Low-impact agents — gets reinterpreted as potential power by → Penalize influence

Low-impact agents, Armstrong's informally-discussed precursor concept, is usually read as a claim about realized change: an agent should not alter its environment more than necessary. Penalize influence quietly relocates the target of that claim from realized change to potential change. Its motivating example is a cleaning robot that should avoid bringing "a bucket of water into a room full of sensitive electronics, even if it never intends to use the water in that room" (concrete-problems, §"Penalize Influence:", pp. 5-6): nothing has happened yet, no state has changed, only the option to cause harm has been acquired. This matters because an agent could satisfy every ordinary reading of "low impact" while still accumulating exactly this kind of latent power, making penalize influence a genuine extension of the low-impact target rather than just one more implementation of it.