Why Clipper Is Good For You
By Stewart A. Baker, Chief Counsel for the NSA
With all the enthusiasm of Baptist ministers turning their Sunday pulpits over to the Devil, the editors of Wired have offered me the opportunity to respond to some of the urban folklore that has grown up around key escrow encryption -- also known as the Clipper Chip.
Recently the Clinton administration has announced that federal agencies will be able to buy a new kind of encryption hardware that is sixteen million times stronger than the existing federal standard known as DES. But this new potency comes with a caveat. If one of these new encryption devices is used, for example, to encode a phone conversation that is subject to a lawful government wiretap, the government can get access to that device's encryption keys. Separate parts of each key are held by two independent "escrow agents," who will release keys only to authorized agencies under safeguards approved by the attorney general. Private use of the new encryption hardware is welcome but not required. That's a pretty modest proposal. Its critics, though, have generated at least seven myths about key escrow encryption that deserve answers.
MYTH NUMBER ONE: Key escrow encryption will create a brave new world of government intrusion into the privacy of Americans.
Opponents of key escrow encryption usually begin by talking about government invading the privacy of American citizens. None of us likes the idea of the government intruding willy-nilly on communications that are meant to be private.
But the key escrow proposal is not about increasing government's authority to invade the privacy of its citizens. All that key escrow does is preserve the government's current ability to conduct wiretaps under existing authorities. Even if key escrow were the only form of encryption available, the world would look only a little different from the one we live in now.
In fact, it's the proponents of widespread unbreakable encryption who want to create a brave new world, one in which all of us -- crooks included -- have a guarantee that the government can't tap our phones. Yet these proponents have done nothing to show us that the new world they seek will really be a better one.
In fact, even a civil libertarian might prefer a world where wiretaps are possible. If we want to catch and convict the leaders of criminal organizations, there are usually only two good ways to do it. We can "turn" a gang member -- get him to testify against his leaders. Or we can wiretap the leaders as they plan the crime.
I once did a human rights report on the criminal justice system in El Salvador. I didn't expect the Salvadorans to teach me much about human rights. But I learned that, unlike the US, El Salvador greatly restricts the testimony of "turned" co-conspirators. Why? Because the co-conspirator is usually "turned" either by a threat of mistreatment or by an offer to reduce his punishment. Either way, the process raises moral questions -- and creates an incentive for false accusations.
Wiretaps have no such potential for coercive use. The defendant is convicted or freed on the basis of his own, unarguable words.
In addition, the world will be a safer place if criminals cannot take advantage of a ubiquitous, standardized encryption infrastructure that is immune from any conceivable law enforcement wiretap. Even if you're worried about illegal government taps, key escrow reinforces the existing requirement that every wiretap and every decryption must be lawfully authorized. The key escrow system means that proof of authority to tap must be certified and audited, so that illegal wiretapping by a rogue prosecutor or police officer is, as a practical matter, impossible.
MYTH NUMBER TWO: Unreadable encryption is the key to our future liberty.
Of course there are people who aren't prepared to trust the escrow agents, or the courts that issue warrants, or the officials who oversee the system, or anybody else for that matter. Rather than rely on laws to protect us, they say, let's make wiretapping impossible; then we'll be safe no matter who gets elected.
This sort of reasoning is the long-delayed revenge of people who couldn't go to Woodstock because they had too much trig homework. It reflects a wide -- and kind of endearing -- streak of romantic high-tech anarchism that crops up throughout the computer world.
The problem with all this romanticism is that its most likely beneficiaries are predators. Take for example the campaign to distribute PGP ("Pretty Good Privacy") encryption on the Internet. Some argue that widespread availability of this encryption will help Latvian freedom fighters today and American freedom fighters tomorrow. Well, not quite. Rather, one of the earliest users of PGP was a high-tech pedophile in Santa Clara, California. He used PGP to encrypt files that, police suspect, include a diary of his contacts with susceptible young boys using computer bulletin boards all over the country. "What really bothers me," says Detective Brian Kennedy of the Sacramento, California, Sheriff's Department, "is that there could be kids out there who need help badly, but thanks to this encryption, we'll never reach them."
If unescrowed encryption becomes ubiquitous, there will be many more stories like this. We can't afford as a society to protect pedophiles and criminals today just to keep alive the far-fetched notion that some future tyrant will be brought down by guerrillas wearing bandoleers and pocket protectors and sending PGP-encrypted messages to each other across cyberspace.
MYTH NUMBER THREE: Encryption is the key to preserving privacy in a digital world.
