Cryptography Reference
In-Depth Information
rather than to spying. People aware of the strength of Soviet mathemati-
cians won't be taken by surprise.
However, we don't know precise details, for cryptology in the Soviet
Union has been locked against the outside.
Actually we don't know many precise details in general — what became of the
captured Enigmas, how many machines were built or sold, who produced them,
and up to when they were in use.
Hinsley also speculates about what turn the war might have taken had the
Enigma texts
not
been deciphered: the invasion of the allied forces would have
happened nevertheless, but perhaps in 1946 or 1947. Unless the Soviet Union
had ended the war by that time the USA would probably have used nukes in
Europe, too (this is one of Hinsley's speculations, but it's worth considering).
Well, history cannot be turned back, nor should it be overly simplified. When
Willi Korn invented the reversing drum, he had no way of knowing what
consequences it was to have for world history.
There is still no end to mystery-mongering. The NSA declassified thousands
of documents about World War II on April 4, 1996 (50 years later!). You can
find notes and quotations on the Internet; references are also included on the
Web site to this topic. The British government has not to date declassified
many important details about deciphering the Enigma. It became known only
in 2000 that ten 'Colossus II' machines, which were superior to the first official
ENIAC computer, were in use. More about this on our Web site (
colossus.txt
and
declassif UK.txt
).
Cryptanalysis Today
Despite it all, I believe that this machine can be broken without using a plaintext
attack considering the current state of computer technology. Since it works with
only 26 letters, we may reasonably assume that it uses normal plaintext. Its letter
distribution is known, and I would launch an attack at this very point. Without
considering the plugboard, a 'ridiculous' set of 1 054 560 possible keys remain
to be tested at various speeds, depending on the program quality and hardware.
Each decryption attempt results in a letter distribution that can be compared
with the expected distribution. We select keys with reasonable results and test
for other criteria to see whether or not the result could be a language. This
trial-and-error process is likely to be fast. But what about the plugboard? The
second transposition — before revealing the plaintext — doesn't change anything
in the distribution, the first does. Without being able or wanting to substantiate
my statement, I think that that's something one could handle.
