On The Net Column
Monday, Sept. 11, 1995
Peter H. Lewis
In terms of its ability to raise the nation's blood pressure, the debate over data encryption has not yet reached the same levels as gun control.
But last week the Clinton Administration appeared to set the stage for an equally divisive national debate over the rights of businesses and individuals to keep secrets when using telephones, computers and other forms of electronic communications.
In two days of public hearings last week in Gaithersburg, Md., home of the National Institute of Standards and Technology (NIST), the Administration in effect unveiled its long-awaited proposals to relax restrictions on the export of cryptographic software.
In effect, the Administration drew a line in the virtual sands of cyberspace, signaling that it is willing to permit Americans to put stronger cryptographic locks on their electronic data only if a spare key to those locks is made available on demand to law enforcement agencies.
There looms the conflict. Although the current debate is about export controls on an esoteric form of software that most Americans do not use, the "export" issue is ultimately irrelevant in today's era of global electronic voice and data networks, where passwords, not passports, are checked at the gates. Simply placing a common privacy program on an Internet-connected computer in Austin, Tex., is effectively no different from sending a shrink-wrapped copy of the program to Moscow.
The real issue is how much privacy the Government is willing to allow its own citizens, and the latest word from the Clinton Administration is that the right to electronic privacy, like the right to bear arms, is not absolute.
***
Cryptography is the science of secret writing. In this digital era, secret writing applies not just to notes handed from one spy to another, but also to telephone calls between individuals, funds transfers between banks, bank and credit card records, electronic mail, faxes, and an endless variety of computer files.
The Clinton Administration has been clear and consistent in outlining its basic position on cryptographic systems. The goal is to allow American citizens and companies to use the strongest possible cryptographic technology, while at the same time preserving the ability of law enforcement agencies to perform court-authorized wiretaps as part of the effort to catch drug dealers, terrorists, child pornographers and other miscreants.
In other words, it favors strong cryptography, but not too strong.
One way to measure the strength of cryptographic software is the length of the software key necessary to encode and decode a message. The longer the key, in terms of digital bits, the harder it is for an unauthorized user to decipher the message.
In recent years, the Government has generally permitted Americans to export cryptographic software with key lengths up to 40 bits. Experts say that 40-bit keys are secure from casual snooping, but will fall quickly to a determined codebreaker. The fact that the Government allows 40-bit encryption systems to be exported is a pretty good indication that the National Security Agency can break them easily.
There are literally hundreds of stronger cryptography programs readily available outside the United States, and these stronger programs are attractive to businesses that want to safeguard their data from Internet bandits and corporate and government spies.
Last week, after more than a year of intense analysis of the software export controls issue, the Government unveiled what it said was the best possible compromise.
Under the new policy, companies can export encryption algorithms using 64-bit keys, which are much more secure, but only if spare keys are made available to law enforcement agents under standard legal procedures. Otherwise, the 40-bit limit continues to apply.
The "spare key" technology, officially known as key escrow, is anathema to many privacy advocates who fear Government abuses. The Government first proposed a key escrow system with its so-called Clipper Chip, a technology that failed to win acceptance even as a voluntary standard.
Unlike Clipper, which was based on a classified algorithm called Skipjack that only a few people outside the Government were allowed to examine, the new policy allows people to use any algorithm they choose -- as long as it uses a key no larger than 64 bits, and as long as the keys are entrusted to a domestic third party accessible to the Government, and as long as the key escrow mechanisms cannot be readily altered or bypassed.
Also unlike Clipper, which required a special tamper-proof microprocessor that would have added cost, complexity and extra power requirements to communications devices, the new proposals can be implemented entirely in software.
Key escrow systems make a lot of sense for most American companies, at least for internal use. Having a spare set of keys lessens the risk of a disgruntled employee or saboteur locking up vital company files.
But key escrow is also unpopular with American computer and software companies, who say it prevents them from competing against foreign companies that have no similar constraints, and with many multinational corporations, who say it prevents them from working with foreign companies that don't especially care for the idea of Uncle Sam holding the keys to their data banks.
"If this was intended to be any sort of compromise, I don't think it achieved its end," said Whitfield Diffie, a Distinguished Engineer at Sun Microsystems who attended the meetings. "I didn't see anybody who was enthusiastic."
***
Raymond G. Kammer, deputy director of NIST, suggested that the hearings last week were intended to elicit public comment, and that the Administration's final positions on cryptographic policy are still under analysis.
However, the emergence of key escrow issues at the NIST proceedings suggest that key escrow is emerging as a non-negotiable demand by some factions of the Clinton Administration, especially the Justice Department and the Federal Bureau of Investigation, led by Louis Freeh. Mr. Freeh sincerely believes that data encryption is a weapon, and has publicly called for domestic restrictions on civilian cryptography.
"If this fails," said one observer familiar with the Administration's thinking on the proposed change in cryptographic policy, "it's going to lead to a very devisive debate. And the irony, for libertarians who oppose key escrow, is that if it fails, I am convinced that Louis Freeh cannot be true to his job without proposing domestic controls on data encryption."
"He's not going to give up without a fight, and neither is the Justice Department," said the observer, who spoke on the condition he not be identified.
Others say they do not think the Clinton Administration has yet arrived at a concrete position, even after more than a year of study and debate. "I don't think it's a final offer," said John Gilmore, a member of the board of directors of Cygnus Support Inc., a computer company in Mountain View, Calif. "It looks to me like a weak strawman, a first offer, a proposal to dance."
The question is whether American citizens and businesses have the patience to wait for the music to start. And the issue may be moot, anyway.
"The Internet Architecture Board has specifically decided to ignore export controls in designing the security infrastructure for the next generation of Internet protocols," Mr. Gilmore said. "The Internet of 1998 will provide automatic, secure, and fully private communication, without key escrow, internationally. The protocols will be implemented independently in a dozen different countries."
In other words, the international Internet community is already planning to jump over the new line in the sand drawn last week by the Clinton Administration. Cryptography that is stronger and better than the Government's proposed system will become an integral part of the Internet, and American companies and individuals would be foolish not to use it.
At that point, millions of Americans will come in direct conflict with Government policy, and the popular gun-control bumper sticker may be replaced by one that says, "If cryptography is outlawed, only outlaws will have cryptography."
Peter H. Lewis, P.O. Box 162490, Austin, TX, 78716-2490
(512) 328-8258 ... "All the Disclaimers That Fit in Print"
plewis@nytimes.com