Why would the desired outcome be a more true test than the actual outcome?