On the Enigma machine: Charles Stevens made the attached cdf. Its outstanding.Taken from this discussion

Attachments:

German Enigma Paper.cdf

Hi Ali,

very nice, thanks for sharing! But I must say that the germans never ever would have encrypted some english text ...

Anyway, your nice code and the fact that you are using just 3 rotors (i.e. there are only 17576 keys) gives me the chance of testing an old idea: An incredibly simple and direct brute force attack - just try all keys and look for english text as outcome, you do not even have to care about the correct key! I guess Alan Turing could not even have dreamed of such an insane approach!

encText = "uubf jw dif oo, jctjgmbn nbtqrang, pvs vsh o orgiya lq lyw svn ssui zcxuxs?";
allKeys = Tuples[CharacterRange["A", "Z"], 3];
allEncodings = ParallelMap[EnigmaEncryption[encText, Sequence @@ #] &, allKeys];
bruteForce = ParallelMap[{#, LanguageIdentify[#]} &, allEncodings];
englishText = Select[bruteForce, Last[#] == Entity["Language", "English"] &];

This should work - but it does not work! I was prepared for some false positive findings (therefore I did not use a While construct), but not for so many (>9%)! Can anybody comment why LanguageIdentify is working so badly?!

Best regards -- Henrik

The problem is that LanguageIdentify, is a simple classifier that always gives an answer. It is the same as:

Classify["Language"]

What you can do: use the underlying classifier and ask for probabilities:

Classify["Language"]["abcdefghijklmnopqrstuvwxyz", "TopProbabilities" -> 1]

and then only accept if the probability is above a high percentage (e.g. 95%). Else, unidentified.

Hi Sander,

an 7 years late answer - as this discussion comes up again: According to your advice (and for the sake of completeness), here is a working version of a most simple brute force attack:

encText = "uubf jw dif oo, jctjgmbn nbtqrang, pvs vsh o orgiya lq lyw svn ssui zcxuxs?";
allKeys = Tuples[CharacterRange["A", "Z"], 3];
allEncodings = ParallelMap[EnigmaEncryption[encText, Sequence @@ #] &, allKeys];
bruteForce = ParallelMap[{#, Classify["Language"][#, "TopProbabilities" -> 1]} &, allEncodings];
englishText = Select[bruteForce, #[[2, 1, 1]] == Entity["Language", "English"] && #[[2, 1, 2]] > .9999 &]

It is interesting to see, what else could be an English text with high probability.

Again many thanks and best regards -- Henrik