The Internet's Original Sin
It's not too late to ditch the ad-based business model and build a better web.
Ron Carlson’s short story “What We Wanted To Do” takes the form of an apology from a villager who failed to protect his comrades from marauding Visigoths. It begins:
What we wanted to do was spill boiling oil onto the heads of our enemies as they attempted to bang down the gates of our village. But as everyone now knows, we had some problems, primarily technical problems, that prevented us from doing what we wanted to do the way we had hoped to do it. What we’re asking for today is another chance.
There’s little suspense in the story—the disastrous outcome is obvious from the first paragraph—but it works because of the poignancy of the apology. All of us have screwed up situations in our lives so badly that we’ve been forced to explain our actions by reminding everyone of our good intentions. It’s obvious now that what we did was a fiasco, so let me remind you that what we wanted to do was something brave and noble.
The fiasco I want to talk about is the World Wide Web, specifically, the advertising-supported, “free as in beer” constellation of social networks, services, and content that represents so much of the present day web industry. I’ve been thinking of this world, one I’ve worked in for over 20 years, as a fiasco since reading a lecture by Maciej Cegłowski, delivered at the Beyond Tellerrand web design conference. Cegłowski is an important and influential programmer and an enviably talented writer. His talk is a patient explanation of how we’ve ended up with surveillance as the default, if not sole, internet business model.
The talk is hilarious and insightful, and poignant precisely for the reasons Carlson’s story is. The internet spies at us at every twist and turn not because Zuckerberg, Brin, and Page are scheming, sinister masterminds, but due to good intentions gone awry. With apologies to Carlson:
What we wanted to do was to build a tool that made it easy for everyone, everywhere to share knowledge, opinions, ideas and photos of cute cats. As everyone knows, we had some problems, primarily business model problems, that prevented us from doing what we wanted to do the way we hoped to do it. What we’re asking for today is a conversation about how we could do this better, since we screwed up pretty badly the first time around.
I use the first personal plural advisedly. From 1994 to 1999, I worked for Tripod.com, helping to architect, design, and implement a website that marketed content and services to recent college graduates. When that business failed to catch on, we became a webpage-hosting provider and proto-social network. Over the course of five years, we tried dozens of revenue models, printing out shiny new business plans to sell each one. We’d run as a subscription service! Take a share of revenue when our users bought mutual funds after reading our investment advice! Get paid to bundle a magazine with textbook publishers! Sell T-shirts and other branded merch!
At the end of the day, the business model that got us funded was advertising. The model that got us acquired was analyzing users’ personal homepages so we could better target ads to them. Along the way, we ended up creating one of the most hated tools in the advertiser’s toolkit: the pop-up ad. It was a way to associate an ad with a user’s page without putting it directly on the page, which advertisers worried would imply an association between their brand and the page’s content. Specifically, we came up with it when a major car company freaked out that they’d bought a banner ad on a page that celebrated anal sex. I wrote the code to launch the window and run an ad in it. I’m sorry. Our intentions were good.
Cegłowski’s speech explains why Tripod’s story sounds familiar. Advertising became the default business model on the web, “the entire economic foundation of our industry,” because it was the easiest model for a web startup to implement, and the easiest to market to investors. Web startups could contract their revenue growth to an ad network and focus on building an audience. If revenues were insufficient to cover the costs of providing the content or service, it didn't matter—what mattered was audience growth, as a site with tens of millions of loyal users would surely find a way to generate revenue.
There are businesses, Cegłowski notes, that make money from advertising, like Yahoo and Gawker. But most businesses use advertising in a different way. Their revenue source is investor storytime:
Investor storytime is when someone pays you to tell them how rich they’ll get when you finally put ads on your site.
Pinterest is a site that runs on investor storytime. Most startups run on investor storytime.
Investor storytime is not exactly advertising, but it is related to advertising. Think of it as an advertising future, or perhaps the world’s most targeted ad.
Both business models involve persuasion. In one of them, you’re asking millions of listeners to hand over a little bit of money. In the other, you’re persuading one or two listeners to hand over millions of money.
