I thought of a unique way to encrypt information.

Every possible key is of course given a binary value.

When the first (and all odd numbered in proper sequence from henceforth) letter is chosen, it is given the last binary value and all other binary values to the right of it are reduced by one. When the 2nd (and all even numbered afterwards) is taken, it is given the first numerical binary value and all values to between the chosen (and empty spot) are moved up one in value.

An example using a 8 character language

A = 000 = 1
B = 001 = 2
C = 010 = 3
D = 100 = 4
E = 011 = 5
F = 101 = 6
G = 110 = 7
H = 111 = 8
If the message is as follows:

ACEGHBDD

We start with the following binary values:

000 010 011 110 111 001 100 100

Then via the act of shifting we get instead

000  (New order then B = 1 and H = 7 and change all between, A now equals 111)
001  (B=2, D=3, E=4 , F=5, C=1 , G=6, H=7 and A=8)
100  (E=8, A=7, H=6 G=5, F=4)
011  (G=1, C=2, B=3, D=4)
101 (H=8, E=7, A=6)
010 (B=1 G=2 C=3)
100 (D=8, H=7, E=6, A=5 G=4)
111 (D=1 now, and change rest up one value each).

There are different ways one can change this, including, but clearly not limited to, pushing all numbers only to the left or right, using an existing encryption tied with this, choosing a variety of 'push to locations' instead of the first and last locations, making multiple entries for the same value (Example, using a 9 bit system, ten repeats of each vowel and other additional keys being multiplied in number, and you change sequentially the one chosen, so that 'drift' does not happen as often from already used characters, and other ways to manipulate the whole.

I considered this carefully when I thought it up. I do not know if this has been created before, so I submitted this for peer review to see if it is an original and if so if some Math Doctorates can validate the efficiency of the core design if 'no one actually knew the code'. Yes it is now public, so it wont be very effective, but I am trying to get validated on my capabilities, not the efficiency of a released code.