The online platforms that we access via our telecommunications infrastructure are fundamentally changing the way advertising works, with significant social consequences. Advertising has an important role to play not just as a form of market signalling but also as a form of social messaging and, at times, as a means of discrimination and social sorting. We know that advertising has been used for progressive as well as pernicious purposes, providing public service messages, but also reinforcing harmful ethnic and gender stereotypes, and in some cases for spreading false or misleading information. For this reason, it is important to hold advertising systems and advertisers accountable for the forms of messaging they promulgate.

Accountability is much harder to achieve in online contexts than in the mass media era, thanks to the combination of data-driven micro-targeting with personalised devices and lagging regulatory frameworks. Commercial messaging on social media takes the form of so-called 'dark ads' that are visible only to those to whom they are delivered. This form of targeting means that the content is not generally available for public inspection and also that the pattern of distribution of ads is non-transparent, even to those who receive the ads. Moreover, online ads are often very short-lived, undergoing constant transformation, which renders them ephemeral and thus difficult to track and hold accountable for their content and their pattern of distribution.

This project develops one model for providing accountability for dark ads on Australia’s most popular social media platform, Facebook. By showing what accountability looks like, the research reminds us just how dramatically the advertising environment has shifted, and of the kinds of questions we need to be asking to ensure that dark ads are not abused in ways that are detrimental to society.

The research tool developed represents the first attempt to make public not just what ads are being served to people online, but how advertising is distributed across demographic groups. The tool enlists volunteers to install a browser extension that captures the ads appearing in their Facebook news feeds. When they install the extension, they are asked to provide some demographic information about themselves. The ads are collated by the tool and displayed in a form that can be filtered by demographic category. This allows researchers to query the database by, for example, sex, or income level, or party preference – or some combination thereof – to see which groups receive what types of ads.