The looping version could keep a static array around to eliminate that difference. I think Fibonacci is a bad example because the student will wonder why they are going to the trouble of dynamic programming when it doesn't save any work. A better example would be the recursive computation of N choose K - because it avoids overflow from computing the factorial-based formula directly - or matrix chain multiplication (http://en.wikipedia.org/wiki/Dynamic_programming#Matrix_chai...).