Subprime Attention Crisis

Tim Hwang, who worked on public policy at Google, argues that the programmatic digital advertising market has structural similarities to the subprime mortgage market before 2008, and that it may be similarly overvalued.

His argument runs through several claims. Attention is bought and sold through automated exchanges as a commodity, in units of impressions whose actual value is poorly measured and highly variable. The market is opaque: buyers cannot see where their advertisements appear or verify that a human saw them. Fraud is substantial, with bot traffic and domain spoofing consuming a meaningful share of spending. Measurement of effectiveness is weak, and the studies that have used proper experimental design, including work at eBay and by Facebook's own researchers, have repeatedly found effects far smaller than the industry reports. Meanwhile ad blocking and user indifference reduce actual attention delivered.

Hwang's conclusion is that the market rests on a belief about the value of targeted advertising that has not been established, that the belief could correct suddenly rather than gradually, and that because digital advertising funds most of the open web, a correction would have consequences well beyond the advertising industry. The book is short and argues by analogy, which is its weakness: the mortgage comparison is suggestive rather than demonstrated, and Hwang does not engage seriously with the case that targeting does work at scale for direct response advertisers. The material on measurement and on fraud is well documented.