@sue when I was at JPL I found it frustrating that we spent months writing specs before writing any code. Most of the specs would be wildly inaccurate by the time you were writing the actual code, because of other parameters you'd need, differences in how you broke out the work into individual functions, etc... So tedious.