In this conversation with RevDem editor Ferenc Laczó, Ben Tarnoff – author of the new book Internet for the People. The Fight for Our Digital Future – discusses how the internet was created and how it has been privatized; how its current version fuels inequality and the rise of the political Right; why finding the right metaphors is crucial; and why the ongoing anti-monopoly push is not enough.
Ben Tarnoff is a tech worker, writer, and co-founder of Logic Magazine.
Internet for the People. The Fight for Our Digital Future is published by Verso.
Ferenc Laczó: You show in your new book Internet for the People that the current internet is a business controlled by big corporations and run for profit. How did that come to be the case? How were “the pipes” and how were “the stacks”, as you call them in the book, privatized over time?
Ben Tarnoff: That’s a set of very big and interesting questions with somewhat long and complicated answers. I should say the book is mostly trying to address precisely these questions, which is how did the internet – which came out of Pentagon research, was a publicly managed and funded project – pass into the hands of the private sector? How did it become the basis for so many wildly lucrative businesses? And what does that have to say, more broadly, about the neoliberal age, about the stage of capitalism we are in? These are, as you would imagine, big and somewhat unanswerable questions.
To briefly sketch its history, the internet was initially developed in the 1970s as a protocol, which in computing simply means a set of rules for how computers should communicate. It was invented as a protocol to enable different kinds of computer networks to interconnect.
There was a military pretext for funding this innovation – the Pentagon was interested in bringing computing power into the field. To do so, it had to find a way for smaller, less powerful computers that might be loaded into a jeep somewhere in Vietnam to communicate with the bigger, more powerful mainframes in Northern Virginia.
That’s the basis for the creation of the internet protocol. But it wasn’t used for that purpose; once it’s created, it’s primarily used to interconnect various fixed-line networks owned by the Pentagon.
Fast forwarding a bit, the internet evolves into a network of networks.
By the late 80s and early 90s, it was still under federal control – it was under the National Science Foundation, which is a US federal agency tasked with supporting basic research. At this point, it was used primarily by academic researchers.
Folks at university campuses, primarily in the United States, but also around the world, were using it to email one another — to collaborate, to exchange ideas, to talk about their research.
The crucial year for the first phase of the privatization of the internet is 1995, at which point the National Science Foundation decides it’s going to turn over the internet to the private sector as comprehensively as possible — which is to say, with no conditions, no compensation, no rules, and no enduring public or non-commercial foothold in this new network. We could talk a bit about the various ideological and economic aspects to this transition, but 1995 is the turning point where privatization begins.
Crucially, the privatization of the internet is a process, not an event. It’s not something that was completed in 1995, it was something initiated in 1995 which will go on for many years. What is at stake in the privatization of the internet is how to program the profit motive into every layer of a network that was built by research scientists and developed under public management. That’s not an easy task.
When I read your book, I had the impression, following the interpretative model developed by Quinn Slobodian and others, that the internet is a key example to understand what neoliberalism is. This is clearly a project that is started by the state and marketization happens through very specific regulations. Would you like to comment on how the story of the internet, in your view, fits or doesn’t fit the theories of neoliberalism more generally?
I think Quinn Slobodian makes the point that what is new in neoliberalism — the basis of the “neo,” if you like — is that the state is enlisted to insulate the market from the demands of a democratic polity. Certainly, that is the case with the privatization of the internet. Telecom lobbying plays an enormous role in determining the shape that the internet will take after 1995. There is relatively little broader public input into this process and, frankly, very little public understanding of what was at stake, because the internet was not particularly well known at the time.
More broadly, thinkers like Mariana Mazzucato have documented how the state, and in particular the United States military, takes on the costs of subsidizing and supporting the basic research that generates breakthrough innovations and brings those technologies to a point of maturity.
At this stage, the private sector takes over, commercializes them, and can make quite a bit of money off them.
The internet is a particularly extreme example, but the entire tech industry as we know it today has been built on this model. Of course, there are many other sectors that owe their business model to federally supported basic research — think of biotech and pharma. This is a common tendency within capitalism: capital finding ways to socialize the costs and privatize the benefits — and that does not necessarily require a neoliberal regime.
