No, because the reduction in-principle would fail if the system's description language could not support explaining the phenomena in the higher level. And going the other way, it could be impossible in-principle to conduct the experiments that get a complete description because the tools required break how it was originally organized.
It's also not just about adding more variables because "it's complicated". Biology for example seems to be highly reflexive, that sort of behaviour can't really be accounted for in basic causal models. At best we have models which resemble fluid dynamics, and fluid dynamics is only as good as it is because we don't reduce the phenomena down to every molecule.
It's also not just about adding more variables because "it's complicated". Biology for example seems to be highly reflexive, that sort of behaviour can't really be accounted for in basic causal models. At best we have models which resemble fluid dynamics, and fluid dynamics is only as good as it is because we don't reduce the phenomena down to every molecule.