Advertising drives the modern digital economy. Whether it’s reading news sites like this one or perusing your social media feeds, advertising is the single most important industry that came out of the development of the web. Yet, for all the tens of billions of dollars poured into online advertising just in the United States alone, how much does that money actually do its job of changing the minds of consumers?
Tim Hwang has a contrarian stance: it doesn’t. In his new book published as a collaboration between Logic Magazine and the famed publisher Farrar, Straus and Giroux, he argues in “Subprime Attention Crisis” that the entire web is staring into an abyss of its own making. Advertising is overvalued due to the opaqueness of the market, and few actors are willing to point out that the advertising emperor has no clothes. Much like the subprime mortgage crisis, once people come to realize the true value of digital ads, the market could crater. I found the book provocative, and I wanted to chat further with Hwang about his thoughts on the market.
Hwang formerly worked at Google on policy and has developed many, many projects across a whole swath of tech-oriented policy issues. He’s currently a research fellow at Georgetown’s Center for Security and Emerging Technology.
This interview has been condensed and edited for clarity.
TechCrunch: Let’s dive straight into the book. How did you get started on this topic of the “subprime attention economy”?
Tim Hwang: There were two incidents where I was like, something is going on here. I was having conversations with a couple of friends who are product managers at Facebook, and I remember making the argument that that there’s a lot of evidence to suggest that this whole adtech thing is maybe just mostly garbage. The most interesting thing that they said was, “Oh, like, advertising works but we can’t really tell you how.” That’s like talking to someone from the national security establishment and they’re like, “Oh yeah, we can stop terrorists but, like, we can’t tell you exactly how that goes down.”
I think one thing that got me really interested in it was how opaque a lot of these things are. The companies make claims that data-driven programmatic advertising really is as effective as it is but then they’re kind of strangely hesitant to show evidence of that.
Second, I was doing research with a lot of people who I think you’d rightly call sort of tech critics — strong critics of the power that these platforms have. I think one of the most interesting things is that even among the strongest critics of tech, I think a lot of them have just bought this claim that advertising and particularly data-driven advertising is as powerful as industry says it is.
It’s a kind of strange situation. Tech optimists and tech pessimists don’t agree on a whole lot, but they do seem to agree on the idea that this sort of advertising works. That was what I wanted to explore in the book.
Why don’t we talk a bit about the thesis?
The thesis of the book is really quite simple, which is you look around and basically our modern experience of the web is almost entirely shaped by advertising. The way social media is constructed, for example, is largely as a platform for delivering ads. Engagement with content is really good for creating profiles and it’s really good for delivering ads. It really has been the thing that has powered the current generation of companies in the space.
As you sort of look closer though, it really starts to resemble the market bubbles that we know of and have seen in other places. So explicitly, the metaphor of the book is the subprime mortgage crisis. I think the idea though is that you have this market that is highly opaque, there’s a lot of evidence to suggest that the value of ads is misidentified, and you have a lot of people interested in boosting it even in spite of all that.
For the book, I wanted to look at that market and then what the internet could look like after all this. Are there other alternative business models that we want to adopt for the web going forward?
Let’s get to the central argument of the book, which is that the ad market looks like the subprime mortgage market. Where does the metaphor work and where does it sort of fall apart?
I think the subtlety with the book is that on one hand you have the subprime mortgage crisis and on the other hand you have the programmatic ad market. I’m not necessarily arguing that programmatic ads are one-for-one like the subprime mortgage crisis, but in fact that both of them are examples of market bubbles. That’s the grounding force — what are the attributes of market bubbles and do we see those market bubble dynamics playing out in the programmatic ad market?
In the mortgage market, you had collateralized debt obligations that were circulating around. It’s really unclear what’s exactly inside those collateralized debt obligations and if you go looking, it doesn’t actually end up looking as great as people say it is. I think it’s similar in the ads case.
