or from creativecrash here.
All thanks to Brian Escribano for his softSelection() code that gathers the id and weights of the current soft selection (basically, all the hard work).
It was simple enough to apply that information to a new cluster.
Extract and place script file into a maya script directory (or python path directory) and run with this
import ld_createSoftCluster as sc sc.ld_createSoftCluster()
and thanks to Valen Wagner for the icon.
07/03/12 - 0.4.0
Alters cluster position to match selection rather than influenced average.