Subprime Attention Crisis

by Tim Hwang

  Imagine that you’re a car company, looking to place an ad promoting your newest model. Using your DSP, you—or more likely an algorithm working on your behalf—identify a perfect opportunity to place an ad on Car Central, a prominent blog for automobile enthusiasts. As far as you know, the system works great: you bid on the spot, you win the auction, and your ad goes up on the website.

  But sometimes, the DSP and the SSP are actually working against you. The DSP obtained the Car Central inventory at a much lower price and sold it to you at a significant, undisclosed markup. The SSP offering the Car Central placement might not even have access to Car Central—they bought the ad inventory from an SSP with access and then offered it at a significant, undisclosed markup.

  In either case, you pay far more for your Car Central ad than you should. The cumulative effect of these undisclosed margins on price is massive. One study by The Guardian suggests that some 70 percent of the money spent by buyers is consumed by the ad tech platform, with the publisher retaining the remainder.24 Had you been able to negotiate directly with Car Central or with an authorized SSP selling inventory on its behalf, you might have been able to avoid the hidden fees.

  This arrangement creates a significant price mismatch between what a buyer pays for an ad placement and what the publisher selling that ad placement actually receives. Some publishers have responded with legal action. In 2017, The Guardian sued Rubicon Project, an advertising exchange, for charging additional fees to buyers of Guardian inventory without disclosing them to the publisher.25 Though the case settled without any admission of liability, Rubicon Project has since eased up on the hidden fees.

  Like the financial institutions that originated, packaged, and sold mortgage-backed securities in the 2000s, the operators of the programmatic advertising infrastructure have perverse incentives to keep prices high and the market hot. The introduction of hidden fees helps to artificially inflate the price of advertising inventory, even as the quality of attention captured by that advertising declines.

  The overall impact of these shenanigans is twofold. First and most obviously, prices will remain inflated, imposing costs on advertisers and publishers. Less obvious but more dangerous in the long run is that these profits eliminate any incentives to change practices within the industry. The money is simply too good. It will keep the market from adjusting quickly and appropriately to structural weaknesses in the value of online advertising, allowing the formation of a bubble that will continue to grow until it pops.

  What Comes Next?

  Market bubbles are the result of harmful, self-reinforcing cycles between buyers and sellers that escalate out of control. These cycles can be surprisingly robust, and bubbles can continue growing long after the underlying economic situation has changed. Mortgage-backed securities continued to inflate home values long after a close look at mortgages would have revealed the inability of a massive number of borrowers to repay the loans.

  The same is true in the online advertising space. Digital marketing is succeeding in spite of the deep structural issues of fraud, opacity, and falling effectiveness. The shrinking of legacy advertising channels has produced a stream of dollars largely unresponsive to these problems. At the same time, agencies and ad technology companies face perverse incentives to avoid slowing this flow of dollars and, quite the opposite, work to constantly juice the marketplace. The result is that the market for digital advertising grows, divorced from the reality of how ads are actually functioning.

  Bubbles pop, of course. And when they do, it’s loud.

  6

  Exploding the Bubble

  Commodifying attention has had a huge impact on the development of the internet. On the one hand, it has enabled massive economic growth and made a wide set of online services accessible to the public in a way that they might not otherwise have been. On the other hand, commodification has introduced a range of structural vulnerabilities that raise questions about the long-term sustainability of the modern model of digital advertising, and therefore the internet itself.

  The previous few chapters have laid out this case, outlining how the confluence of opacity, falling asset values, and bad incentives seem to echo previous crises. The financial crisis of 2008 has served as an uncannily good guide in highlighting the corrosive market dynamics playing out in today’s online attention markets.

  So what should be done about the precarious state of this economy? Advertising has been the dominant model for making money on the internet in the last few decades. But if the market is fundamentally unstable over the long term, as the financial markets were before the subprime mortgage crisis, what should we do about it?

  We should see these structural instabilities as an opportunity. Online advertising has long exerted a corrosive effect on the design of the internet, and the perverse incentives to inflate the bubble make it difficult for alternative business models to emerge. Rather than trying to fix a broken market, we should work toward a controlled demolition that reduces its influence in the long run. The vulnerabilities I have laid out chart a path forward, not toward fixing what has long been a problematic system, but toward starting the internet anew.

  Against Programmatic Advertising

  There are good reasons to dislike the online advertising economy. Many of these critiques are well established and have existed since advertising emerged as the primary business model for funding the internet.

