Positional Mimic Cipher (PM-Cipher)

[quimqu]

But multiple encodings would probably drive overall entropy up?

I've tested two tables, one for the first lines of the paragraphs and the other for the rest (as it is also known that they are quite different). The model fits a bit better, and the entropy remains low.

If we switch residuals, it should go up a bit. But if we find the correct natural language, the cipher would fit better (not 100% because that would be a direct substitution). So, the residuals would be minimal, and the general entropy would be even lower, so switching residuals would make it a bit higher, but starting from a lower point (due to the very low degree of residuals).

I don't know if I have explained myself properly..