4

In Bruce Schneier's book Applied Cryptography (1994?) he writes

The NSA uses its power to restrict the public availability of cryptography, so as to prevent national enemies from employing encryption methods too strong for the NSA to break.

My first question: Is this still true today?

My second question: If this is still true today, what are the ways that the NSA restricts the public's access to secure cryptography?

(I can only think of the export restrictions as one of the ways in question 2.)

  • 3
    It might be more accurate to say that the NSA restricts the public's access to secure cryptography (as seen in the recent RSA scandal) – KnightOfNi Apr 18 '14 at 3:41
  • @KnightOfNi: Yes, I am guessing that this is what Schneier meant. – Thomas Apr 18 '14 at 13:02
7

This quote is quite misleading in several ways. Let's recall some facts:

  • Bruce Schneier writes from an American (as in "USA") point of view. This is rather clear: when someone says "national" without specifying what nation he is talking of, then chances are that he means the USA.

  • USA, like many other countries, have export regulations for cryptographic material; they also have other laws regulating usage, import, publication... in a number of more or less complex cases. See this site for a comprehensive survey of cryptographic laws around the World. Take note that these regulations evolve at a fast pace, just like the technology that they are trying to regulate; 20 years are a long time.

  • US export regulations are controlled by the Export Administration Regulations, an emanation from the Department of Commerce, but in 1994 the relevant administration was ITAR, linked to the Department of State, with inputs from the Department of Defense during the review and approval process.

  • In 1994, there was an ongoing criminal investigation with regards to a supposed unauthorized export of "munitions", namely the source code for PGP. In order to work around the export regulations, the source code of PGP was printed as a book (because export of books from the USA is not restricted), and meant for automatic scanning once beyond the USA boundaries (so this was only technically a book; it would be fairer to call it a "floppy disk made out of paper"). It is actually unknown whether the export was really done with the book and a scanner, or simply with a floppy disk or network, using the "book" as a plausible deniability token (since the book exists, it could have been used, and therefore the existence of the source code in non-US countries does not automatically imply that it was exported through any other, regulated means). Anyway, the case was dropped a few years later.

  • The NSA is one of the multiple intelligence agencies of the USA, specially dedicated to communications, in both offensive and defensive roles. This means that the NSA is likely to be involved on cryptography questions, when it comes to breaking other people's systems but also, for US own usages (government and business), promoting encryption systems that other people won't be able to break. In the latter role, standards are established by NIST (which depends from the Department of Commerce) with inputs from NSA (which comes from the Department of Defense) without any formal subservience.

The rest is mostly speculations. The quote from Applied Cryptography translates Schneier's point of view, or, more accurately, what was his point of view back in 1994. These speculations include the singling out of "the NSA" as an all-powerful puppet master, able to curb the other parts of the US Federal administration to its will; that the NSA wants to spy on American people (that one is true, although their mandate is for targeted individuals) and will thus want to kill strong cryptography altogether (that's the speculative part: it completely ignores the other role of NSA as promoter of strong cryptography); that everything which happens in the field of cryptography is (or was) a direct result of a struggle between the arch villain Darth NSA, and a bunch of disorderly youngsters fighting for their freedom (of speech). This is, at best, an atrociously simplistic cliché. It also makes the highly implausible assumption that a huge administrative structure like the NSA is both competent and structurally efficient.


Like most defense-related organizations which have lived throughout the Cold War, the NSA and the myriad of other similar agencies has long had a structural tendency to think of "the Enemy" as definitely Red and Siberian. However, the field has changed. In particular, War is now much more a question of economy and commerce than soldiers and nuclear warheads. Classic spying involves a fake plumber with big ears in an embassy; modern spying will more use a fake intern at some big business. This is a tragic turn of events that established intelligence agencies find hard to live: they are no longer spearheading the spying effort. Their playground has been stolen from them, and has shifted to private corporations who now wage that war between themselves.

This forces intelligence agencies to reevaluate their goals and methods, which is easier said than done. To some extent they found some relief in the definition of new "normal" enemies (albeit with less communism, but longer beards and a sunnier tan). But they still have to tilt toward their "defensive" role. In the field of cryptography, this can be seen by comparing DES with AES. The old encryption standard was designed by IBM. NSA was involved, and, from what was said afterwards (in particular by Don Coppersmith), we can conclude that the NSA inputs influenced the design in two ways:

  • The (effective) key length was shortened, from an initial 64 bits to 56 bits (which mechanically implies a lower resistance to exhaustive search).
  • The DES "S-boxes" were strengthened against differential cryptanalysis. More accurately, both IBM and NSA has independently discovered that kind of cryptanalytic method, and they worked together to make strong S-boxes; NSA furthermore "convinced" IBM not to publish the method (which was ultimately rediscovered by Biham and Shamir in the 1980s, and then published).

These design influences are the mark of a self-concious leader: the NSA of the 1970s knew (or at least assumed) that it was more powerful than the rest of the World, both in computing power (they thought they could go through 256 possible keys faster than anybody else) and in mathematics (hence the ban on publication of differential cryptanalysis). They still mixed in both offensive weakening (reduction of key length) and defensive strengthening (improved S-boxes), thereby fulfilling their dual mandate.

Such advances could not last for long, especially since private business finally decided to embrace computers. Even talking only raw money: current NSA budget is estimated around 10 billions of dollars, a hefty sum; but Apple's cash reserve are up to 160 billions (as of early 2014). Back in the late 1990s, US intelligence agencies (including NSA) admitted that they could no longer expect to be in the lead, thus dooming the strategy of enforcing "slightly weak" algorithms. Instead, they went for the design a new encryption algorithm which was so strong that nobody would be able to break it; not them, but nobody else either. This was the AES. With such goals they could even afford an open, international competition, and select an algorithm designed by foreigners ("reasonable" foreigners, mind you; preferably westerners from a non-controversial country, in this case Belgium).


From all this, we can conclude that the NSA is not really working to prevent "the public" from accessing "secure cryptography". Instead, they are working hard to ensure that:

  • the "good" people (US businesses) have access to strong cryptography that the "bad" people (their non-US competitors) won't be able to break;
  • they can keep track of all cryptography products that are in use, so that they can at least know whether any given algorithm can be recommended for "good" people use;
  • they still get enough legal weapons to trigger search warrants, assets freezing or other restrictions on people that they really don't like.

NSA also realized that in a given security system, cryptography is often the strongest part, not the weakest, and concentrating on buffer overflow exploits and concealed backdoors was much more likely to yield tangible results than trying to enforce weak crypto everywhere. In particular, the Internet quite kills any hope of effectiveness for an import or export embargo.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.