IN THE FOLLOWING conversation, I ask Nick Srnicek, the author of Platform Capitalism, about the nature — and dangers — of digital technology platform companies like Google, Facebook, Amazon, and Walmart.
TOBIAS HABERKORN: Your recent book, Platform Capitalism, ends with the suggestion that we’d have many good reasons to nationalize big platforms like Google or Facebook. Let’s start from the beginning: what do you mean by platform?
NICK SRNICEK: The platform is actually a quite old business model, but it has become much more pervasive with the rise of digital technology. Effectively, a platform is the intermediary between two or more different groups. We might think here of early market squares, but the platform as a model has really taken off with digital technologies in the past 10 years. Facebook, for example, is an intermediary between advertisers on the one hand, and users, software developers, and companies who create pages and chatbots on the other. Facebook brings together all these different groups and gets its value from them, and that’s quite novel compared to more traditional businesses. And these platforms are becoming central to contemporary capitalism: they are increasingly the most profitable, the wealthiest, and the most powerful companies in the world.
What would older examples be?
The classic example would be something like a shopping mall. A shopping mall brings together a number of individual shops on the one hand and a large group of customers on the other. It functions as a physical platform to bring together these two groups. Physical platforms have existed for a while but with digital platforms, the business model is becoming much more widespread for one crucial reason. As an intermediary, the platform is able to collect and extract all the data about the behavior of the individual groups that interact on it. All of these data resources can be siphoned off very easily, and that helps explain why we are starting to see platforms pop up not just in the technology sector, but everywhere. Monsanto and John Deere in agriculture, Siemens and GE in manufacturing, Rolls-Royce in the jet engine business, and so on. In various ways, these firms are all adopting the platform as a business model.
Can you explain in what way Siemens or GE are becoming platforms? It is pretty intuitive to conceive of Facebook as a platform since contents and profiles are almost literally “on” Facebook. But how can an industrial firm like Siemens function in a comparable way?
The idea for a platform behind Siemens or General Electric is having something like an industrial internet. In Germany, they call this Industry 4.0: the idea that you can hook up all the individual components of a factory, the machinery, the products that are going through it, to a local intranet. The promise of this approach is to make factories more automatable (reducing labor costs), more efficient and productive, and less prone to breaking down (since sensors can constantly monitor wear and tear). GE and Siemens are currently inventing billions in a race to build the cloud computing and the software analytics necessary to do this, and the tools that allow developers to create new software for these things. One way to think about it is that Siemens and GE are competing to build an App Store for factories. Instead of every individual factory having to build up software from the ground up, they simply adopt Siemens’s or GE’s platform and then purchase the data analytics apps they might need. For Siemens and GE, this means every other manufacturing company becomes dependent on them, both economically and politically. The result is a handful of monopolies potentially dominating much of global manufacturing.
The car industry for example, which accounts for roughly 800,000 jobs in Germany, has had distributed production chains for a long time. Firms like Volkswagen or Daimler use a network of suppliers and build modules into models of different car brands. Aren’t they platforms already?
Not really. The traditional business model of the automobile industry is much more linear. You are taking supplies from a supplier, bring it to your factory, build it in the cars, and then sell the cars to the customers. There are different sizes of that market, but the key thing is that they are not interdependent in the way that they are for a platform. A brief example: Uber’s two sides are massively interdependent. You have to have enough drivers to get enough riders and enough riders to get enough drivers, otherwise the platform won’t work. These two things cycle around each other, and this is what is really unique to the platform. Moreover, platforms give rise to network effects. At the most basic level, this means that the more people are involved in a platform, the more valuable the platform becomes to everybody else. You may not like Mark Zuckerberg, you may not like Facebook’s privacy scandals, you may not enjoy their surveillance, but, if you’re going to join a social network site, it’s probably going to be Facebook simply because all your friends and family are already there. As more people use the platform, it becomes more valuable to everyone. This is why the platforms tend to create monopolies. With network effects, it becomes very difficult for competitors to get any power, any influence, whatsoever.
