According to every textbook I know of, magnetic forces do no work (e.g. David Griffiths Pg. 207). Yet this problem causes this to be hard to believe:
If I take two magnets, I can set them down on a table (with a little friction). I then slowly push them toward each other, then at some point...