The ways of knowing modern audiences, broadcast ratings, is now big billion dollar business, with Google in alliance with Nielsen, one of the world's leading media researchers, to literally auction audiences off to the highest bidder. Ironically, 'the black box' of survey and sampling methodology that delivers audiences remains a mystery to most, just as Art Nielsen's original 'black box', the Audimeter, for measuring radio sets baffled onlookers. The black box, though, is in crisis. It is beset by a rapidly changing media and consumer environment, heightened demand for ever more precise information, increasing measurement difficulty and declining respondent participation. Sampling has also become the subject of political and methodological battles of how to represent people, including everything from the Census to the Internet. Recent US research on non-response shows that technology rich households are rejecting participation in audience surveys. The joint Arbitron and Nielsen Project Apollo, designed to collect everything about audience behaviour also found the people were rejecting both the intrusion into their private lives and the tasks given to them. In this paper the authors outline the contemporary survey and sampling wars in modern audience ratings, including especially the proliferation of different ways of measuring audiences and audience rejection of many of them.