[Cryptography] A question on Shannon's entropy

[mok-kong shen]

Given 2 alphabetical sequences A and B of the same length. If A
has higher entropy than B, then (if I don't err) this is commonly
interpreted in crypto to imply that A is more random than B. Now
suppose A has entropy u. Sort A to result in C such that all a's
are in front of all b's etc. Since the frequencies of the characters
are not affected by the sorting, C has also entropy u according to
Shannon's formula. However C is evidently less random than A. How
could one resolve this prolem? One could eventually employ pairs
of characters as unit and compute a similar measure with the
formula. But evidently the probllem would at best be weakened but
not basically resolved that way.

M. K. Shen