Best definition of cryptography

From: Anton Stiglic (stiglic_at_cs.mcgill.ca)
Date: 04/29/04 

Date: Thu, 29 Apr 2004 11:21:08 -0400

I often read a definition of the word "cryptography" and get annoyed with
the fact that it restricts the science to that of encrypting and decrypting
messages. (Another thing that often annoys me is seeing explanations of
signature scheme based on the use of an encryption scheme, but that's
another topic).

We all know that cryptography is much more than that. Probably
the best known goal of cryptography, other than confidentiality, is
integrity. Integrity in it's essence has nothing to do with encryption and
decryption, although you can built mechanisms that allow you to verify
the integrity of a message from a block cipher, in CBC mode for example,
there are other ways as well, like HMAC which has nothing to do with
cryptography. You can also use a MAC to build an encryption scheme,
"Chaffing and winnowing", but it's not a very "efficient" reduction and in
general I would say that encryption schemes and integrity schemes are
not the same beasts.
Other than integrity there are allot of other things that cryptography can
provide, such as

- identification and/or authentication
- random number generators
- zero-knowledge proofs
- oblivious transfer
- bit and string commitments
- multi-party computations
- anonymity providing systems
- etc...

A good definition of cryptography should be general enough to include
all of the above.

Some definition mention the four classic goals of cryptography,
confidentiality,
integrity, authentication, non-repudiation, but that doesn't cover
everything
either, and "non-repudiation" is a controversial subject.

My favourite definition is from Gilles Brassard: Modern Cryptology
- A Tutorial Springer 1988

I don't have the book with me, and it has been a while since I read it,
but if I recall correctly, it defines cryptography as something like this:

"The science of providing secure communication over insecure
 channels".

This is the best definition I have seen, the word "secure" is ambiguous
enough to allow for various definitions. For example, you might
want to establish a communication channel that is in clear, and only
assures integrity, that could be the definition of a secure channel in
some scenario. In another scenario, secure might be related to the
property of anonymity, or reliability.
In a far fetched sense the definition can also include stuff like
zero-knowledge proofs and multi-party computation.

My question is what is your favourite definition of cryptography?

--Anton

Relevant Pages

  • Re: Best definition of cryptography
    ... > We all know that cryptography is much more than that. ... Integrity in it's essence has nothing to do with encryption ... You can also use a MAC to build an encryption scheme, ...
    (sci.crypt)
  • Cryptography FAQ (08/10: Technical Miscellany)
    ... How do I use compression with encryption? ... What does ``random'' mean in cryptography? ... WordPerfect encryption has been shown to be very easy to break. ... where the ciphertext is something like the ...
    (sci.crypt)
  • Cryptography FAQ (08/10: Technical Miscellany)
    ... How do I use compression with encryption? ... What does ``random'' mean in cryptography? ... WordPerfect encryption has been shown to be very easy to break. ... where the ciphertext is something like the ...
    (sci.crypt)
  • Cryptography FAQ (08/10: Technical Miscellany)
    ... How do I use compression with encryption? ... What does ``random'' mean in cryptography? ... WordPerfect encryption has been shown to be very easy to break. ... where the ciphertext is something like the ...
    (sci.crypt)
  • CryptoSurvey -- Results ..
    ... Many same or similar behavioral barriers for the ... effective utilization of many security solutions still exist limiting ... applications of encryption technologies currently in commercial ... Many people do not care about cryptography and/or security products ...
    (sci.crypt)