TurnTrout comments on How much can value learning be disentangled?