The major advantage of this "bug" is that you can begin your sequence with double of the same note, which you cannot do in the normal documented way unless you enter a full 16-step sequence (i.e. you would have to hold some arbitrary note, like lowest C key, then punch in your double notes and fill out your sequence from there, but you have to enter in the full 16 steps so your first "hold note" gets overwritten).
With the regular documented method there is no way (that I am aware... anyone correct me if I'm wrong?) so enter a sequence "A, A, B, C" since if you hold down that first A note, you can't repeat it... or if you hold down a DIFFERENT note, then punch in double A, you end up with something like C, A, A, B... which might SOUND the same but technically will not be "lined up" / synced correctly.
So with this bug, you can enter your sequence in literally exactly as you want it, starting with double notes if desired, AND without having to punch in a full 16-note sequence.
At least that's the way I understand it... I have not actually TRIED the "bug" yet
but these are the deficiencies I ran into when using the ARP sequencer, and from all descriptions of the "bug" I see, this seems to be one notable advantage.