Zheng Wang comments on Why and When Interpretability Work is Dangerous