I think the second thing that is a very big parallel is just sort of the conflicted actors. When you get into the subprime mortgage crisis, the banks were like “Yep, you should keep buying these collateralized debt obligations because they’re a great asset” and they were compensated for those sales. I think you have a similar dynamic playing out in adtech, where we need to sell you on the value of this data-driven advertising regardless of whether or not it is actually providing real value.
I don’t want to oversell the metaphor. But, I think the question in the book is what would a crisis of confidence look like that would be big enough to create that kind of panic in the marketplace?
We’ve had a few scandals in the adtech space — Facebook famously inflated the view metrics of its video advertisements and ultimately settled a lawsuit on the subject. Why didn’t those issues pop your ads bubble?
My main hypothesis there is two-fold. One of them is that a lot of the media companies believed this story that essentially Facebook says everybody’s watching video on Facebook, and you need to hire a bunch of video producers to produce content for the newsfeed. They overstate the amount of time that people are actually spending looking at these things, and I think there my theory is that the impact just wasn’t as massive. It wasn’t really widespread through the entire marketplace of advertising.
Second, I’ve been looking into things like privacy laws such as GDPR as popping the bubble in some ways, because there’s a potential to basically show that the building blocks of advertising (such as personalized data) might not be worth as much as people think. It’s the structural changes that will be the most dangerous for the ad market.
Another major counterpoint to this thesis is look, you have real-time auctions for these ads. If they weren’t useful or effective, companies and buyers would bid less, and then these ads would therefore be worth less. Why doesn’t the free market just work itself out?
I think this is actually where I’m hoping to have a debate with the adtech folks. We have seen cost per impression go down over time as clickthroughs have gone down over time. So, price has been been changing, there’s no doubt about that. The question is how much does it actually reflect the current value of the asset? Certainly, I think the value has gone down, but it’s hard to tell because the market is extremely murky.
I talk to people who are in the tech industry who are like this is totally right and the price is reflecting the real value. And I think the companies are very reticent to give up the data that would allow you to find a really definitive answer to that question.
You cited Tim Wu’s “The Attention Merchants,” and I’m curious: Does advertising ultimately work? Does all this money and time spent getting in front of consumers actually lead to … better sales or what have you?
I’m glad you brought up “The Attention Merchants” because I do think Tim Wu’s historical views on these issues is really good. I think the argument is to say like, maybe it’s not that digital advertising doesn’t work, but it may just be as bad as earlier types of ads. The claim that data-driven programmatic advertising is going to solve the kind of long-standing questions with ads is, I think, not the case.
I think that position is maybe a little bit more nuanced. We clearly have examples of ads working. The question is, do you think that’s the norm? Do you think that the 80%-90% of ads delivered on banners and search and online work? I’m a little skeptical.
I think one of the really interesting things that comes out of the academic literature is the effect is so small and that it’s really expensive to run experiments to know whether or not it has an effect.
One of the points that you make in the book is that our current experience of the web is basically contingent on decisions that had quite a bit of range in the past but now are quite locked in because our mental model of what an experience on the web can be is built around the ad economy. Do you think there’s potential to migrate to alternative models of what the web could look like?
Well, this is actually one of my worries. I think one of the things that I observed is that people expect social platforms to be designed in ways that are built on top of this sort of advertising presumption, and I do think it actually does constrain the ability for new things to enter the marketplace. To be optimistic I have to believe that we can get people to adopt new platforms, but I do think ads make the climb a little bit steeper than it otherwise would be.
One of the things I have been thinking a lot about since the book came out though is the idea that advertising in some ways allowed us to put off some pretty hard questions about the internet. You’ve been offered it all for free through advertising and the psychological cost of this is zero dollars and zero cents.
Finishing up here, do you think the largest ad companies like Google and Facebook are ready to have a deeper conversation around this?
No, I don’t think they are. I think they have resisted third-party transparency, which has been a really interesting fight between the buyers and the ad platforms.
I’ve already heard some anecdotes from people since the book came out. They were people who work for the big tech companies who were raising these questions and they finally got the cold shoulder, and that doesn’t surprise me because I think it’s kind of a dangerous question to these companies.