Yes. Of the fields I've implemented algorithms from papers in, computer vision seems to be particularly adverse to discussing the (often critical) downsides of their algorithms. Often its something like "Camera calibration isn't exactly perfect? Well this scene reconstruction technique simply wont work."