The problematic aspect of Facebook would then be that they become the monopolist for engineering the public sphere, which transforms it into a very powerful platform vulnerable to manipulation. We’ve seen the controversy around interference with the American (US) elections, Facebook ads targeted to certain audiences and so forth. Now that elections are at play, politicians seem to have become aware of how powerful platforms have become for public opinion. What do you make of the recent attempts in regulating and controlling Facebooks on the political level?
I think there have been two broad responses so far. First, Facebook insisting that they can and will self-regulate. Their attempts to label fake news, or their promises that AI will solve their problems, are examples of this trend. The problem with this approach is that their entire business model is reliant on advertisement. And advertisement in the internet economy is based upon surveillance. So Facebook has to monitor people, it has to collect data, and the more private data they can get, the better it is for their bottom line. Despite the best intentions of individual Facebook employees, the structural necessity is that they have to collect more and more data. And I don’t think self-regulation is going to be able to ever overcome that hurdle.
On the other hand, you increasingly have regulators (particularly in Europe) pressing for more stringent controls over Facebook. While there are more radical noises about breaking Facebook up, or enforcing data sharing policies, in practice these regulations have been relatively minimal so far. Perhaps most problematically, a lot of the discussion assumes that Facebook — and other private platforms — should be making decisions about censorship and content and access. As you mentioned, platforms like Facebook are increasingly taking on public utility functions, and as such, they need to be governed by the public, not by a private company and its executives. This means entirely changing the governance structure of these platforms.
An interesting aspect of Mark Zuckerberg’s recent statements was that he said he’s expecting users to spend less time on Facebook and to see certain engagement measures go down. He says he wants to make time on Facebook more socially valuable, “time well spent.” Do you think this is empty rhetoric or actually a reckoning that he has pushed the business model too far?
I think that partially this is a matter of data quality versus quantity. To see what this means, we can point to research on the data that people give to Google and to Facebook. When people present themselves in social media, they do so in a mediated way, presenting idealized image of themselves — which means that Facebook likewise tends to get more mediated data on people. That has impacts on how well Facebook knows these people, how well they know them for advertisers. Google, by contrast, is a relatively more anonymous platform. You may search, for instance, about medical conditions that you have or search for things that you would never talk about on public social media. The result is that Google arguably has better data on people — its quality is higher.
To go back to Facebook’s decisions around “time well spent,” I think partially what is going on here is that they recognize if fake news and low-quality materials are allowed to continue spreading wildly, that it’ll have a detrimental impact on their data quality. They’ll get artificial antagonisms, they’ll get people leaving, and they’ll get people who just stop posting because they don’t find it useful anymore. All of that is actively harmful to Facebook’s core business. So I think what is happening is actually a strategic play to give up some quantity of data in return for better quality data.
I think we all understand the tendency of Facebook toward a monopoly of information and political news is problematic. But if we take, for example, a platform like Amazon, which is enabling an enhancement of user service, enabling as a platform to condense the entire network of independent bookstores into one platform where you can then find books for reasonably good prices, et cetera — this is how they started — then you have to admit that from the point of view of the individual user, they have arguably had a positive impact in many ways. Do you think that once they have acquired a dominant, almost monopolistic position in many aspects of ecommerce, that they would necessarily abuse their power position?
You are right that, on the consumer level, these things have often brought significant benefits. Uber is much more convenient than traditional taxis; if you use Gmail, you benefit from having you mail information automatically imported into your calendar; and so forth. There can be a lot of benefit in the consolidation of these things. These benefits, in turn, feed the monopolistic aspect of these companies — but they also form the basis for thinking that these companies should be, at a minimum, regulated as public utilities, if not brought into common ownership (since breaking them up would be counterproductive).
