Subjecting a text to a transformation is a lot like doing arithmetic. Thus, I view as highly intrusive any attempt to make the private use of encryption illegal, even if certain forms of hardware encryption with government access features are excepted.
On the other hand, the use of any form of encryption over mobile radio links, such as Amateur Radio or Citizens Band radio, has long been banned; and this does not seem unreasonable, both because the use of the airwaves is regulated in many ways, and the use of radios to co-ordinate a bank robbery or outside assistance to a prison escape is an obvious possibility. Thus, while the idea of putting a Clipper chip on every telephone in the U.S. (this assuming every home in the U.S. had an ISDN line to hook up a digital telephone to) naturally raised many eyebrows, using that same technology in the limited province of cellular telephones doesn't seem unreasonable. This would even have been a good place to start for a government with an agenda to put that chip in other places; a claim the chip was for mandatory use in one specific area, highly subject to regulation in any case, would be believable. A claim that it was for voluntary use (what good is it to be able to tap the phones only of people who don't mind) was simply not credible, and steered the whole thing into a sufficiently deep quagmire of public objections that I must confess to being unsure that the U.S. government was ever really serious about Clipper.
Since even an old 8-bit computer, such as a VIC-20, can be used to encrypt text securely, using algorithms considerably more complex than those used by the cipher machines of World War II, and since basic knowledge about cryptography is widely available (unlike programs that implement DES, even the export control regime of the U.S., which goes beyond the Wassenar agreements in several respects, does not restrict the export of descriptions of DES or similar algorithms), and since how to program a computer is a basic skill that is also very common, it is difficult to see how a compelling national security need is met by placing restrictions on the export of encryption software, particularly in source code form.
However, encryption hardware, not only offering additional security features, but also being ruggedized for use under adverse battlefield conditions, is obviously military in nature, and requires advanced technology to produce.
Even in the nineteen-twenties, financial institutions used cryptography to protect funds transfers. However, codes and ciphers have traditionally been primarily the province of rulers, diplomats, spies, and especially the military. Today, the basic insecurity of the Internet has stimulated public interest in cryptography for much the same reason as the telegraph did in its early years.
The development of public-key cryptographic techniques is also responsible for the increased current popularity of cryptography, as it has made setting up secure communications much more convenient, and hence practical in situations where it would previously have been awkward.
Is cryptography a weapon? Its relation to military operations seems to be rather that of a support function. Jeeps that don't break down, radios that work under harsh conditions, food that won't spoil, clothing suitable to the weather that will last: an army needs all of these things, just as it needs a way to report observations, and recieve orders, that cannot be overheard by its adversaries.
I cannot, therefore, deny that it is militarily useful, even if it is in the category of a dual-use technology. Export controls impose costs, by on the one hand denying sales to the commercial software industry, and on the other hand restricting the availability of freeware and shareware software to consumers containing encryption, since this kind of software is distributed in ways which make preventing export difficult to do.