Cryptography Reference
In-Depth Information
In contrast, the program cut its teeth over a short Caesar riddle similar to the
one above!
Such games are helpful in learning to understand problems and possibilities
of automatic cryptanalysis that have been a classic domain of national intel-
ligence organizations. This is the learning effect. And the aspired Aha effect
should have happened when the author's kids dead cert secret writings gradu-
ally revealed themselves on the screen. All right, I openly admit, that was my
motivation for the entire effort in the first place. Unfortunately, the kids had
long left the house before my modest program finally finished.
2.2.2 First Improvement: Homophone Substitutions
We have seen that simple substitution ciphers can be broken by frequency or
pattern analysis, even when the messages are relatively short. A trick helps
you make this task harder. The trick is the so-called
homophone substitution
,
which let's you assign one plaintext character to several ciphertext charac-
ters. The ciphertext alphabet includes more than letters; for example, it can
also include numbers and special characters. You will want to assign several
symbols particularly to the most frequent characters, such as 'e', 'n', 'i', 'r',
and 't'. (Blanks are always omitted in classic cryptology.) In the ideal case,
all characters would occur more or less equally often in your ciphertext. The
homophone substitution is better than the simple substitution, but it still has
serious drawbacks:
•
First, there are no defined rules for selecting one out of several possible
ciphertext characters. The quality of such rules determines the quality of
the algorithm. Look at this simple example:
Assume we want to substitute the character 'e' by either 'b', '4', or '!'. A
stupid or unqualified code writer would take his choices cyclically: in the
first cycle, he replaces 'e' by 'b', in the next by '4', then by '!', and then
by 'b' again. If the adversary is aware of this approach, which is normally
the case, then he will search a lengthy text (using a computer) for groups
of characters that always occur in the same cyclical arrangement. And
the encryption algorithm will quickly have been identified to be simple
substitution.
A random, 'unreliable' selection by hand would definitely be more secure.
This is the way the method was used in practice. Living in the computer
age, we have different demands.
