On 2012-02-12, adacrypt <
austin...@hotmail.com> wrote:
> Two points in space can be defined by the position vector that
> connects the two points.
>
> In everyday standard usage one of the points is taken to be fixed at
> (0,0,0) and every other point in the whole of space can be expressed
> as being relative to this point.
>
> But users are not bound to always uses (0,0,0) as the standard
> reference and they can agree between themselves to use another private
> reference point at say (x,y,z ) to define points in space whenever
> they need to. The position vector is then totally different to what
> it would be relative to (0,0,0). Only the users know just how
> different because they alone are privy to (x,y,z) which could be any
> point in the infinity of space.
In one dimension this is the Caesar cypher. It is easily broken.
Similarly in three, assuming that xyz are the same for the whole
message. It is breakable because the message contains masses of
redundancy, and possibly known plaintext. Ie, If I know the coordinates
of three points in your xyz translated system and in the 000 system
(known plaintext) I also know what xyz are and thus can decrypt the
entire message.
If you change xyz for each and every point, and those changes are
completely random, then this is a one time pad, with all the problems of
such (the necessity of conveying to your partner in a totally secure
manner, the whole range of xyz values. If you have such a secure way of
conveying them why not use that means of conveying the message?) If the
xyz are created by some algorithm, this is just a stream cypher. with
the usual means of attack on the generating function for the stream of
xyz.
>
> With knowledge of (x,y,z) any person who knows this can navigate to
> (0,0,0) and find the correct position vector relative to (0,0,0) when
> that is needed but (x,y,z) could be any one of an infinite number of
> confusingly different values in the whole of space when users decide
> to keep it secret.
>
> There is no mathematical means whatever by which (x,y,z) can be
> deduced only the users know and they alone can provide the information
> (but of course they are not telling).
Sure there is. If you know the values of the point for an xyz
displacement and you know what its value should be (with respec to 000)known plaintext)
youknow what xyz is.
>
> When the users are Alice and Bob in a secure communications scheme
> this ploy is called a change-of-origin. I liken it to a transfer of
> data from a human memory to a computer memory that they alone can
> implement and supply the correct values of (x,y,z) that enable
> decryption to proceed with a correct result.
>
> They calculate ciphertext according to (0,0,0) but publish it as being
> relative to (x,y.z) and go back to (0,0,0) again at decryption time.
>
> I am calling this a definitive one-way mathematical function in
> cryptography ? one-way simply because that is what it is i.e. non-
> invertible by any mathematical means albeit a proper mathematical
> function per se at the same time.
Nope. A one way function is one where, even if you know what xyz is, you
cannot figure out what the values are for 000.
Ie, in crypto, a one way function is one where, even if you can
perfectly encrypt a message ( in your case you know what xyz is and can
make the change of origin) you cannot decrypt it.
Clearly not true of your scheme.
>
> This powerful ploy guarantees my cipher against the attack described
> as attack 1) that is foremost in basic design priority, called known
> ciphertext attack.
>
> I have no evidence that there is such a thing in academia as a ?one-
> way? function in mathematical parlance. (Never mind the Cryptography
> Handbooks ? they are not quotable outside of cryptography).