In the 30 years or so that we have been developing a safety methodology to accompany the growing presence of software in safety-related systems, understanding of the software development process itself hardly seems to have advanced. We still use in most part the same languages we did then but duly bloated to match the uncontrollable growth of software itself. We still teach entire paradigms without any basis in the scientific method whatsoever whilst the amount of software continues to grow alarmingly, particularly in the automotive industry. The result is an absence of any real forensic basis for understanding failure and subsequently avoiding it. This short essay looks at some of the reasons why, and demonstrates from recent results in information theory, that this stems from the fact that we are probably barking up the wrong tree.