So Shannon developed the channel capacity concept into a complete one by involving it w/ SNR - a breakthrough which I think goes far beyond Hartley's Law.
Based on the equation C = Blog2(1+S/N), what I can see are
1. For NRZ, it implies a SNR of 3 even w/o any noise present?
2. Does 1+SNR actually take into account of M?
LDPC is puzzling...