Subprime Attention Crisis by Tim Hwang

Rating: β˜…

This book bugged the shit out of me.

The central argument of Subprime Attention Crisis is that programmatic advertising is overvalued and poised to collapse similar to the 2007 subprime mortgage crisis, and that this will result in chaos on the internet. It’s not a good argument. I was very, very frustrated with the book because I read it shortly after The New Lombard Street and The Code of Capital. Unlike those books, Subprime Attention Crisis is imprecise and sloppy when it deals directly with finance and economics. If you're going to make a direct comparison to the subprime mortgage crisis, you better get it right.

One cool idea:

Before I get to more of the bad, I do want to bring up one neat idea: "Publishers want to streamline sales by providing a simple menu of attention inventory to buy. This requires that online platforms be actively architected in a way that standardizes user engagement and attention." So the broad range of expression that the internet could enable is limited to what can serve ads. "In this sense, advertising is complicit in restricting the grammar of social interaction online."

An incomplete list of things I found suspect about the book:

  • The author speculates that ads may have as much systemic importance as banks. This totally ignores 1) how much bigger banking is and 2) how much more interconnected banking is. Unlike in banking, if one ad seller blows up, it doesn't cause a contagion of failure.
  • Some of the only economists cited are the disgraced Reinhart and Rogoff.
  • Extremely bad passage: "In financial parliance, both CDOs and online advertising inventory are derivatives - they derive their value from an underlying asset. CDOs draw their value from the mortgages they contain; online ad inventory draws its value from the attention that it represents." The author literally defined how ads are comparable to commodities a few pages earlier. This is like saying that commodity corn is a derivative because it's abstracted from individual kernels. It's all the more absurd because true ad derivatives are being proposed like ad futures.
  • I wish the author would just make value judgments instead of saying things like this: "If it turns out that the data collected by advertisers are not in fact accurate, or that the data are not in the end all that useful for shaping perceptions and behavior, then programmatic advertising is not 'better' in any real sense" – if you think programmatic ads are bad, just say so! Things can be effective and harmful.
  • The author seems to have a general inability to distinguish supply- and demand-side forces. The author says several times that ad sellers keep prices high, but this totally ignores that ads are being bought. The demand side is buying – the "true" value of these ads is what the market will support!