the box of ack rather than individual shells isnt really the problem for me, its the instant tracking which needs fixing. and it could be a fairly easy fix. my suggestion (which simulates the real director operation):
1. target designated, target location A (xA,yA,zA) noted.
2. wait 10s, target new location B (xB,yB,zB) noted.
3. wait 10s, ack bursts at location C, predicted from A and B (xC,yC,zC) = (2xB-xA,2yB-yA,2zB-zA) and ack continues to burst along the predicted path.
etc.