It's worth remembering that even psychohistory was "wrong". It's a good example of a model that was "usefully" right but not mathematically true.