Basically, if you mess the the equation, you have to be very sure you aren't simulating something silly. Which is easy to do, unfortunately I've done it often.
I still need to read the original paper in detail to confirm, but if the post is correct, the N-body simulation might have some issues.