Encryption is secret writing. Codes and ciphers. Spies. Encryption was originally used by military and diplomatic organizations; Julius Caesar invented an encryption scheme. In the last century, electronic communication (telegraphy and radio) made it widely useful, and computerization has made it extremely cheap. Widespread public networking has made it useful to everyone, for everything from putting "envelopes" around your email for privacy, moving money around the net safely, to proving that you're really you when you're halfway around the world.
The US government is deathly afraid of its own citizens (and non-US-citizens) having access to good encryption. This fear extends all the way up to the Vice President and the head of the FBI, who personally get involved in creating encryption policy. Everyone in government refuses to tell us why, saying it's classified and the national security is at stake. Rubbish! The security of the nation is already gone when its government violates the basic rights of its own citizenry, as these agencies do every day. They are "burning the Constitution in order to save it". (My own belief is that what's really at stake is a wiretap-based power base that J. Edgar Hoover and the classified spy agencies have built up for their own benefit.)
The most Byzantine set of laws, regulations, policies, departments, and practices you've ever heard of are employed by the National Security Agency and three or four other Executive Branch departments in an attempt to keep good crypto from bad guys. Unfortunately, they have also succeeded in keeping good crypto from good guys who have Constitutional rights. I instigated a
lawsuit to correct this, with Dan Bernstein as plaintiff and the Electronic Frontier Foundation backing him up. I was a technical advisor to the lawyers in the case. On December 6, 1996, Judge Patel decided that the export regulations are unconstitutional. The government appealed, and on May 6, 1999, the 9th Circuit Court of Appeals
agreed with her. The government appealed to an 11-judge panel in the 9th Circuit (an "en banc review"), which was granted, and then the government "voluntarily" changed the encryption export regulations so that most free software and academic research, and a lot of proprietary encryption software as well, can be easily published from the US. The "en banc review" of the old regulations became moot, and the case has been handed back down to Judge Patel, who ultimately ended it. The new regulations are even more complex than the old ones, and carry the same old harsh penalties for inadvertent violation. They need to not just be "reformed" but scrapped.
The government claims to retain the right to change those rules whenever it wants, and restrict encryption software again if it chooses. Congressman Judd Gregg announced support for doing so in the week of hysterical reaction after the World Trade Center was destroyed by hijacked airliners, but was shouted down by the people who'd spent a decade fighting this battle before he could gather any political support.
I have had an interest in encryption since childhood, and have spent a lot of time working on
crypto export control issues.
I led the team that built the world's first publicly announced
DES Cracker, a machine that finds the secret key used to encrypt messages in the government's favorite encryption scheme, the Data Encryption Standard (DES). The National Security Agency intervened when the scheme was being standardized in the early 1970s, shortening the secret keys so that they could build their own DES Crackers. But they spent the next 25 years lying to us about how secure the scheme is, to encourage everyone to use it -- and we did. This left NSA able to secretly eavesdrop on anyone who used DES, which includes the entire financial community, and most computer and network security systems. Technology has advanced to where anyone with $200,000 can break the code, leaving all of our DES-protected infrastructures at risk.
Thanks NSA! By 2002 much of the older DES-based software has been replaced, though there are numerous places that still use it, and its use is an option in many new protocol implementations even though it is known to be insecure. NIST has standardized a new algorithm with much longer keys, which has not been studied nearly as long as DES, but which has resisted all attacks so far. Smart people have stopped designing DES into new systems. Triple-DES or AES seem to be the preferred replacements.