I think when something is very uncertain (because it's hard and/or the engineer is inexperienced), it's not possible to develop a mental model that lets you pla

I think when something is very uncertain (because it's hard and/or the engineer is inexperienced), it's not possible to develop a mental model that lets you plan a course of action. However, taking a slightly directed, non-destructive action can sometimes reveal more information to you. The idealization of this is coming up with some hypothesis, and testing it, which I think is covered by your approach. But what if you don't know enough to even know the right questions to ask. You may be dead in the water if you don't allow an unplanned action.