Unbreakable Cryptography Theory...

[daxmax]

So in the end, your "unbreakable Cryptography" isn't unbreakable. It's simple unfeasible to break due to time\ machine power constraints.

[Cheeseball]

Originally posted here by Haikka
I mean a bruteforcing like this:



Encrypted string: sasfgfiealskdmlasieeqskmdapwowmqwfdmafoewpq
The program starts encrypting loads of strings:
a
b
c
...
hello, this is a tesr -> asddamopemlaspfomsföasmfpaeomsgalfmadfsoöae
hello, this is a tess -> alsdaispodmaoifmsdlkmfapowepqoweialksömdpofa
hello, this is a test -> sasfgfiealskdmlasieeqskmdapwowmqwfdmafoewpq [same as the known encrypted string]

-> bruteforcing complete <-

Oh well it would be different if the encryption wouldn't use a fixed key...

This is mostly inaccurate. Most modern algorithms 'scramble' they key and the plaintext between or during each round of encryption. The only real way the example you have provided would work is if the system used a mono-alphabetic system. With any type of poly-alpha system you will not know if you get close to the correct answer, simply because the set of answers to the system can be (with some 0x00 padding) "hello, this is a test", "hi, you evil evil person", "what is this about?", and any other message of the correct length.

As far as this cryptosystem goes, for some reason people always walk up to neural networks for crypto, most of the time trying to make a ANN that will break crypto (which isn't as easy as it looks because there is only one right answer and the key changes everytime, thus resetting all the wieghts to 0, so it's almost pointless to try unless you find a 'shortcut'). This is definately a 'creative' thought...I'd like to see an implementation of it as well as a white paper on the algorithm design so cryptoanalysis would be possible.