Predicting the actual results of real unpublished experiments with a 0.9 correlation factor is a very non-trivial result. The human forecasts comparison is not the central finding