the problem, as I understand it (and I think have a pretty fair understanding of this) is that, in a -non- emissions controlled truck, that the coil generates a spark when the circuit is broken, which happens as the key is moved from the 'on' position to the 'crank-start' position, before the starter ever engages the fly wheel, which in a motor with good compression the motor will be slightly before TDC in it's rotation. this will cause a spark that causes the engine to ignite whatever gases were left in the cylinder, and the engine will be moving in the opposite direction as the starter Bendix engages the flywheel.
This does not happen in trucks built since 2004, as the emissions computer is what powers the ignition circuit, and it does not provide any power to the ignition until the engine has provided a signal from the crank sensor to assure the motor is turning (2 revolutions before any spark).
the time delay (PN that is suggested above), when properly wired in, allows the starter to engage/crank for 0.5 seconds before any spark is allowed to the ignition.
I have seen far too many time when the wiring was misunderstood and the time delay was wired in such a manner as to make the starter wait 0.5 seconds after the ignition circuit came on (which almost works right, since any 'rock back' caused by the single stray spark may have stopped before the starter engaged.