1 of the things I work on regularly was originally created with a fairly consistent data set. We had a very good idea of what fields were going to be in the data. As a result, we thought we knew where exactly we needed to check for null data, and when we didn’t need to know. Then the data set expanded, and the new data wasn’t nearly as consistent as what we started with. That’s when we realized that we were very much wrong in thinking we knew what needed null checks.