Even people who don't believe that they are likely to be part of future resistance movements have nonetheless been persuaded that encryption is the key to preserving privacy in a networked, wireless world, and that we need strong encryption for this reason. This isn't completely wrong, but it is not an argument against Clipper.
If you want to keep your neighbors from listening in on your cordless phone, if you want to keep unscrupulous competitors from stealing your secrets, even if you want to keep foreign governments from knowing your business plans, key escrow encryption will provide all the security you need, and more.
But I can't help pointing out that encryption has been vastly oversold as a privacy protector. The biggest threats to our privacy in a digital world come not from what we keep secret but from what we reveal willingly. We lose privacy in a digital world because it becomes cheap and easy to collate and transmit data, so that information you willingly gave a bank to get a mortgage suddenly ends up in the hands of a business rival or your ex-spouse's lawyer. Restricting these invasions of privacy is a challenge, but it isn't a job for encryption. Encryption can't protect you from the misuse of data you surrendered willingly.
What about the rise of networks? Surely encryption can help prevent password attacks like the recent Internet virus, or the interception of credit card numbers as they're sent from one digital assistant to another? Well, maybe. In fact, encryption is, at best, a small part of network security.
The real key to network security is making sure that only the right people get access to particular data. That's why a digital signature is so much more important to future network security than encryption. If everyone on a net has a unique identifier that others cannot forge, there's no need to send credit card numbers -- and so nothing to intercept. And if everyone has a digital signature, stealing passwords off the Net is pointless. That's why the Clinton administration is determined to put digital signature technology in the public domain. It's part of a strategy to improve the security of the information infrastructure in ways that don't endanger government's ability to enforce the law.
MYTH NUMBER FOUR: Key escrow will never work. Crooks won't use it if it's voluntary. There must be a secret plan to make key escrow encryption mandatory.
This is probably the most common and frustrating of all the myths that abound about key escrow. The administration has said time and again that it will not force key escrow on manufacturers and companies in the private sector. In a Catch-22 response, critics then insist that if key escrow isn't mandated it won't work.
That misunderstands the nature of the problem we are trying to solve. Encryption is available today. But it isn't easy for criminals to use; especially in telecommunications. Why? Because as long as encryption is not standardized and ubiquitous, using encryption means buying and distributing expensive gear to all the key members of the conspiracy. Up to now only a few criminals have had the resources, sophistication, and discipline to use specialized encryption systems.
What worries law enforcement agencies --what should worry them -- is a world where encryption is standardized and ubiquitous: a world where anyone who buys an US$80 phone gets an "encrypt" button that interoperates with everyone else's; a world where every fax machine and every modem automatically encodes its transmissions without asking whether that is necessary. In such a world, every criminal will gain a guaranteed refuge from the police without lifting a finger.
The purpose of the key escrow initiative is to provide an alternative form of encryption that can meet legitimate security concerns without building a web of standardized encryption that shuts law enforcement agencies out. If banks and corporations and government agencies buy key escrow encryption, criminals won't get a free ride. They'll have to build their own systems -- as they do now. And their devices won't interact with the devices that much of the rest of society uses. As one of my friends in the FBI puts it, "Nobody will build secure phones just to sell to the Gambino family."
In short, as long as legitimate businesses use key escrow, we can stave off a future in which acts of terror and organized crime are planned with impunity on the public telecommunications system. Of course, whenever we say that, the critics of key escrow trot out their fifth myth:
MYTH NUMBER FIVE: The government is interfering with the free market by forcing key escrow on the private sector. Industry should be left alone to develop and sell whatever form of encryption succeeds in the market.
In fact, opponents of key escrow fear that businesses may actually prefer key escrow encryption. Why? Because the brave new world that unreadable encryption buffs want to create isn't just a world with communications immunity for crooks. It's a world of uncharted liability. What if a company supplies unreadable encryption to all its employees, and a couple of them use it to steal from customers or to encrypt customer data and hold it hostage? As a lawyer, I can say it's almost certain that the customers will sue the company that supplied the encryption to its employees. And that company in turn will sue the software and hardware firms that built a "security" system without safeguards against such an obvious abuse. The only encryption system that doesn't conjure up images of a lawyers' feeding frenzy is key escrow.
But there's a second and even more compelling reason why the key escrow initiative can't fairly be characterized as interfering with private enterprise: The encryption market has been more or less created and sustained by government. Much of the market for encryption devices is in the public sector, and much of the encryption technology now in widespread use in the private sector was funded, perfected, or endorsed by the federal government.
