Okay, I just glanced through this entire thread, and I had to mention something:
I thought Russ's original statement was that if you executed a trojan that was stored on computer A from computer B, computer A would be infected. That's not possible - it's simply a shared device - anything you run from that drive on your local computer will only affect your local computer. For example, if you double-clicked an MP3 on your friend's shared drive, would the MP3 start playing on his computer? Therefore, I'm with josphstalinator on that one.
On the other hand, like Russ is indicating, it _is_ possible that the person who placed the trojan on computer A could, using his write-access, rig computer A to execute the trojan upon next restart. However, I don't think that's what you guys were arguing, was it?
Correct me if I'm wrong, but that's my interpretation of this thread!