How To Choose A Good Scientific Problem