And not by accident, either. Good encryption is expensive. It isn't just a matter of coming up with a strong algorithm, although testing the strength of an algorithm can be enormously time-consuming. The entire system must be checked for bugs and weaknesses, a laborious and unglamorous process. Generally, only the federal government has been willing to pay what it costs to develop secure communications gear. That's because we can't afford to have our adversaries reading our military and diplomatic communications.
That's led to a common pattern. First, the government develops, tests, or perfects encryption systems for itself. Then the private sector drafts along behind the government, adopting government standards on the assumption that if it's good enough for the government's information, it's good enough to protect industry's.
As encryption technology gets cheaper and more common, though, we face the real prospect that the federal government's own research, its own standards, its own purchases will help create the future I described earlier -- one in which criminals use ubiquitous encryption to hide their activities. How can anyone expect the standard-setting arms of government to use their power to destroy the capabilities of law enforcement -- especially at a time when the threat of crime and terror seems to be rising dramatically?
By adopting key escrow encryption instead, the federal government has simply made the reasonable judgment that its own purchases will reflect all of society's values, not just the single-minded pursuit of total privacy.
So where does this leave industry, especially those companies that don't like either the 1970s-vintage DES or key escrow? It leaves them where they ought to be -- standing on their own two feet. Companies that want to develop and sell new forms of unescrowed encryption won't be able to sell products that bear the federal seal of approval. They won't be able to ride piggyback on federal research efforts. And they won't be able to sell a single unreadable encryption product to both private and government customers.
Well, so what? If companies want to develop and sell competing, unescrowed systems to other Americans, if they insist on hastening a brave new world of criminal immunity, they can still do so -- as long as they're willing to use their own money. That's what the free market is all about.
Of course, a free market in the US doesn't mean freedom to export encryption that may damage US national security. As our experience in World War II shows, encryption is the kind of technology that wins and loses wars. With that in mind, we must be careful about exports of encryption. This isn't the place for a detailed discussion of controls, but one thing should be clear: They don't limit the encryption that Americans can buy or use. The government allows Americans to take even the most sophisticated encryption abroad for their own protection. Nor do controls require that software or hardware companies "dumb down" their US products. Software firms have complained that it's inconvenient to develop a second encryption scheme for export, but they already have to make changes from one country to the next -- in language, alphabet, date systems, and handwriting recognition, to take just a few examples. And they'd still have to develop multiple encryption programs even if the US abolished export controls, because a wide variety of national restrictions on encryption are already in place in countries from Europe to Asia.
MYTH NUMBER SIX: The National Security Agency is a spy agency; it has no business worrying about domestic encryption policy.
Since the National Security Agency has an intelligence mission, its role in helping to develop key escrow encryption is usually treated as evidence that key escrow must be bad security. In reality, though, NSA has two missions. It does indeed gather intelligence, in part by breaking codes. But it has a second, and oddly complementary, mission. It develops the best possible encryption for the US government's classified information.
With code breakers and code makers all in the same agency, NSA has more expertise in cryptography than any other entity in the country, public or private. It should come as no surprise, therefore, that NSA had the know- how to develop an encryption technique that provides users great security without compromising law enforcement access. To say that NSA shouldn't be involved in this issue is to say the government should try to solve this difficult technical and social problem with both hands tied behind its back.
MYTH NUMBER SEVEN: This entire initiative was studied in secret and implemented without any opportunity for industry or the public to be heard.
This is an old objection, and one that had some force in April of 1993, when the introduction of a new AT&T telephone encryption device required that the government move more quickly than it otherwise would have. Key escrow was a new idea at that time, and it was reasonable for the public to want more details and a chance to be heard before policies were set in concrete. But since April 1993, the public and industry have had many opportunities to express their views. The government's computer security and privacy advisory board held several days of public hearings. The National Security Council met repeatedly with industry groups. The Justice Department held briefings for congressional staff on its plans for escrow procedures well in advance of its final decision. And the Commerce Department took public comment on the proposed key escrow standard for 60 days.
After all this consultation, the government went forward with key escrow, not because the key escrow proposal received a universally warm reception, but because none of the proposal's critics was able to suggest a better way to accommodate society's interests in both privacy and law enforcement. Unless somebody comes up with one, key escrow is likely to be around for quite a while. That's because the only alternative being proposed today is for the government to design or endorse encryption systems that will cripple law enforcement when the technology migrates -- as it surely will -- to the private sector. And that alternative is simply irresponsible.
For more information on the Clipper standard you can access Wired's Clipper archive via the following WIRED Online services.
Stewart A. Baker is the National Security Agency's top lawyer. He worked briefly as Deputy General Counsel of the Education Department under President Jimmy Carter, and he practiced international law at Steptoe & Johnson, in Washington, DC. He has been at the NSA since 1992.