And should the model prove more or less totally wrong, how many more times should he be allowed to discuss forecasts?