The key part of investor storytime is persuading investors that your ads will be worth more than everyone else’s ads. That’s because most online ads aren’t worth very much. As a rule, the ads that are worth the most money are those that appear when you’re ready to make a purchase—the ads that appear on Google when you’re searching for a new car or for someone to repair your roof can be sold for dollars per click because advertisers know you’re already interested in the services they are offering and that you’re likely to make an expensive purchase. But most online advertising doesn’t follow your interest; it competes for your attention. It’s a barrier you have to overcome (minimizing windows, clicking it out of the way, ignoring it) to get to the article or interaction you want.
A back of the envelope analysis from Felix Stalder gives a sense of how little these ads are worth. Last quarter, Facebook reported that it had 1.32 billion users, collected $2.91 billion in revenue and made a profit of $791 million, for a profit margin of 27 percent. Facebook is clearly doing a great job making money from ads. But the profit per user is just under $0.60. That’s a fascinating figure, because Facebook reports that users spend 40 minutes per day on the site, or roughly 60 hours per quarter.
Stalder is interested in the idea that users are working for Facebook, generating content that the company profits from without getting compensated. But even if we ignore the important idea of “free cultural labor” that makes a business like Facebook (or Tripod!) possible, it’s striking that our attention, as viewers, is worth only a penny an hour to Facebook’s advertisers.
Don Marti uses the same set of Facebook earning numbers to demonstrate that print newspapers make roughly four times as much money in advertising as Facebook does in the United States. Print advertising generates these enviable, if shrinking, numbers despite capturing only about 14 minutes a day of Americans’ attention. This “print dollars, digital dimes” problem is an apparent paradox: Why are targeted digital ads worth an order of magnitude less than untargeted print ads, in terms of “attention minutes”? Marti argues that advertising in a public place, like a newspapers, builds brands in a way that private, targeted ads can’t. (I’ve argued that this is a legacy effect and that print ads will fall in price once there are efficient digital ways to reach the majority of consumers in a market.)
Cegłowski tells us that it doesn’t matter.
The poor performance of digital ads just makes investor storytime more compelling. After showing how poor YouTube’s targeted ads are in understanding him as a consumer, he explains, “Of course, for ad sellers, the crappiness of targeted ads is a feature! It means there’s vast room for improvement. So many stories to tell the investors.”
Most investors know your company won’t grow to have a billion users, as Facebook does. So you’ve got to prove that your ads will be worth more than Facebook’s. In 1997, I argued that Tripod’s users were more valuable to advertisers than the average web user because I could use algorithms to analyze the home pages they posted and target ads to their interests and demographic data. Facebook makes a vastly more sophisticated version of that argument, and faces problems much like those we faced almost two decades ago. Targeting to intent (as Google’s search ads do) works well, while targeting to demographics, psychographics or stated interests (as Facebook does) works marginally better than not targeting at all.
Demonstrating that you’re going to target more and better than Facebook requires moving deeper into the world of surveillance—tracking users' mobile devices as they move through the physical world, assembling more complex user profiles by trading information between data brokers.
Once we’ve assumed that advertising is the default model to support the Internet, the next step is obvious: We need more data so we can make our targeted ads appear to be more effective. Cegłowski explains, “We’re addicted to ‘big data’ not because it’s effective now, but because we need it to tell better stories.” So we build businesses that promise investors that advertising will be more invasive, ubiquitous, and targeted and that we will collect more data about our users and their behavior.
I have come to believe that advertising is the original sin of the web. The fallen state of our Internet is a direct, if unintentional, consequence of choosing advertising as the default model to support online content and services. Through successive rounds of innovation and investor storytime, we’ve trained Internet users to expect that everything they say and do online will be aggregated into profiles (which they cannot review, challenge, or change) that shape both what ads and what content they see. Outrage over experimental manipulation of these profiles by social networks and dating companies has led to heated debates amongst the technologically savvy, but hasn’t shrunk the user bases of these services, as users now accept that this sort of manipulation is an integral part of the online experience.
