24-07-2025, 02:27 PM
This is the way character set reduction works, with the new character set of 11 symbols per character entropy cannot be larger than 3.45, but also this encoding is heavily skewed towards using N very often.
It easy to even bring the entropy well below 1 (I think), just encode any code N as a sequence of N zeros and mark the end of the sequence with 1.
It easy to even bring the entropy well below 1 (I think), just encode any code N as a sequence of N zeros and mark the end of the sequence with 1.

