That's a reasonable way to look at it, I think. Except imagine an ARP where there were thousands of very highly paid network engineers constantly monitoring the tables.