Users have been so well trained to expect surveillance that even when widespread, clandestine government surveillance was revealed by a whistleblower, there has been little organized, public demand for reform and change. As a result, the Obama administration has been slightly more transparent about government surveillance requests, but has ignored most of the recommendations made by his own review panel and suffered few political consequences. Only half of Americans believe that Snowden’s leaks served the public interest and the majority of Americans favor criminal prosecution for the whistleblower. It’s unlikely that our willingness to accept online surveillance reflects our trust in the American government, which is at historic lows. More likely, we’ve been taught that this is simply how the Internet works: If we open ourselves to ever-increasing surveillance—whether from corporations or governments—the tools and content we want will remain free of cost.
At this point in the story, it’s probably worth reminding you that our intentions were good.
What I wanted to do was to build a tool that allowed everyone to have the opportunity to express themselves and be heard from anywhere from a few friends to the entire globe. In 1995, there weren’t a lot of ways to offer people free webpage hosting and make money. Charging users for the service would have blocked most of our potential customers—most of the world still doesn’t have a credit card today, and fewer did in 1995. E-payment systems like PayPal didn’t come online until 1999. But because Tripod’s services were free and ad supported, users around the world found us and began posting webpages they could not host elsewhere.
In 1996, we noticed that the majority of our users were coming from four countries: the United States, Canada, the U.K., and Malaysia. Since none of our content was in Bahasa Malay and since we’d never done any outreach to Malaysian users, this was a surprise. I started printing out heavily trafficked webpages posted by Malaysian users and brought a sheaf of them to a professor at nearby Williams College, who read them over and informed me that we had become a major vehicle for expression for Malaysia’s opposition political group, Anwar Ibrahim's Reformasi movement.
The adoption of Tripod by Malaysian activists was not directly due to our use of an ad-supported model, but it was an unintended, positive consequence. We couldn’t find a way to make money from advertising to Malaysian users, and we had internal discussions about whether we should “cut our losses” and provide services only to users in countries where we could sell advertising, conversations that Facebook and other ad-supported companies are now wrestling with as they expand in the developing world. I’m glad that we made the right decision (morally, if not fiscally) and that Facebook, thus far, has done so as well.
The great benefit of an ad supported web is that it’s a web open to everyone. It supports free riders well, which has been key in opening the web to young people and those in the developing world. Ad support makes it very easy for users to “try before they buy,” eliminating the hard parts of the sales cycle, and allowing services like Twitter, Facebook, and Weibo to scale to hundreds of millions of users at an unprecedented rate. This, in turn has powerful network effects: Once all your high school classmates are on Facebook, there’s a strong temptation to join, even if you don’t like the terms of service, as it’s an efficient way to keep in touch with that social circle.
In theory, an ad-supported system is more protective of privacy than a transactional one. Subscriptions or micropayments resolved via credit card create a strong link between online and real-world identity, while ads have traditionally been targeted to the content they appear with, not to the demo/psychographic identity of the user. In practice, part of Facebook’s relentless drive to ensure all users are personally identifiable is to improve targeting and reassure advertisers that their messages are reaching real human beings.
And while we’re considering the benefits of an ad-supported web, it’s worth exploring the idea that the adoption of advertising as a business model normalized the web much more quickly than it otherwise would have spread. Companies like Tripod worked to convince massive companies that were at least a decade from selling online, like auto manufacturers, that they needed a presence on the web to build their brand in an important new media. (Second Life unsuccessfully tried the same strategy a decade later, persuading Pontiac to open dealerships for virtual cars.) Taking a small portion of the auto industry’s vast ad budget allowed companies to persuade investors that online advertising would be huge and brought those companies online years before they had a business need to do so.
An ad supported web grows quickly and is open to those who can’t or won’t pay. But it has at least four downsides as a default business model.
