Mentoring junior engineers is kind of like building a particle accelerator out of tin cans.  Clunky and unscientific, and mostly it doesn't work, but if you get them going fast enough sometimes they veer wildly off course and collide with something and there are explosions!  And then you learn something.