The Cryptography Wars
By Kenneth W. Dam and Herbert S. Lin
The problem is this: Individuals and businesses have a legitimate need to protect information from interlopers through the use of cryptography. But law enforcement officials fear that drug dealers and terrorists using cryptography will be able to thwart legally authorized surveillance and search warrants. National security officials are concerned that encrypted communications may frustrate intelligence collection against parties that might be building nuclear or biological weapons for use against the United States.
But this conflict describes only part of the picture. After all, protecting a company's proprietary information against industrial spies is very much a part of law enforcement. Protecting critical national information systems and networks against unauthorized intruders is a key responsibility of national security. Thus, the use of cryptography can help law enforcement and national security as well as hinder them, as we pointed out in a recent report of the National Research Council (NRC).
In a June 10 editorial ["Global Village Cops?"], The Post disagreed with that report, suggesting that law enforcement and national security interests require that current restrictions on cryptography be maintained. The Post asserted that it is "too soon" to accept that encryption can help law enforcement and national security.
Reasonable people can disagree about the weights used in balancing cryptography's help vs. hindrance to law enforcement and national security. But arguing that it is premature to believe some uses of encryption do benefit law enforcement and national security simply denies reality. We emphatically reject The Post's implication that we "sacrificed" law enforcement and national security considerations in favor of economic interests.
We also counsel against ineffective and self-defeating "solutions" sometimes proposed such as bans on the use of cryptography and stringent controls on exports of cryptography. A ban on encryption cannot be enforced, and it would put American companies at a significant disadvantage in a global information society.
As for export controls on cryptography, they have for many years helped to deny the benefits of cryptography to foreign adversaries. But today's controls work to reduce the domestic availability of strong encryption and restrict U.S. sellers of technology from exporting products with such capabilities, even when foreign customers can buy them elsewhere. We believe export controls should be progressively relaxed, thereby strengthening the market leadership of U.S. vendors, itself important to our national security.
To counter criminals and terrorists, the Clinton administration has pushed a plan whereby keys enabling messages to be read would be placed "in escrow" with a third party. In theory, this sort of "escrowed encryption" would allow legitimate users to protect their information while also giving law enforcement authorities the access they need. Critics feel this places the needs of law enforcement and national security authorities for information gathering and surveillance above the needs of businesses and individuals for information protection.
Escrowed encryption seems to us to be a promising technology. But it is unproven: For now, government should view it as a tentative concept to be explored. Instead, the administration has sought in recent months to make the adoption of escrowed encryption the quid pro quo for a liberalization of export controls. We believe such linkage is unwarranted.
Cryptography policy over the long term rests on four fundamentals: First, there is no cryptographic nirvana. The trade-offs between better information security and better government access to information are real. No politically feasible policy will fully satisfy all stakeholders.
Second, individuals, business and government need a high degree of information security in today's world. Trade-offs that might have been appropriate in an information-poor society are not necessarily appropriate for an information-rich one.
Third, government should help law enforcement and national security authorities adjust to the technical realities of the information age. Support for new technical capabilities will almost certainly help these authorities more than promoting escrowed encryption to a resistant market.
Fourth, and perhaps most important, classified information is not necessary to carry out a rational discussion of national cryptography policy. The Post's editorial suggested that the law enforcement and national security cases for maintaining current restrictions on exports of cryptography depend on classified arguments. However, the cleared members of the NRC study committee (13 out of the 16) were given access to this information, and concluded that the considerable strengths of the law enforcement and national security arguments on encryption can be demonstrated on the basis of information that is in the public domain. Indeed, only a fully open and inclusive public discussion can lead to the national consensus upon which any successful cryptography policy will depend.
Acceptance of these principles could lead to a cease-fire among unhappy stakeholders. Such a cease-fire would be an important first step toward the national cryptography policy our nation needs so urgently.
Kenneth W. Dam, a former deputy secretary of state, recently chaired the National Research Council Committee to Study National Cryptography Policy. Herbert S. Lin was study director for the committee.