The New York Times today, in the sort of story that reminds us of why newspapers still matter, led with a lengthy report on the Obama Administration’s developing plans to enhance the government’s ability to eavesdrop on internet communications. The devil is always in the details in these things, and I’ll reserve judgment on the substance of the plan until we see what the Administration really wants. But to those of us who have been around the tech world for a couple of decades or more, the story had a disturbingly familiar ring.
Back in 1993, the Clinton Administration was worried that the growing availability of encryption software would make it difficult for law enforcement to keep tabs on the bad guys (back then, the focus was more on voice telephony than text.) So the Administration, with Vice President Al Gore leading the way, proposed a scheme of escrowed key encryption. As a carrot, businesses would be able to export products containing encryption software freely–encryption technology was then subject to stringent export controls–but companies or individuals would have to provide the government with a copy of a private key that could be used to decrypt data with a court order (the exact nature of how this escrow would work changed several times over the proposal’s short and unhappy life.) Encryption was to be implementyed through a classified algorithm called Skipjack but the scheme was better known as Clipper, which was actually the name of a hardware implementation designed for encrypting voice phones.
The Skipjack project was highly controversial, in no small part because the underlying algorithm was classified and because, at least under initial plans, could only be implemented in secure hardware. At a time when online commerce was just beginning to take off, many feared that securing transactions with such a cumbersome and expensive scheme would cause online buisness to be stillborn. And, of course, there was considerable fear about potential government abuse of its ability to eavesdrop.
In the end, the project fell of the weight of its own technical flaws. The most serious blow was struck by Matt Blaze (then of AT&T Research, now at the University of Pennsylvania), who demonstrated in a 1994 paper, that a critical component of the escrow scheme, the Law Enforcement Access Field, would easily be spoofed, allowing unauthorized retrieval of escrowed keys. In 1996, the enterprise was formally abandoned and the Skipjack algorithm was declassified in 1998.
Eventually, the U.S. government bowed to the inevitable and not only ended most export controls on commercial cryptography, but sponsored a competition to choose a replacement for the aging Data Encryption Standard. The result was the Advanced Encryption Standard, which is widely used to secure online transactions and encrypt sensitive data (including most data exchanged using Secure Socket Layers.)
What goes around, it seems, comes around.