Would you be willing to comment on key challenges related to conceptualizing the internet? How does your book aim to change the conversation when it comes to the key terms we use to try and grasp what is going on?
Metaphors matter when we think about the internet. In many ways they’re not just a useful tool for thinking, but the primary lens through which we see something like the internet – we obviously can’t see the internet from a bird’s eye view. If you think about many of the other essential infrastructures on which our lives depend — power plants, the electric grid, sewage treatment plants, highway systems, train systems — you can visualize them. Maybe not comprehensively, but it’s fairly easy to fire up Google image search and get something that looks reasonable. If you plug the internet into Google image search, you just get nonsense — glowing planet shapes with lights buzzing around.
This is a common problem in computing, which is that the only real thing that is happening are very small surges of electricity somewhere deep in the machine that are encoded as zeros and ones. Everything built on top of that is an abstraction.
When it comes to the internet, what is the abstraction that would help us talk about it most accurately? What I came up with was that the internet is a language; I think that’s the best way to understand what the internet is.
It’s a language that lets different computer networks speak to one another across their distinct dialects. You can think of the internet as a kind of digital Esperanto that makes possible a form of universal communication.
Now this language is implemented as a protocol which, as we discussed, is a set of rules for how computers can communicate, and it’s implemented as software that runs on your computer. This language is spoken across various structures, which are things you can actually touch — fiber optic cables, routers, various machines that stitch the networks of the internet together and make this language possible.
That’s where I landed in terms of how to think about the internet. Now there are other forms of life within the online realm, such as the so-called platform, that I also attempted to provide my own metaphors for. I realize this is kind of a fool’s errand — I don’t expect that this book will cause people to stop using the term platform, for instance, even if I consider it misleading.
I risk being a bit of a pedant here, but my hope is that I can show that there are real stakes to the words, to the metaphors, to the abstractions that we use. They are not a small matter and have a lot to do with how particular firms and particular actors have acquired so much power. In part, their power has to do with how they have led us to think about the internet in a particular way.
A moot question is what kind of business the internet is. You argue that the online mall “facilitates interactions, it writes the rules for those interactions, and it benefits from having more people interact. But in the online mall, these interactions are recorded and interpreted. This is what distinguishes online malls from their precursors. They are above all designed for making, and making use of, data. Data is their organizing principle and essential ingredient.” Let us perhaps approach this issue in a rather general way first: how does the internet make money from what people do online? And how come data has become so valuable in financial terms?
The question of how to make money from what people do online was the central question behind the dot-com boom, which began in 1995 with the IPO of Netscape – Netscape was the creator of the Netscape Navigator, the most popular graphical web browser of the time – and haunted that era until the bust in 2000 and 2001. They tried to figure out how you build a business model on the web.The breakthroughs that would unlock the profit potential of the internet mostly didn’t arrive until after the bust. It’s really out of the rubble of the dot-com bust that a set of new complex computational systems that we commonly call platforms emerge, and begin to make a lot of money.
I use the term “online malls” to think about how these systems function. I draw on the work of Jathan Sadowski who points out that real malls are in the rental business; they are renting spaces in the mall to merchants. Online malls are also in the rental business, but they extract two kinds of rents: they extract monetary rents – if you take an Uber ride, for example, there is a fee that goes to Uber. But, more importantly, they extract data rents. All the interactions that transpire within the walls of the online mall are occasions for manufacturing data. And these interactions can be commercial – you could be buying something, just as you would in a real mall – or they can be social. If you’re a suburban teenager, you probably went to malls to hang out with your friends. Similarly, online malls can host a wide variety of interactions but, crucially, all these interactions can create data, can be occasions for the manufacture of data, and this data, in turn, can be monetized in a variety of ways. The way in which we are most familiar with this in the mainstream conversation is so-called targeted advertising. This is the basis of Google and Facebook’s business model which in practice means that data is manufactured about a user and that data is then applied to sell advertising that claims to be able to target particular demographics.
That’s not the only way that data is monetized. Data can be used to train machine learning models that generate completely new types of services and products – like chatbots, for instance. They can be used to optimize labor processes to manage workers – think about all the data that a company like Uber generates and that is being fed into its algorithmic management systems that are designed to extract more and more value from each worker. Suffice to say there are a variety of data monetization strategies.
