned. One very trivial but interesting fact is that people were unable to get cash because everyone was so dependent on ATMs (automatic teller machines). After the hurricane the infrastructure was not there, and it was a problem for some time.
ANDRE VAN TILBORG: I wonder if you might comment on how much of the security problem might be related to not having the analogue of something like building codes for information and telecommunications systems, versus how much really requires deep, new insights and research. If you had, for instance, a national electrical code—not for electricity or electrical appliances or Underwriters Laboratory, but for computing systems and telecommunications systems—you might be able to cover a large fraction of the troublesome areas in ensuring that your systems remain stable and work. You would still have that part where a very determined adversary can get through, even though you have a good building code. I wonder what your thoughts on that might be.
TRAUB: I think what you are suggesting is standards. It has been suggested that there be a core communications system in case of a national emergency, and there has been some discussion about that. The things we have with which to protect our homes will at least keep out the amateurs, although they will not keep out a professional, determined burglar. The big problem is that everything is changing so quickly that this is an almost impossible area, I believe, to standardize. Standardization, it seems to me, requires a certain maturity and a certain stability. I am not sure we can do that in this area.
ROBERT KAHN: The good news is that the notion of the national information infrastructure is in the public consciousness. The bad news is that we really do not know what it is or might not recognize it if we saw it. To some, the NII is 500 channels of cable TV and to others it is the Library of Congress on every desktop—in some ways very mutually incompatible goals.
Some people think we have always had an information infrastructure, or at least maybe since 1844 when the telegraph was invented, if you want to focus on electrons. Others think we clearly have it now. Still others will wake up in 10 years and be totally shocked that we do not have an infrastructure yet and have been talking about it all these years.
It seems to me that the one thing that has really been missing, apart from understanding what it is, is any notion about getting coherence among all the pieces, so that the infrastructure really becomes the mechanism to lower the barriers to productivity in a broad sense. I do not think we are there yet. The big objective over the next decade, perhaps many decades, is trying to figure out how to achieve interoperability to lower these barriers. By getting coherence in the system, you make it more of a target for the kind of information warfare that you are talking about.
So my question to you is, How do we go about designing this coherence for interoperability into the system, while at the same time worrying about protecting against the kind of information warfare you are discussing as a social process in this country?
TRAUB: That is a very good question, Bob. I am sorry, but my time is up.
WILLIAM WULF: Let me just make a couple of comments. First of all, the balancing of privacy and societal protection was mentioned. Probably the most sensitive report that CSTB has ever undertaken is going to be released imminently, and it is one that was requested by Congress. 5 It addresses national cryptography policy. It will be a completely unclassified report. This fact is very, very important. There is not going to be a classified annex to the report. We wanted the report to be completely unclassified.
My second comment has to do with the national information infrastructure and natural disasters. We absolutely agree with Dr. Bonametti's remarks. If you look in the CSTB brochure, you will see that one of CSTB's 1996 studies (Computing and Communications in the Extreme: Research for Crisis Management and Other Applications) is looking at how we can use information technology to save lives and property.
In his early remarks, Joe pointed out that, from the outset, the Board has been concerned with both technology and policy issues. If anything—and this is a personal perception—the increasing recognition among people in both the executive and the legislative branches of the relevance of information technology to virtually every problem that the country faces has reinforced the correctness of that original decision.