Hey Tom!
I really liked your article, it was very interesting! I am going to try this for sure.
However, this part is completelly false:
CAS latency is measured in cycles, not in time. Thus, while higher speed memory tend to have a higher number of cycles of delay, every cycle is faster!
For exemple, 2400mhz at CL 12 has a cycle time of 0.833 ns, giving us a total of 10 ns of latency.
The 1600mhz at CL9 has a cycle time of 1.25 ns, giving us a total latency of 11.25 ns!