I
Since the public viewing of the Snowden affair is 1 year old, there are some articles in the press, on Snowden, but also on Turing and encryption/decryption, and I stumbled on this one (in German): http://blog.zeit.de/mathe/allgemein/d-day-normandie-enigma-zweiter-weltkrieg/
I never understood the Enigma, especially since every "explanation" about it you can find in the web, either is written by experts for experts, or by non-experts who don't understand the Enigma themselves, and the link above falls into the second category, but some comments there rise some interesting points.
It seems Turing began his decryption work on the Enigma after one functional Enigma machine fell into British hands.
It seems the "code for the day" was communicated using the previous code.
It seems some primary code needed for the real encryption, then to be made by the machine, with the help of the "code for the day", was a stable arrangement of the abc chars, and the British tried myriads of possible sequences for this, e.g. qwerty, and so on, but the Germans, "incredibly", just used "abcde", and it seems the mathematician Marian Rejewski did find out this, not Turing, and the commenter in the above link who brings this info into the discussion, muses that Rejewski had worked in Göttingen, Germany, beforehand, so had some first-hand info on German psychology / way of thinking, which enabled him to take into consideration the Germans might do it in the utmost basic, primitive way, a possibility excluded by Brits just admiring the machine but without intimate knowledge of German "national character" - I very much like this observation.
(EDIT: And of course, this over-emphasizing/relying upon the over-obvious resp. the "really-too-easy" reminds us of that E.A. Poe short story...)
It seems the breakthrough was then made by Turing's reflection that by the way the machine obviously worked, on a physical level - direct current was sent thru the rolls in one direction, then in the other direction - no character (a...z, etc.) could be replaced by itself, this drastically reducing the machine's encryption possibilities / possible permutations; in fact, the cited article is primarily about this phenomenon of "Selbst-Bewichtelung", no English translation found, just this transcription, "Players who receive their own gift in 2/3 of all secret Santa games."
Some commenter over there claims the biggest U.S. employer for mathematicians is the NSA - very funny and very convincing, even while no proof is given.
II
I, just some days ago, had mused about specific file formats countering effective encryption. Let's say you use MS Word files, or some other file formats where quite lengthy passages are, more or less identical, but highly standardized at the very least, for every one of your encrypted files, AND the decryptor knows (or can safely assume, from the presence of these applications on your system, or simply by the ubiquity of some applications, like MS Word, and its rather few "replacements", ditto for spreadsheets, etc.) which (few) applications you will have used to produce the encrypted data:
Then, this might drastically reduce the theoretical power of your specific encryption, since the decryptor (assuming even he doesn't have a way (which might exist, without us being aware of such possibilities) to determine where one of your file ends, and the next one begins, which would further cut the possible permutations into a mere fraction of their theoretical potential-by-strong-password) would try to decrypt those "standard passages" first, and even allowing for your individual data within these "standard passages", intimately knowing the "format" of the latter, incl. possible lengths of different such individual data in-between, and once these "file headers" are decrypted, your key will be known.
This would mean that usage of any application not producing just naked ansi files, but putting "processing data", "meta data" into the file, too, should be prohibited if you really want your data safe (= necessary but not sufficient condition)...
Since the public viewing of the Snowden affair is 1 year old, there are some articles in the press, on Snowden, but also on Turing and encryption/decryption, and I stumbled on this one (in German): http://blog.zeit.de/mathe/allgemein/d-day-normandie-enigma-zweiter-weltkrieg/
I never understood the Enigma, especially since every "explanation" about it you can find in the web, either is written by experts for experts, or by non-experts who don't understand the Enigma themselves, and the link above falls into the second category, but some comments there rise some interesting points.
It seems Turing began his decryption work on the Enigma after one functional Enigma machine fell into British hands.
It seems the "code for the day" was communicated using the previous code.
It seems some primary code needed for the real encryption, then to be made by the machine, with the help of the "code for the day", was a stable arrangement of the abc chars, and the British tried myriads of possible sequences for this, e.g. qwerty, and so on, but the Germans, "incredibly", just used "abcde", and it seems the mathematician Marian Rejewski did find out this, not Turing, and the commenter in the above link who brings this info into the discussion, muses that Rejewski had worked in Göttingen, Germany, beforehand, so had some first-hand info on German psychology / way of thinking, which enabled him to take into consideration the Germans might do it in the utmost basic, primitive way, a possibility excluded by Brits just admiring the machine but without intimate knowledge of German "national character" - I very much like this observation.
(EDIT: And of course, this over-emphasizing/relying upon the over-obvious resp. the "really-too-easy" reminds us of that E.A. Poe short story...)
It seems the breakthrough was then made by Turing's reflection that by the way the machine obviously worked, on a physical level - direct current was sent thru the rolls in one direction, then in the other direction - no character (a...z, etc.) could be replaced by itself, this drastically reducing the machine's encryption possibilities / possible permutations; in fact, the cited article is primarily about this phenomenon of "Selbst-Bewichtelung", no English translation found, just this transcription, "Players who receive their own gift in 2/3 of all secret Santa games."
Some commenter over there claims the biggest U.S. employer for mathematicians is the NSA - very funny and very convincing, even while no proof is given.
II
I, just some days ago, had mused about specific file formats countering effective encryption. Let's say you use MS Word files, or some other file formats where quite lengthy passages are, more or less identical, but highly standardized at the very least, for every one of your encrypted files, AND the decryptor knows (or can safely assume, from the presence of these applications on your system, or simply by the ubiquity of some applications, like MS Word, and its rather few "replacements", ditto for spreadsheets, etc.) which (few) applications you will have used to produce the encrypted data:
Then, this might drastically reduce the theoretical power of your specific encryption, since the decryptor (assuming even he doesn't have a way (which might exist, without us being aware of such possibilities) to determine where one of your file ends, and the next one begins, which would further cut the possible permutations into a mere fraction of their theoretical potential-by-strong-password) would try to decrypt those "standard passages" first, and even allowing for your individual data within these "standard passages", intimately knowing the "format" of the latter, incl. possible lengths of different such individual data in-between, and once these "file headers" are decrypted, your key will be known.
This would mean that usage of any application not producing just naked ansi files, but putting "processing data", "meta data" into the file, too, should be prohibited if you really want your data safe (= necessary but not sufficient condition)...