  First, an advertising-driven online economy relies on effectively invading the privacy of consumers, a model that critics have labeled “surveillance capitalism.”1 Second, incentives exist for online platforms to continuously manipulate user behavior and seize user attention in ways that may be harmful to mental health and personal development.2 Third, online advertising incentives promote the creation of media that is shocking or reaffirming to the viewer, producing polarization and supporting the formation of echo chambers.3 These long-standing critiques have been an aspect of commentary about the internet for decades. One early web pioneer involved in the creation of the pop-up ad, Ethan Zuckerman, has called advertising the internet’s “original sin.”4

  These are general arguments against online advertising. The rise of programmatic advertising—as a distinctly commodified, financialized means of buying and selling attention online—adds a unique set of ills to this list.

  Commodification implies standardization. In online advertising, buyers want to know exactly how much attention they can purchase with their available budget. 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.

  In Seeing Like a State, James C. Scott explores a helpful notion of what he terms “legibility.” In order to administrate at scale, governments and large bureaucracies need to be able to see the world effectively. The result is that the world is actively shaped in order to enable administration. To set up a system of taxation, for instance, it is necessary to create a system of fixed identities so that the government can track over time which people have paid their taxes. Establishing a legible system of fixed identities may necessitate cultural changes, like introducing the concept of a last name to cultures that previously did not have one.5

  Legibility has shaped the media channels through which advertising has flowed. The creation of the television sitcom, with its consistent, recognizable setting and characters, is historically linked to advertising sponsors’ need to approve content and promote products.6

  Social media is no different. The need to create a liquid market in human attention influences the architecture of the social spaces of the web. Commodification requires attention to be legible: in other words, the internet must structure “engagement” in a way that is easy and accurate to measure.

  Social interaction between people is mediated by structured tags such as “like” and “favorite” because these render sentiment easy to measure. Even features that we take fo
r granted, such as requiring user registration to create a profile, are building blocks designed to support the delivery of advertising online.7

  We’ve lived for so long in an online social universe purpose-built for advertising that it is difficult to imagine what an alternative might look like. Consider for a moment an alternative social media platform that we’ll call Super Social Media 3000 (SSM 3000), the bizarro opposite of the advertising-legible sites we use every day. It consists of a single page on which everyone interacts and where everyone sees the same thing. Rather than having structured text boxes, users manually draw shapes and words with their cursors. There are no user profiles, and you do not need to be logged in to use it.

  This is an advertising nightmare! The contributions of users are all jumbled up into an unrecognizable mess. The system logs no demographic information relevant for delivering advertising. In contrast to the discrete, measurable likes on a post on Facebook, a given section of SSM 3000 provides advertisers with only a difficult-to-interpret doodle.

  SSM 3000 would assuredly be a social experience. Users would interact with one another, and would likely make friends and even build communities. But light-years of difference would exist between it and what we currently understand as “social media.” The difference is a distinct lack of the features that advertising has incentivized and helped mold. By and large, we don’t have platforms like SSM 3000. This is because the broad range of expression that the internet might otherwise enable has been limited to ways of connecting that are consistent with the financial needs of advertising. The free-form scribblings of SSM 3000 are financially unsustainable compared to the shallow paradigm of likes, retweets, and short comments. In this sense, advertising is complicit in restricting the grammar of social interaction online.

  Even worse, commodification is similar to other processes of standardization in that it forces new entrants to comply with the existing state of play in the market. The next Facebook or Google, to the extent that it is also driven by advertising, will need to structure attention in a way that makes its attention inventory salable in the broader marketplace. This means that securitization may be a one-way street: decisions made in the formative stage of the web limit what is possible in the future.

  This is more than simply enforcing a certain kind of product design. We have been taught to interact with other people online by platforms built to buy and sell attention. One wonders if that will constrain the social possibilities of the future. At first glance, it might seem that no one would want to use SSM 3000: the anonymity and lack of clear individual spaces might degrade into a digital wall of bathroom graffiti in a few hours (or less). I don’t doubt that it would.

  But that deterioration says less about fundamental human nature than it says about how we have learned to interact online. Society has only constructed online communities within the context of a web structured by online advertising, and so no norms and practices currently exist for using a space like SSM 3000. In this sense, we may be locked in. Our ingrained approach to interacting with others online assumes the features of an advertising-driven internet. This may make it hard or impossible to build alternative online social networks that do not collapse into anarchy.

  These numerous arguments amount to a condemnation of the financialized infrastructure that enables much of the web to function. There is, if anything, a strong ethical imperative to allow the collapse of global surveillance capitalism rather than attempting to save it, because it might clear the deck for something better to emerge.

  Controlled Demolition

  The companies built on digital advertising are an imposing part of the economic landscape. Google and Facebook appear to occupy an impregnable position, sucking in an ever-increasing amount of cash with no end in sight. Throughout the web, programmatic advertising remains pervasive, a dominant paradigm for making platforms financially sustainable if not extraordinarily profitable.

