[Cryptography] A software for combining text files to obtain high quality pseudo-random sequences in practice

[Jerry Leichter]

> I do not think it is viable, because it is not random; it depends
> entirely on the input text. One article puts total size of the
> web at 1.2 petabytes
> http://www.sciencefocus.com/qa/how-many-terabytes-data-are-internet
I didn't see where the original proposal suggested using the entire Internet's text corpus as input.

However:  This reminds me of an old proposal - I don't recall where I saw it - for something of this sort as a much more secure version of a book code.  Take a high-resolution image as a source of bits; I think the original suggested some NASA image of the moon.  A key is several starting points and step sizes in a fixed linearization of the image.  Each starting point and step size gives you a stream of bits.  XOR together and then use the result as input to a stream cipher.

Yes, insecure in theory - you're reusing a "one-time pad"; worse, the contents of the "one-time pad" are public!  But the only plausible attack appears to be brute force, and there are a *lot* of keys....
                                                        -- Jerry