Stuart_Armstrong comments on Maximizing a quantity while ignoring effect through some channel