“According as circumstances are favourable, one should modify one's plans.”
A plan is a hypothesis; when the ground shifts, the plan must shift or it becomes a liability.
With counsel accepted, Sun Tzu adds the necessary counterweight: modify plans as circumstances prove favourable. The verse rescues Chapter 1 from the misreading it would otherwise invite — the temple calculations are not a doctrine to die for but the best available map, redrawn as the territory reports back. Planning and revision are one continuous act, not phases.
A product roadmap commits to an enterprise tier by Q3. In Q2 the data returns something unexpected: the bottom-of-funnel usage that was supposed to be a stepping stone is itself the product. The team that "modifies according to circumstances" kills the roadmap item and doubles the funnel — the team that treats the plan as a promise spends Q3 building the wrong thing well.
Flexibility is not plan-rot. Sun Tzu says modify as circumstances warrant — meaning the circumstances must genuinely warrant it, in evidence, not in mood or in the discomfort of difficulty. The test cuts both ways: a plan never changed is dogma, and a plan changed every week was never a plan. The constant is the assessment discipline; only its output moves.