So perhaps we've been focusing on the wrong series of problems when it comes to software development productivity. Doing the HOW is not the bottleneck, but rather determining (rigorously enough) the WHAT that is the bottleneck.