Image talk:Cipher-taxonomy.png

Matt, I've just noticed your diagram. Sorry I took so long. I have to say that I think it misleading as it mixes imprecise indeterminate common word usage with determinate diagrams. In short the labels are wrong. In addition, there is an attempt to mix mechanism with concept, and with several dimensions of concept, all of which make diagrams more difficult.

We should have something like the following (in primitive ASCII art):

encryption -- code / cypher

   one book -- |            | -- symmetric key 
   two book -- |                  | - substituion; 
                                  monoalphabetic / polyalphabetic
                                     | -- hand (lots of kinds)
                                     | -- machine (rotor etc)
                                  | --- transposition
                                     | -- hand (rail fence)
                                     | -- 'machine' (scytale)
                                  | -- combination 
                                     | -- most block cyphers
                                     | -- some hand cyphers (JN-25)
                            | -- asymmetric key 
                                  | -- public key / private key (RSA)
                                  | -- two private key (lost exmpl)

Some folks follow diagrams for organization clarity (a fellow I knew in college learned taxonomy solely from diagrams he made) others rely on textual material. Both should be accurate, but they should surely match lest the cognitive dissonance in those who look at both be too extreme. ww 15:32, 19 Jul 2004 (UTC)

I understand your concerns, but I disagree; the taxonomy splits ciphers into the common historic classification, not an abstract description; this is most useful for the reader, because it follows how the terms are used in the field. The first "three way split" is into the three rough stages of development of ciphers were used; 1) classical ciphers, 2) rotor machine, / electromechanical encipherment and 3) modern ciphers. Authors agree on these stages in cryptography (although some merge rotor machines into classical). Within these stages, there are subdivisions commonly given, as represented in the diagram; classical into 1.1) substitution and 1.2) transposition; and modern ciphers into 3.1) symmetric key and 3.2) asymmetric key. The substitution / transposition division makes no sense in the context of modern cryptography, as the concept doesn't really come into play — all ciphers are substitution ciphers (cipher components are a different matter). Equally, the symmetric / asymmetric division is not relevant in the context of classical ciphers, even though the concept might apply in an abstract sense. — Matt 20:34, 19 Jul 2004 (UTC)

Matt, Thanks for your prompt reply. I think then you're attempting something I didn't understand, and that fact presents its own problems. If the idea is, as I take it from the above, to use historical importance in practice as the groupings, then it needs to be more clearly so. But I think that's not what in fact you did quite as you have mixed operating characteristics (eg, private key v public key) in as part of the classification scheme. Given the potential for confusion (even of a not so naive reader) there is need of some modification. And codes are left out (which echoes our discussion of some time ago, I suppose). ww 16:06, 20 Jul 2004 (UTC)