First, while advertising without surveillance is possible—unverifiable advertising was the only type of advertising through most of the 20th century—it’s hard to imagine online advertising without surveillance. The primary benefit of online advertising is the ability to see who’s looking at an ad. Simply paying for online advertising requires surveillance, if only to eliminate clickfraud. And if Cegłowski’s theory is true, there’s no apparent escape from escalating surveillance to create more attractive business propositions.
Second, not only does advertising lead to surveillance through the “investor storytime” mechanism, it creates incentives to produce and share content that generates pageviews and mouse clicks, but little thoughtful engagement. Clickbait has become so prominent that even Upworthy, popularizer of spreadable media as a tool for social change, is asking advertisers to consider how much attention readers are paying to content, rather than how many pageviews it generates. Some new media empires are so attached to advertising metrics that they are giving writers days off from “traffic whoring” duty to allow them to produce content that has greater social and informational value. While many newspapers are shielding their reporters from statistics about whether their stories are being read, the increasing importance of digital news outlets to the public sphere suggests we may get less news that helps us engage as citizens and more news designed to get us to click the “next page” button.
Third, the advertising model tends to centralize the web. Advertisers are desperate to reach large audiences as the reach of any individual channel shrinks. A generation ago, you could reach a significant fraction the the American population by buying ad time on four television networks. Very few companies can offer that “Superbowl ad” reach today. Advertisers purchase ads scattered across hundreds of sites, buying demographic targeting at the lowest rates available. Companies like Facebook want get as much of that money as possible, which means chasing users and reach. Using cash from investors and ad sales, they can acquire smaller companies that are starting to build rival networks. (See Facebook’s acquisition of Instagram and, to a lesser extent, Whats App.) This centralization has dangers for online speech—it means decisions these platforms take to ban speech are as powerful as decisions made by governments, as Rebecca MacKinnon has eloquently documented.
Finally, even attempts to mitigate advertising’s downsides have consequences. To compensate us for our experience of continual surveillance, many websites promise personalization of content to match our interests and tastes. (By giving platforms information on our interests, we are, of course, generating more ad targeting information.)
This personalization means that two readers of The New York Times may seen a very different picture of the world, and that two users of Facebook certainly do, shaped both by our choice of friends and by Facebook’s algorithms. Research suggests that these personalized sites may lead us into echo chambers, filter bubbles, or other forms of ideological isolation that divide us into rival camps that cannot agree on anything, including a set of common facts on which we could build a debate. While many have written on this topic (and I wrote a book on it), few have shown the implications of overpersonalization as well as Gilad Lotan did in this recent analysis of media consumption in Israel and Palestine, where he describes the view participants in the current Gaza war have of the conflict as “personalized propaganda.”
It’s easier to rant about technology than it is to propose solutions. To Cegłowski’s credit, he closes his talk with a set of practical suggestions about limits we might put on the use of digital data by advertisers. He demands that we be given a right to review and delete data companies hold on us, proposes a time limit on how long data can be held and how it can be shared. Implementing these regulations, of course, would require finding a regulator with teeth, and it’s not clear that the Federal Trade Commission would be willing to enforce such constraints on companies that are becoming powerful actors in Washington.
More importantly, Cegłowski offers us a way forward through his own actions. Cegłowski wrote and maintains Pinboard.in, a simple and powerful bookmarking service with an unusual business model. Each user of the service pays a one-time fee, which rises a fraction of a cent with each new user. (When I signed up for Pinboard, it cost $5, and now costs a bit more than $10.) The cost has the benefit of keeping the service spam-free—Metafilter has seen some of the same benefits from their nominal membership fee—and has meant that the service has been profitable since it was launched. Users can upgrade to a $25-per-year version that archives every webpage you bookmark, creating a permanent, searchable archive of your journeys through the web. Cegłowski promises that he will never sell ads on the site and never sell data to third parties, reminding us, “If you’re not paying for your bookmarking, then someone else is, and their interests may not be aligned with yours.”
Pinboard launched in 2009, in part in reaction to changes at Del.icio.us, a beloved bookmarking site started by Joshua Schacter and sold to Yahoo, which slowly ran the site into the ground. Many of Pinboard’s policies can be read as Cegłowski’s attempts to protect himself—and anyone else—from having precious personal data held hostage when a company changes hands. But these principles are also an invitation to think about, and perhaps to create, a web that works very differently.
