We would like to start by exposing a paradox: the 'audience' (or, increasingly, audiences) has been a major focus of television research for the last 30 years. Viewers have been scrutinized, questioned, observed and interviewed according to different research traditions, focusing on different groups (e.g. notably the family, Morley, 1986) and using different methodologies. Different representations of viewers have been produced: passive, active, mobilized, semi-oppositional, etc. At the same time, the 'audience as figures', in other words the audience which is central to the television industry, has been for the most part ignored by the academia or hastily dismissed as an 'administrative', marketing-based vision of audiences.Academics tend to narrow down the activity of television audience measurement (TAM) to some of the numbers it produces, known as 'ratings'; and those are routinely criticized for their effects on culture or on a specific cultural field, such as journalism (notoriously, but without originality, by Bourdieu, 1998). In a less pessimistic tone, the link between ratings and the degradation of culture has been made in the US as well (e.g. Gross, 1997Gross, : 1348. Critics consider ratings with a strange mixture of respect (for their 'accuracy') and rejection (for their 'effects').There have been some more sophisticated approaches to ratings. Smythe (1977, 2006 [1981]) adopted a neo-Marxist approach to analyse the part played by the 'commodity audience' (measured by ratings) in the political economy of the media, inviting his colleagues to pay less attention to media texts, which for him were a 'free lunch' meant to attract audiences to the media in order to sell them to advertisers. This has triggered what is been known as the 'blind spot' debate. However, like Smythe himself, most authors