The crucial thing that enables these firms to unlock the profit potential of the internet is data. In my view, it enables privatization to be pushed up “the stack” from “the pipes”, from the physical infrastructure of the internet up into the application layer where we experience the internet. What we get is the mass manufacture and mass monetization of data.
You seem to insist in the book that the promise of sending personalized advertisements via these apps and major sites may be an illusionary one. There is in fact little evidence that you can really sell products in great quantities via such a method. What has perhaps happened is that companies managed to convince other companies that it’s worth investing in these types of advertisements rather than really finding a new marketing technique that would have changed people’s behavior on the ground. How do you view that gap and what it implies?
I rely on the work of Tim Hwang here who has published a wonderful book called Subprime Attention Crisis. Tim makes the point — and he draws on a lot of research that substantiates this view — that online attention, as it is packaged and sold by firms like Google, Facebook, and others within this extraordinarily baroque advertising ecosystem, is wildly overvalued. He draws the analogy to the subprime mortgage crisis where traders thought they were buying something of value and the asset that was supposed to be behind the derivative they were purchasing turned out to be worthless.
There is indeed a lot of data that suggests that online advertising is not particularly effective and does not really deliver on its promises; people are not seeing the ads because they’re not placed in the right place on the site, or they’re using ad-blockers, or they’re not going to the so-called targeted destinations that they’re supposed to go to.
Obviously, the monopoly power of firms like Google and Facebook and the opacity of the online ad market makes it even harder to understand exactly what’s going on. But it is an important corrective that Tim provides us, and one that I wanted to emphasize in my book. These systems are quite complex. They have all sorts of destructive downstream social consequences that I discuss in my book, but it’s important not to impute a kind of metaphysical power to them.
This is where I think certain critics go astray. Shoshana Zuboff, for example, often implies that Mark Zuckerberg has built some kind of mind-control machine, and that’s simply not the case. Zuckerberg, of course, wants his customers, who are advertisers, to believe that, but it’s not true — we’re not simply rats in a kind of Skinner Box where we are responding to various inputs that are being delivered in very precise ways. If you work in software, you know that much of the software simply does not work.
But it’s a balance. I think we do need to acknowledge that these are complex systems, they do require close attention, and they do cause harm. But the people who create them also have made certain promises — and these promises are integral to their business model — which are not always kept. We need to be able to keep all those thoughts in our mind simultaneously.
You call the contemporary internet an inequality machine and argue that “what the shipping container has done to much blue-collar work, the internet has done to much white-collar work.” You also underline that there is a symbiosis and a rather delicate balance between social media and rightist forces in politics. Could you elaborate on those points? How is the dominance of a privatized internet connected to growing inequality and the rise of political Right?
We need to try to find some kind of common characteristic that ties these various computational systems commonly called platforms together because they are quite complex and they are quite distinct from one another. Facebook is quite different in its technical composition from Uber or Amazon, and so on.
If we go up a level of abstraction and try to think about what it is that all these systems share in their social effects – even while keeping in mind that those precise effects vary and work through different entanglements of social, financial, legal forces — we can see that they all reinforce and amplify inequality in one way or the other. They are all inequality machines.
In the case of a company like Uber, how it amplifies inequality is by facilitating the exploitation and indeed the immiseration of app-based workers. Through forms of algorithmic management workers are compelled to work harder while they are often making close-to-poverty wages. An occupation, the taxi business, that formerly provided a lower middle-class livelihood has in many cities been pretty close to decimated by Uber and Lyft.
That case belongs to a broader history of using networks to move work around and to manage workers at a distance. The shipping container comparison that I make is connected to that discussion. I think it’s always important to historicize and Uber certainly doesn’t come out of nowhere. Ever since we had local area networks (LANs) and wide area networks (WANs), companies have been using them to move workers around, to move back-office functions off-site and then offshore.
One of the virtues of networking is not just being able to move work away to places where workers have lower wages, which is what the shipping container enabled manufacturing firms to do, however. Networking also enables the employer to retain control over those dispersed workers through software, which is what we now call algorithmic management.
