if we solve distributional shift, do we also solve ontological crises?