As for whether or not these companies will use their power for more nefarious reasons, I think this is already happening. Amazon’s hunt for their second headquarters is a great example. They had several cities compete against each other to provide Amazon with the best tax breaks, the best incentives to place their headquarters there. A race to the bottom in order to get Amazon to come to town — and this is only possible by virtue of the power that Amazon has at the moment. And that’s just one example.
Maybe one can say this is the business model they found somewhere around 2008. There was a point where they seriously understood that they didn’t have a business model and then they went into that, right? It is maybe difficult for these types of firms, Facebook particularly, to distinguish their initial drive from how they evolved as a business and then tried to become profitable.
Right, there’re only a few seemingly viable business models for the internet (at least under capitalist conditions), and many of them center around extracting personal data and using it to sell ads. This has become the dominant approach — surveillance capitalism, to use Shoshana Zuboff’s term — for a lot of firms.
Yet at the same time, most of these major internet-based companies are actually unprofitable. LinkedIn has never been profitable; WhatsApp was never profitable; Snapchat has never been profitable; Uber has never been profitable — the list goes on and on. All of these companies have made losses every single year of their existence. So I think there’s increasing desperation among some of these companies to find a profitable approach — along with all the impacts on privacy and security and manipulation of society that the rabid drive for profits entails. Taking a longer view, we might wonder whether the internet is even suited to capitalism. Perhaps, to really see the internet’s potential flourish, we need to escape (surveillance) capitalism.
In Platform Capitalism, you express a lot of skepticism toward the viability of advertising platforms. You say that a firm like Amazon, which is grounded in ecommerce, in the commerce of concrete goods, has a much more viable business model than Google and Facebook. Why?
I think there are a couple of problems with the advertising model. First of all, it’s heavily subject to economic fluctuations. If the economy goes into a recession, one of the first things that businesses cut is their spending on advertising. There is a well-known correlation between economic cycles and advertising spending. This means that Google and Facebook are highly susceptible to economic fluctuations. The other problem for them is that the growth we have seen in digital advertising is still continuing, but that growth rate is declining every year. As more and more traditional advertising moves to the online world, there ends up being less old advertising money to pull over. So advertising platforms are still growing revenues, but the medium-term poses problems for that model. Another aspect is that Facebook and Google have basically carved up the online advertising sphere for themselves already. They control 50 or 60 percent of the lucrative Western online advertisement market as it is (estimates vary from country to country, but they are typically 50 percent and above). But this means that the easy growth they had taking revenues from other firms is basically gone now. Every year now, 80 to 90 percent of the new spending in digital advertising is going to Facebook and Google. The only place they can take growth from now is to cannibalize each other. This all means that there are significant limits to how much revenue they can bring in. And if advertising is the fundamental resource generator for the rest of the business — because again, the rest of these businesses are unprofitable — it places limits on what you can do.
By contrast, a firm like Amazon has much more room to grow profits — whether it be through cloud computing or through their more traditional ecommerce business. The latter, for instance, is still a relatively small part of the American retail sector — just under 10 percent according to the official figures. So there is a lot of room to grow for Amazon, and relatively little for Facebook and Google (at least for their core business).
Even if we take aside the question of the limits of growth — is advertising actually a business model of the future? Advertising functions by creating desires and giving people the choice or suggesting that they can choose the best products. What tech firms are now pushing for are personal assistants at every instant in the chain of consummation, a type of service where the wish is fulfilled at the very moment it is formed, so that there is no need for advertising anymore.
Yes, I think this is probably their ideal sort of thing, then you’ve got the little Amazon buttons you press and you get the product — that’s the sort of idealized end goal: where ultimately prediction takes the place of advertising, and the products show up on your door even before you recognize you need them.
This would in fact eliminate the core principle of market capitalism from the point of view of the consumer, which is: You have a choice between several options, you can choose the one that has better qualities in your eyes, and a lower price. In a system where we don’t have these instances of choice anymore, do we still have competition and at what level?
I think you have competition, but it’s at the level of algorithms fighting among each other rather than firms competing to grab individual choice through advertising. It’s an interesting hypothesis (and something that’s been explored in much more depth by others).
