Cart
[Home] [By Thread] [By Date] [Recent Entries]
Perhaps one way is to distinguish 'data for processing' vs. 'data for reporting'. Software vs. wetware. In the first case, if a process has a requirement for both first-order and second-order information, little alarm bells should ring. If many processes are deriving second-order results from from first-order input, how do you ensure such results are the same, or that discrepencies are handled correctly? In some cases, the calculating or deriving process description may represent first-order data, also, and should be passed forward as such (if it can be serialized or referenced). For reporting, just about anything goes, I imagine. Humans have amazing cognitive abilities. Nothing in this business is cut and dried, but there is an impact on the art of requirements analysis and the analyst's responsibility to make sure all the loose ends are tied and consequences are anticipated fully.
|
