The US scientific community faces a dilemma that has not arisen since World War II. It is being asked voluntarily to censor a field of free and open research -- the mathematics of cryptology.
To acquiesce would impose a prior restraint on publication reminiscent of the self-censorship of atomic scientist 40 years ago. To refuse could invite even more onerous legislated restrictions and possibly do some damage to US security interests.
This is not an issue concerning secret or restricted military or commercial research. It is a question of what to do about open basic research in certain mathematical fields, such as number theory, which makes possible what appear to be unbreakable secret codes. Such research is quite fundamental and can be done by any competent mathematician specializing in the relevant fields.
The National Security Agency (NSA) of the US Department of defense, which is responsible for communications intelligence, has been concerned that other countries might develop more secure communications which NSA then would have more trouble decoding. Thus NSA has been looking for a way to edit research publication without seriously damaging traditional scientific freedom.
This hope has been embodied in recommendations of the Public Cryptology Study Group of the American Council on Education. These outline a system whereby researchers and journal publishers would voluntary get NSA approval for publishing research results in what is ostensibly nonsecret work, much of which is funded by the National Science Foundation (NSF).
However, the NSF Mathematics and Computer Advisory Committee opposes this as a "direct and serious threat to NSF's charter of furthering basic research." Rather than submitting research papers for approval (prior restraint), the committee recommends merely informing NSA of the content of such papers. It would then be up to the agency to take legal action to prevent publication if it felt this necessary.
To the committee, the distinction between "prior restraint" and "informing" NSA is crucial. It notes that there should be "clean separation of the procedures for funding and otherwise promoting basic research from the procedures for handling national security and other nonscientific considerations." Compromising this separation for mathematics, it fears, could encourage similar intrusions of secrecy considerations in other basic research fields.
The atomic scientists stopped publishing their basic research when, awed by its implications, they say war approaching and when the government itself had not yet awakened to the implications. There is no such urgency with cryptology today. The NSF and the academic community should be wary of poisoning the basic research a tmosphere with so-called "voluntary" restraint.