But you wouldn’t have competition between Alexa and Google Home, would you? You wouldn’t have both these assistants and ask them, “Who can get me spaghetti for a better price,” right?
It’s possible to have competition between Alexa and Google Home, where consumers multihome on these platforms — but it adds more complications and costs for users, so it doesn’t seem likely to become widespread. But there would be a lot of competition to be the single product that is presented as the fulfilment of that desire. Already, Alexa has a tendency to present Amazon-branded goods when people ask to shop for basic items, and you can imagine the sorts of monopolistic manipulations of consumerism that could occur here.
At the end of your book you speak about platform wars, the prospect that the big tech firms will start competing against each other. But doesn’t it rather look like they have happily carved up the internet and are getting along rather well without interfering too much? Google for search and ads in the open internet, Facebook for social interactions, Amazon for ecommerce, and Apple for the hardware — there are, of course, others, but let’s stay with the Big Four here.
Each of these Big Four started historically in a particular area. Whether it be search engines, social media, ecommerce, or hardware and operating systems. But as the data imperatives have compelled them to expand and find new areas to collect data, they are starting to impinge upon each other. I think ecommerce is a really good example. Facebook is trying to develop ecommerce capacities on its website, as a platform. Google is also trying to do ecommerce capacities on their platform. And these all are directly countered to Amazon’s platform. So we’re starting to see some competition there. Google has begun collaborating with Walmart, which is one of the most direct competitors to Amazon (and Walmart has purchased Flipkart, Amazon’s biggest competitor in India). So there are these partnerships and alliances being built up for more or less indirect competition at the moment, but competition nonetheless (and one that’s marshalling billions of dollars behind it).
In the market for smartphones, Amazon tried to get into the action with the Fire Phone in 2015 — a failure in the end, but an attempt to directly compete with Apple. Facebook is trying to carve away Google’s hold on the open web by redirecting users to their closed app (e.g., webpages that open in Facebook’s app, rather than Google’s browser). We can see similar lines of competition being drawn in all sorts of places: driverless cars, messaging platforms, voice assistants, internet of things, virtual reality, and so on.
I think we’re also starting to see the inklings of competition in AI. All of the major platforms are investing heavily in AI and recognize that AI is the future for their businesses (and likely much of the economy). Moreover, to do AI today requires things that only a handful of companies have access to, and the big tech firms recognize that they can build a monopoly position within the AI sphere. If they can achieve that, they’ll have the power to basically rent out AI to any other company and charge an exorbitant amount of money. So we’re seeing significant outright competition in the key factors that go into contemporary AI: data and skyrocketing salaries for the skilled workers.
This leads us to a scenario that leaves little room for resistance, because opting out, even collectively with a group of friends or a small organization, seems pointless. You are missing the network effects and you are losing traction. Every act of resistance seems to be too small because of the sheer scales of these firms. What do we do?
It’s incredibly difficult, and a lot of the resistance options are heavily individualized. We can personally lock down our social media profiles, we can turn toward privacy-saving alternatives, we can deploy algorithms that fool other algorithms, we can fool facial-recognition surveillance with specialized makeup and outfit, and so on. But these all tend to rely on individuals making a choice, rather than any approach to trying to systematically undermine the big platforms. It is incredibly difficult to think about what can be done, in part because the size and scale of these companies is very difficult to challenge from the ground up. There are some attempts though — for example, the platform cooperative idea of “we’ll produce an alternative, and we’ll get users to migrate to a nice, humane version of Uber,” for instance …
Are there any functioning examples of that?