Moving to social media, let’s take a company like Facebook: how does Facebook amplify inequality? It does so somewhat differently, through its effect on our informational environment. My argument is that Facebook’s contribution to inequality lies in providing a toolbox, a valuable resource, a megaphone to the political forces who are committed to promoting inequality — which are those of the Right.
I walked through various research that shows that the Right has been extraordinarily successful, particularly on Facebook, in spreading its propaganda and making use of some of the algorithmic imperatives that are wired into Facebook to maximize user engagement, and which tend to favor provocative and sensationalistic content that the Right produces.
They have also conducted effective pressure campaigns on the company through Republican Congressmen. There’s a lot of evidence to substantiate the view that Facebook has become a machine for the proliferation of right-wing propaganda.
Before we talk about potential change and potential progress in all these areas, I wanted us to maybe open the perspective a bit and address the global dimension. Your book focuses on the U.S. where you observe a lightly regulated oligopoly which results in a huge concentration of power. You also mention that a set of linked internets appear to have emerged by now that are centered on the main digital hegemons, the U.S., the EU, and China. Would you be willing to discuss how similar or different these linked internets are? How do you view the place and role of the U.S. in the global system today?
The United States retains a great degree of power in the global internet. Of course, the internet is far less American than it was in the 90s. Most internet users are not Americans anymore, so that is an enormous change. At the same time, most of the big tech firms remain American. There are still some governance functions too that remain in American hands — ICANN, which is a nonprofit tasked with overseeing the domain name system, which is essentially the address book of the internet, is based in Los Angeles, for example. The U.S. still exercises a great degree of influence over the contemporary internet, but, as you mentioned, there are new online hegemons that have emerged.
For instance, the Europeans, whose influence is primarily exercised not through industry — because they have, with few exceptions, not generated any major tech firms — but through regulation. GDPR, the major European data regulation, has had an effect far beyond Europe and has been an inspiration to certain efforts in the United States. And there are now more regulations appearing in Europe, such as the Digital Services Act or the Digital Markets Act. It’s a separate question how effective these regulations are, but it’s certainly something that the Europeans are keen to distinguish themselves on.
The Chinese, on the other hand, have pursued an industrial policy model of building their own internet industry through the erection of the Great Firewall which tightly filters traffic coming in and out of China.
There is obviously an authoritarian-securitarian aspect to the Chinese approach which is often emphasized in the Western media. But there is also an economic aspect: it’s essentially a kind of protective tariff that has enabled a native internet industry to emerge and become quite varied and successful.
Interestingly, more recently, there has been a government crackdown on tech in China for various reasons: for violations of new Chinese data protection laws, because of the perception that they are contributing to inequality, or because they’re not generating the kind of dynamism that Xi Jinping thinks is necessary to propel the Chinese economy forward. Xi has a term, “common prosperity,” which has been discussed widely in the Western press, and seems to consist of trying to find a more equitable form of growth, an emphasis which poses challenges for the tech sector. That is an interesting governance direction t and one that, again, distinguishes the Chinese from both the Europeans and the American models.
I would say it is a somewhat mixed picture globally. There are also what I would call sub-hegemons. For instance, India has an interesting tech scene, and there are some interesting things happening in Latin America. But I think it’s fair to say that the U.S., the Europeans, and the Chinese are the ones who are mostly setting the pace in terms of what the contemporary internet looks like.
You mention towards the end of the book that the everything machine that is the internet has now started to become an everywhere machine too. As the internet has become more fluid, ubiquitous, and diffuse in recent years, the digital surveillance of the physical world has greatly expanded. There is also greater recognition among members of the political elite in the U.S. that the current internet is broken, and reform measures would be necessary. How much worse have the dangers posed by the internet to democracy gotten? How would you assess the current “techlash” and the ongoing reform attempts aimed at regulating the internet and introducing anti-monopoly rules?
In the United States, an important consequence of the “techlash” that has emerged in 2017-18 is the rise of a robust anti-monopoly current within the Democratic Party and even among some Republicans. This is coming out of the work of younger scholars like Lina Khan, who is now leading the Federal Trade Commission under President Biden, as well as a handful of others who have been pushing for much more robust anti-trust enforcement for things like breaking up big firms, banning mergers and acquisitions, introducing various forms of regulation that would promote what they call fair competition. This is often focused on tech, because tech is the profit center of the American economy, but it’s not limited to tech.