  This makes discussions of how to structurally alter the web feel futile. It seems impossible to tear out the money machine powering the internet, and of little use to create alternatives when the existing model is so ingrained and so profitable. Programmatic advertising simply seems like the state of affairs that will govern the structure of the internet forever. The problems that come with these monetization models, too, appear to be chronic, always mitigated but never truly resolved.

  Against this backdrop, the structural issues at the heart of programmatic advertising present not a threat but an opportunity. The collapse of the global digital advertising markets would produce a great opportunity for alternative business models to take shape, and a chance for the internet itself to take a different shape as well.

  We could take an approach of benign neglect here. If one buys the argument that ad fraud, ad blocking, and an ever-increasing indifference toward ads will eventually break this marketplace on their own, why not just sit back and let it self-destruct?

  Patiently waiting for programmatic advertising to break is an attractive position because it demands little of us, but it misses the bigger picture for a number of reasons. First, bubbles grow larger over time. The bigger the bubble, the harder the fall. From an economic management perspective, it is far better for a bubble to collapse early than for it to grow to a size where a market panic could produce massive collateral damage throughout the economy. As we have seen, the industry is well aware of the structural issues that threaten to upend the viability of the programmatic advertising markets, but is unlikely to make the radical changes necessary to address these problems. This will lead to partial, temporary fixes that only prolong the growth of the bubble and the associated problems that come with it.

  Second, waiting for the bubble to burst on its own deprives us of the ability to distribute the social costs of such a downturn in a just and equitable manner. It is important to remember that it is not just huge advertisers and dominant digital platforms that rely on the continued functioning of the online advertising ecosystem. Like it or not, advertising is a critical, if tenuous, force for funding journalists and a vast universe of smaller media outlets and niche media.8 They too would feel the costs of a widespread failure in the online advertising markets.

  These linkages go beyond the media. One of the perverse developments of online advertising is that its outsize profitability has allowed it to float enterprises that might otherwise be insolvent. Like the banks during the subprime crisis, dense links between advertising-driven businesses and the rest of the economy mean that vulnerability in the attention marketplace may produce harm in other, less expected places. The sudden collapse of the profitability of online advertising in, say, Google Search might affect the availability of free services like Google Scholar or reduce overall investment in self-driving vehicle technology.

  This is not to mention the impact on society at large. In a distressed situation where previously free services throw up paywalls and limit features to premium tiers of service, accessibility will turn on the ability to pay. Not everyone will be able to afford a subscriber-only version of the web, and such a transition could deny vulnerable segments of the population access to the critical services that they have come to rely on. Failing to address these issues in advance would be a cruel kind of irresponsibility.

  Third, an uncontrolled popping of the bubble is less likely to lead to permanent change. In the absence of a clearly articulated, workable alternative, one might imagine a serious downturn after which the existing advertising ecosystem dusts itself off and resumes along the same path again. That would mean the return of the same markets with the same faults. Ensuring robust, sustained change in the structure of the web requires creating the space and opportunity for real alternatives to programmatic advertising to emerge.

  Like in the demolition of a building, the safest approach might be to bring about a well-considered and structured implosion, rather than allow for an out-of-control collapse that might harm bystanders and create unintentional damage elsewhere. To that end, we may no
t want to simply wait for the crisis to arrive. Instead, it may be important for us to actually accelerate and bring about the collapse of the programmatic advertising bubble in ways that allow us to control the consequences.

  Starting a Manageable Crisis

  Programmatic advertising is a bubble. The potent combination of opacity, perverse incentives, and ever-eroding value has produced this bubble. But ultimately, financial crises depend on a crisis of confidence: a spreading panic that a highly sought-after asset is perhaps worth far less than originally speculated. How might we provoke such a crisis of confidence in the online advertising economy, particularly when billions of dollars are steadfastly committed to ignoring these warning signs? Can we control that crisis in such a way that its consequences do not spin beyond our control?

  We’re aided by one significant difference between the mortgage-backed securities market of 2007–2008 and the programmatic advertising market of today. In the subprime crisis, the mortgages at the root of a complex system failed. Loans were issued to home buyers who were unable to pay, resulting in defaults that upended the global economy. The panic over the viability of the assets being traded in the market was, in other words, built into the very structure of these marketplaces. Once this trip wire was triggered, the resulting cascade was impossible to manage.

  In contrast, the troubles that plague the advertising economy are creeping threats. People pay less and less attention to advertisements online and steadily adopt ad blocking. Click fraud has proven to be pernicious, eroding the value of advertising inventory and resisting attempts to curtail it. Trust continues to decline between ad buyers and the wide array of players in the ecosystem with conflicting goals: the agencies, ad exchanges, online platforms, and publishers that facilitate the distribution of ads online. This slow, multifaceted creep of a range of structural problems means that the market can be stepped down in stages over time, with a series of targeted crises helping to slow and eventually reverse the flow of money into the ecosystem.