To my knowledge, one of the more interesting is probably Ride Austin in Austin, Texas. It’s not quite a platform cooperative, but it is a nonprofit platform that aims to pay and treat its workers well. It has had some success, but the story of that success is what makes the case particularly interesting. What happened was that Uber and Lyft were banned from the city at one point, thanks to a failure to adhere to local regulations. With these platforms, and their attacks on competitors, gone, new alternatives came up, including Ride Austin. Now they had some success as a result, but eventually Uber and Lyft were allowed to operate in the city again, and the ridership numbers for Ride Austin dropped by around 70 percent. So the lesson here is that to compete against the major platforms, they may have to be banned in some cases. Simply put, companies like Uber are willing to sabotage competitors (e.g., Uber has been accused by one competitor of booking rides with them and cancelling, thereby clogging up their taxi service) and undercut prices (through access to deep venture capital pockets and tax avoidance measures), for it to ever be a level playing field between a platform cooperative and these platform monopolies.
While this has tended to happen at municipal levels (Austin is one example, while London is currently threatening to ban Uber as well), this can also take place at state levels. Ethiopia is a fascinating example here, where they’ve decided to ban foreign ride-sharing platforms as part of their economic nationalism, and as a result there’s been a number of domestic alternatives that have risen up, tailored to local conditions (e.g., sparse smartphone penetration, slow internet, and a lack of mobile payment options). It does show what’s possible when governments remember they do have power over these platforms.
You could see it probably in very restricted industries, because Uber is a very different platform from Google and Facebook. If London would decide that Uber should be heavily regulated or even banned, there would be room for a publicly run taxi app. The smaller the field, the larger the leeway.
I entirely agree, and I think when we talk about what can be done it’s important to distinguish between these single sector platforms versus something which is spanning a number of different markets like Google, Amazon, or Facebook. The problem with those multisector platforms is much more difficult. At a minimum, we can think of public utility regulation to try and channel them toward benefiting the common good. But we can also think about some sort of democratic public ownership as an ideal solution. The problem is that as soon as you say that, you run into all sorts of technical hurdles around what that would entail. It’s not entirely clear what it would mean to have Google and Facebook nationalized, technically, economically, and politically. So I think we’re only at the beginning stages of starting to parse out how to take control and ownership of these means of production.
You are comparing the situation to early 20th-century collectivization of basic infrastructure such as water and electricity. Can you maybe talk about the limits of this analogy? The idea seems so simple: digital platforms are providing infrastructure that is so useful for everyone, and a part of the daily life of everyone, that we need to publicly control it. Therefore, we should nationalize it such as we have nationalized water and electricity supplies at certain points — even if we have reprivatized them since.
I think the justification behind bringing them into collective ownership and control is largely the same: they are natural monopolies providing a public good and as a result need to be taken into democratic control. But in terms of the mechanics of actually enacting that, I think there are a few key differences. The first is that earlier nationalizations occurred in single markets — oil, railways, water, and so on. But the monopolists of today are spanning several industries — Amazon alone spans ecommerce, manufacturing, logistics, cloud computing, grocery stores, computer hardware, and so on. The simple fact of the matter is that today’s monopolies span a much broader cross-section of the economy than any early monopolies did, and that poses challenges for thinking about bringing them into collective ownership.
The other salient difference is scale. Back in the day, monopolies were much more nationally and territorially localized, and it was clear which jurisdiction would have regulating and nationalizing power. Today, the platform monopolies are more global in many ways — and subject to overlapping jurisdictions at times. Data on EU citizens, for example, may be stored in the United States, but then subject to EU and American laws (a tension that is being played out in international negotiations around data protection and privacy). And platforms like Google don’t have to be located physically within a country in order to provide services to that country — is there any sense in which nationalization can make sense for a country like that? Which is all to say that while “collectivize the platforms!” makes a nice demand, there’s still work that needs to be done to figure out what this means in practice.
Which public institutions could take that on these questions? These firms are global, so we would need a transnational political unit that would take on these firms.
A transnational body would be ideal, but again there’s no existing institutional appetite for that, nor is there any social basis for pushing for it. As a result, at the moment these are mostly free-floating ideas, though that’s still important in terms of expanding the sense of what’s possible.
