The decay is not correlated with the distance between two points. It represents the envelope decay. To be more precise it is the time needed by the peak meter to reach -20 dB when feeding it with a 0dB / -oo dB step signal.

The peak meter sends automation parameters at a frequency of 20 Hz to avoid the host to receive too much information. Afterwards the hosts usually simplifies the data, and this really depends on the host (for example Cubase SX 1 does an awful simplification on automation, and Sonar does it pretty well).

Do you have an issue with the distance between two points?