Biden, in addition to appointing Khan and other anti-monopoly advocates to positions in his administration, issued an executive order that directs various executive agencies to pursue pro-competition initiatives. There are also two bills in the Senate that would take aim at so-called self-preferencing by tech firms – things like if you Google something, how Google will show you its products in the search results. Interestingly, this is the same thing that Microsoft got hit for in its anti-trust saga some twenty years ago – for bundling its terrible web browser with Windows.
Whether the new anti-monopoly initiatives will succeed is an open question. The level of dysfunction in the American political system right now is severe and I think the prospects of getting anything done are slim. We could have one of those Senate bills get through in some form, and there will certainly be continued efforts by figures like Khan at the FTC to make anti-trust enforcement and anti-monopoly in general more robust, but there are various powerful forces arrayed against them, as you can imagine.
In terms of my hopes and my own position, I have sympathy and critical support for the anti-monopoly position. I think we need to curb the power of these big firms and shrink their footprint. And there is a role, certainly, for more robust regulation in contributing to that goal. But my view is that monopoly power is not the root of the problem.
In other words, the reason that the internet is producing these destructive social effects – the reason that the internet is broken, as I argue in my book – is not because markets are excessively consolidated.
The problem is the market itself.
My view is that we don’t exactly need to make markets more competitive – I am not convinced that doing so will automatically generate better social outcomes – but rather to make markets matter less, to shrink the space of the market, to diminish the power of the profit motive. We need to create spaces and structures that encode real democratic control, where people can participate in the decisions that most affect them and build the internet they want.
You state in the book that de-privatization is the right target and that this is obviously a political rather than just a technological question. You explain that one of the goals should be to have social media that are decentralized, self-governing, well cared for, anchored in public libraries, and enriched by public media. Why do you see de-privatization as such a crucial aim? And what might a social and political movement to democratize the internet look like?
De-privatization in my view is a political project that is committed to creating an internet where people, are not profit, rule. What would that mean specifically? First of all, it doesn’t mean turning the clock back. The point is not to return to the internet of 1990 or 1986, or anything like that.
The privatization of the internet was a creative process – it took a relatively small public network and turned it into a powerhouse of global capitalism. De-privatization must be no less creative.
What might that look like in practice? Our strategies for de-privatization are going to depend on the layer of the internet that we’re talking about. The internet has different layers. If we’re talking about the bottom layer, the physical infrastructure of the internet, what I call the pipes in my book, we have a pretty good model for what it might look like to de-privatize it. In the United States, we have more than 900 so-called community networks which are publicly and cooperatively owned broadband networks – they could be owned by your local municipality or by the users themselves in the case of cooperative networks. Research has shown that these community networks provide better service at lower cost than their corporate counterparts, and the reasons for that are fairly straightforward: they’re not spending billions of dollars on stock buybacks and dividends, and they’re not spending tens of millions of dollars on executive pay packages either.
These networks exist within capitalist markets and sell a commodity, of course. They have to make money in order to survive, but they are nonetheless able to deviate, in some cases slightly, in some cases significantly, from the imperative to accumulate, from the principle of profit maximization.
They are able to value social goals like universal connectivity and, crucially, they are able to encode democratic practices into their everyday operation.
Cooperatively owned networks are often what are called Rural Electric Cooperatives in the United States. Many of them date back to the New Deal era, are member-owned, and enable people to get electricity. Many of these cooperatives have now moved into the provision of broadband. In order to retain their federal tax exemption, Rural Electric Cooperatives must hold regular elections for their board.
That’s just one example of how, when we develop publicly and cooperatively owned alternatives, we can create spaces for democratic decision making so that the users of the infrastructure get to participate in the deliberations about how it will be deployed and used. In other words, executives and investors no longer get to monopolize these decisions. These decisions could be located where they belong, which is with the people themselves.
The transcript has been edited for length and clarity.
In cooperation with Lorena Drakula.