There are some existing tendencies toward nationalization of at least parts of the digital infrastructure, though these are driven more by national security concerns instead of any shift toward democratic control and ownership. A good example here is that the Trump administration was reportedly briefly thinking about creating a nationally owned and operated 5G network in the United States. It goes against their core economic beliefs, but the argument was being pushed by a group of people in the military who tried to justify it on the basis that (a) you need a 5G network to maintain technological sophistication, and (b) it also needs to be nationalized so that you can make sure China doesn’t get access to it. So geopolitical competition is driving the idea that you have to nationalize, and I think we’re likely to see much more of that — especially as autocratic leaders seem to be on the rise everywhere right now.
If you are an activist who wants to push for a less capitalist internet, an internet that would be oriented toward public interest, et cetera, a traditional way would be to act on the ideological level by putting pressure on the European and national parliaments, raising consciousness among users, et cetera. Is this still an effective mode of action? Or are the challenges technical rather than ideological, and is it therefore much more important to learn to code rather than to engage in discourse and procedural politics?
You need to do both (as has always been the case). I think that one challenge still is raising awareness. It goes back to the consumer benefits that we see from these platforms and the simple fact that most people (myself included) get a lot of useful benefits from having them around. But at the same time, we can run these companies in better ways and we can orient them toward incentives that aren’t simply “gather as much personal data as possible in order to sell targeted advertisements.” We can do better, which is the basic premise of socialism.
So raising awareness can have an important role in getting people to recognize that. Users don’t always realize the ways in which data may manipulate their job chances, their chances at getting credit, their chances of renting a home. They don’t see these impacts, because they’re indirect. Most users may also not think about the long-term impacts, when these platforms have consolidated their power even more. If states are already struggling to rein them in, what do we do in 10 years’ time? At that point, it becomes much more difficult to overturn their power. All of this is good reason to try and raise consciousness now — while at the same time thinking about the technicalities involved in resistance and in building alternatives.
It maybe depends on cultural factors. I think that in terms of privacy and control, the German public for example is much more concerned than the French or the British public. When we speak about nationalizing and the carving up of the internet into different political spheres, examples like Russia or even more China come to mind. From the beginning, these countries have been more restrictive on social media, they have built their own, nationally owned platforms. Is that something the European Union could take an example from?
There’re definitely national differences: Germany has a strong history of privacy after its experience of the Stasi, and China has maintained some distance from global capitalism and subscribed to a more economically nationalistic vision.
China is a good example of what I mentioned earlier: that to build up alternatives to the platform monopolies, it seems necessary to block them out. China is the best exemplar of that approach, and they’ve been wildly successful at it — creating not only copies of the major platforms, but in many cases improving upon them.
How have they done this? Have they just seen emerging platforms in the Anglophone or American ecosphere and then done their own version of it?
To some degree, yes. Sometimes it’s been quite a direct copy — intellectual property rights being justifiably thrown out the window. And in other cases, they’ve had to adapt to local conditions and introduce significant variations. The result today though is that many of the Chinese platforms are more sophisticated, more extensive, and more embedded into everyday life than the American ones.
Of course the problem with the Chinese approach, and why it’s not such a simple model for Europe to copy, is that all of this has been tied in with heavy government surveillance. While the fears over the social credit system are overblown (and often miss that most of it already occurs in Western countries), it is true that there is a close relationship between the Chinese platforms and the Chinese state, and that this is often a relationship oriented toward surveillance and control of the population.
Others say that this is actually enabling communism to finally implement planned economy on scales that we have never seen before.
Very much so, and there’s a book from two excellent scholars on this topic, coming out soon. In it, they make the point that Walmart is a bigger centralized economy than the USSR was. And Amazon is even more sophisticated. Internally, it’s a planned economy that has millions of items that can automatically change prices in response to fluctuations in demand, it can recommend what you need, it can predict consumer demand, and then it can have items shipped and delivered with increasingly less human labor involved.
Let’s be clear that none of this solves some of the most difficult problems with economic planning, but it does point to ways in which we might start rethinking what a planned economy means in the 21st century.