Cryptography expert Martin Hellman, co-inventor of Diffie-Hellman public-key encryption, says he never encrypts his e-mail. It's just too much trouble.
"There's a lack of integrated, automatic, transparent crypto," says Hellman, professor emeritus of electrical engineering at Stanford University. "If security is an add-on that people have to do something special to get, then most of the time they will not get it."
But there are developments afoot that during the next five years may bring some of the integration, automation and transparency Hellman is looking for.
"Cryptographic operations will disappear into the infrastructure," predicts Tom Berson, a principal scientist at Palo Alto Research Center Inc. (PARC) in California. "The complexities of cryptography and of cryptographic key management will be hidden from users."
Encryption will be seamlessly integrated into virtually every computing device and piece of communications software, Berson says.
And it will happen not a moment too soon. Increased connectivity in general and the rapid rise in wireless communications in particular are leaving users with files, messages and telephone conversations vulnerable to loss of privacy and confidentiality.
PARC researchers have several projects under way to encourage the use of encryption. One, called Quicksilver, is a program aimed at convincing vendors and users that cryptography is no longer slow and difficult to use. In its Quicksilver Manifesto, PARC exhorts users to "demand that your interests take priority over obsolete beliefs about what can and cannot be done to secure your information."
PARC is also working on making encryption more user-friendly. For example, it has proposed an intuitive scheme for user authentication in ad hoc wireless networks. The scheme uses public-key cryptography, but without a public-key infrastructure (PKI) that requires users to have digital certificates signed by a trusted third party.
"The problem with PKI is the I, not the PK," says PARC researcher Diana Smetters.
Stanford researcher Dan Boneh is developing a public-key system based on Identity-Based Encryption, by which users can use their e-mail addresses as their public keys. Recipients of e-mail encrypted by the system wouldn't need pre-established keys or certificates. The system would also allow the creation of messages that can be read only at a specified future time.
And Anna Lysyanskaya, a computer science graduate student at MIT, is working with IBM to develop privacy-oriented "anonymous credentials." For example, users could sign up for an online service without divulging their identities. The service provider could verify that a user is registered, but it couldn't track his activities and track them back to him.
Years of Development
Encryption is based on complex mathematics, and it often takes years for experts to satisfy themselves that there are no serious flaws in a cryptosystem. For example, the new federal Advanced Encryption Standard (AES) algorithm, which is replacing the old, less secure Data Encryption Standard, is based on techniques developed in the mid-1990s. But the U.S. Department of Defense won't widely deploy AES cryptography until 2007.
"Within three years, we'll see a reasonable deployment of AES," including in Sun Microsystems Inc. products, predicts Susan Landau, a senior staff engineer at Sun. She says that Sun is also likely to embrace Elliptic Curve Cryptography (ECC), a public-key cryptography created in 1985.
ECC has advantages over the more widely used RSA encryption scheme. These advantages principally include shorter key lengths and faster performance for a given level of security. Landau says that may make it the cryptosystem of choice for mobile devices short on processing power and memory.
"Pretty much all of the new standards coming out of places like the [Internet Engineering Task Force] are being written to support ECC," Smetters says. "You will start to see people adopt ECC as an option; after it's a pretty common option, it will start to become the default."
Cryptosystems based on existing algorithms are likely to remain in use for many years. They can be strengthened against brute-force attacks launched by ever-faster computers by increasing the length of the encryption keys they use. But, experts say, there is a threat looming for all public-key systems: ultra-fast, massively parallel quantum computers. Very primitive quantum computers have already been built at IBM.
However, Hellman says it's likely to be at least 10 to 15 years before quantum computers can be scaled up to crack modern encryption keys. "So, while prudence dictates that we keep our eye on quantum computing, there is no immediate threat," he says.