SYSE.19:2 - Problem

Two shortcuts cause most of the damage. One repeats every downstream activity merely because the source label changed. The other keeps every decision merely because the engineered System did not change. Both avoid the central question: which relied-on claim changed enough to alter engineering action?

Other recurring failures appear when the source episteme cannot be identified, reliance is known only at document level, a semantic difference is confused with a text diff, revalidation uses the wrong configuration, or a passing test is treated as permission or release. The project then either spends effort without a decision or keeps a decision whose premise no longer applies.