Read it hoping the topic was going to be sub-gradient optimization (or any kind of technique that follows an "interesting" direction when the gradient is either

Read it hoping the topic was going to be sub-gradient optimization (or any kind of technique that follows an "interesting" direction when the gradient is either not available or "boring"). It's unfortunately not presenting anything new.