There is a limit on gamma, the correlation factor, so that the upper-bound of the subset-estimator COV is lower than the MCS COV. But is this limit useful? Since the actual COV is usually much lower than the upper-bound. Besides, theoretically gamma can be very large; which means the COV bounds are very large. However, empirically we found that gamma is usually < 10. Wondering what all these mean. Wondering whether all these are useful, meaningful, and not in vain? Of course they are not in vain, are they?