Excellent question! Stated rhetorically, "If a GFCI only cares about the current imbalance, and the traditional style breaker opens to due to a big spike in c

Excellent question! Stated rhetorically, "If a GFCI only cares about the current imbalance, and the traditional style breaker opens to due to a big spike in current, why care about what path it takes through the building wiring back to the pole transformer?" If a sub panel does not have the jumper removed, then the EGC and GSC between that sub panel and the main panel will form a current divider. Let's assume they are the same wire gauge, basically the same resistance, and so essentially split the return current. This leads to a couple problems: 1. The EGC may no longer able to carry its full rated current load, due to the baseline "normal" current that the EGC is also bearing. You want to bear as much current as possible to quickly trip a conventional breaker in event of a fault to the EGC. This is real, but relatively minor concern. 2. All EGCs "upstream" of the bonding point have been "contaminated" by the normal neutral current that got on to the EGC at the mistaken sub panel bond. This could "put current" on the surfaces of other equipment, exposed ground wires, conductive raceway, etc. This is the more major concern, as these surfaces would have the full voltage potential available to them should there be yet an additional current path. In the main panel the EGC, GSC, and GEC are all connected at a single point. And from this point there is only one path back to the transformer, via the 0V conductor of the GSC. The transformer side of this 0V potential conductor will match the local Earth voltage, set by the GEC, and wiggle from 0V nominal by the fluctuations caused by current (V=IR) in the GSC conductor between the panel and the pole. Make sense?