Its a trope at this point, but a major point missed by the author is opportunity cost. Haven't we reached a point where optimizing that many basis points of incremental sales isn't worth what new solutions we could build with all these engineers' time?