Support versus substitution is how the framework tells healthy assistance from the slow hollowing of a participant. Support helps a person carry their own geometry; substitution carries it so completely that the capacity never develops or is gradually lost. The same tool can do either — the same navigation aid can teach someone a city or ensure they never learn it — so the distinction is not about the technology but about its effect on future capacity.
From it follows the design rule that, outside a genuine emergency, assistance should default toward capability transfer: repeated help should leave the participant more able to act without it, not less. This is the mechanism underneath competence-preserving dependency, and it is why the framework wants dependency kept visible — the difference between support and substitution is easy to miss from the inside, because a capacity someone has quietly stopped exercising still feels like one they possess.
In care, presence, and art, who does the work is part of its worth, so people are owed disclosure of whether a human or an AI is doing it.