The web is celebrating a 25th anniversary, but that celebrates the invention of the HTTP protocol by Tim Berners Lee. In practical terms, the web as we know it is less than 20 years old. Many of the services we rely on, like Twitter, are less than 10 years old. Yet it’s often hard to imagine making deep, structural changes to the web. It’s easy to assume that aspects of the web’s architecture and business model are inevitable: We will inexorably move towards a web that is centralized, ad supported and heavily surveilled.
Part of the celebration of the Web’s 25th anniversary is The Web We Want, a campaign to open the dialog about how the Internet is structured and governed to voices from around the globe. I think it’s at least as important to consider how we want the web to make money, as these decisions have powerful unintended consequences.
One simple way forward is to charge for services and protect users’ privacy, as Cegłowski is doing with Pinboard. What would it cost to subscribe to an ad-free Facebook and receive a verifiable promise that your content and metadata wasn’t being resold, and would be deleted within a fixed window? Google now does this with its enterprise and educational email tools, promising paying users that their email is now exempt from the creepy content-based ad targeting that characterizes its free product. Would Google allow users to may a modest subscription fee and opt out of this obvious, heavy-handed surveillance?
Users will pay for services that they love. Reddit, the lively recommendation and discussion community, sells Reddit Gold subscriptions that give users special privileges and the ability to turn off ads. (Those ads, by the way, are a lot less intrusive than those on most social networks.) Reddit advertises Gold both by detailing benefits of membership and setting “daily goals” for gold subscriptions, telling readers, “We believe the more reddit can be user-supported, the freer we will be to make reddit the best it can be.” A culture has developed of giving a month’s Reddit gold membership to someone whose comments or content you’ve especially appreciated, rewarding both the individual and the community as a whole. It would be interesting to see if Facebook could support a premium model. I suspect many people use Facebook because they feel they have to, not because they love it, as they love Reddit.
There are numerous consequences, not all intended, not all good, to using fee for service as a default model on the web. Many users would abandon services that weren’t worth paying for—we’d see usage numbers for sites shrink as well as growing inexorably. Most sites would have much smaller userbases. This likely means we’d have a harder time finding our exes on Facebook, but might mean we’d see more competition, less centralization and more competitive innovation.
If we want to build a web that’s really global, we need to rethink online payment systems. Visa and Mastercard may never become pervasive in India and sub-Saharan Africa as mobile money already has a strong market share. Still, systems like M-Pesa suffer the same problems as credit cards and PayPal: high transaction costs. The model Ted Nelson dreamed of with Xanadu, where hyperlinks would ensure authors were cited and compensated for their work, required a micropayment system with low transaction costs.
We may be nearing such a system with Bitcoin and other cryptocurrencies that permit transactions with extremely low transaction costs. (In theory. In practice, Bitcoin transaction costs are currently still a significant fraction of a dollar.) Projects like Stellar are focused on mainstreaming cryptocurrency and ensuring these systems aren’t open only to the digerati. If Stellar takes off (a big if) and if transaction costs drop low enough (a very big if), we might see an Internet supported on micropayments of a fraction of a cent to compensate the operators of services or creators of content. This sort of architecture might provide a viable alternative to advertising for new businesses starting online. (Consider Jeremy Rubin’s Tidbit project, which quickly ran into legal trouble—viewers of a webpage would pay a micropayment through giving cycles of their computer’s CPU to mine Bitcoins as a way of paying the page’s owner.)
There is no single “right answer” to the question of how we pay for the tool that lets us share knowledge, opinions, ideas, and photos of cute cats. Whether we embrace micropayments, membership, crowdfunding, or any other model, there are bound to be unintended consequences.
But 20 years in to the ad-supported web, we can see that our current model is bad, broken, and corrosive. It’s time to start paying for privacy, to support services we love, and to abandon those that are free, but sell us—the users and our attention—as the product.