Stuart_Armstrong comments on Optimization Regularization through Time Penalty