Vili Lehdonvirta is Professor of Economic Sociology and Digital Social Research at the Oxford Internet Institute, University of Oxford. He just released the book Cloud Empires: How digital platforms are overtaking the state and how we can regain control, which analyzes the rise of the platform economy into statelike dominance over our lives and proposes a new way forward.. Lehdonvirta proposes to understand platforms as institutions as powerful as the State. Recognizing this can be the beginning of the work of democratizing them.
In this interview given to Rafael Grohmann, he discusses what institutions are, institutional approach to platform economy, State, global online labor market, network effects, collective action, digital middle class, among other subjects.
DIGILABOUR: What are the gains to an institutional approach to platform studies and platform economy and what would this institutional approach be?
VILI LEHDONVIRTA: Institutions are widely used in different social sciences. One basic definition of institutions that is very popular comes from an economic historian Douglass North. He says that institutions are the rules of the game, that structure interactions in society. Then, you have another definition that comes from people like Anthony Giddens. This is a more sociological definition, in which institutions are just understood as any persistent patterns of social action that are repeated and reproduced across time and space. economists use institutional change to explain changes in the efficiency of an economic system such as a market. They try to explain changes in efficiency by reference to changes in institutions, changes in the rules of the system and political scientists often use institutions to explain distributional outcomes, distribution of surplus and distribution of power in society depending on the rules of the game. Distribution between different social groups, who gets the benefits? Who gets the risks? It depends on the rules of the game. Then sociologists sometimes use institutions just to explain an institutional change, to explain how outcomes are stabilized broadly speaking. Sometimes we have periods of disorder, in a social system and then things fall into order. And sociologists may use the concept of institutions to explain how was disorder achieved and what is that order based on. And unlike economists, they might not say that this order is more efficient or less efficient. They will just say that our order has been achieved, but maybe sociologists, like political scientists are interested in the distributional nature of the order. Who gets one?
I think an institutional approach to studying the platform economy is really useful because it asks us to investigate how tech companies are changing the rules of the game. This can mean examining the rules of a particular platform. We examine how the rules of an online labor marketplace, how they make markets possible, how they make transactions between people who don’t know each other possible by addressing problems such as the problem of cooperation and information costs and so on. And how they also allocate risks and benefits. So we can study the rules of a particular platform, but then we can also study how technology is implicated in changing the rules of society. We have these questions that I know you have talked about a lot around how gig economy platforms may actually be circumventing or breaking the rules, that we have in terms of employment law and labor law and the introduction of a digital platform to a city which allows work to be conducted under this new rule set may be completely circumventing the old rule set. Then we can study the political fight over which rules that prevails. And we know that tech companies spend a lot of money on lobbying national policymakers and local policymakers to make exceptions to the rules or the formal to the laws and also to ignore in terms of enforcement not enforce rule the existing laws and there’s results in institutional change in terms of the rules enforced, because law is only an institution if it is actually enforced and if it is effective. A law that is just a word without any effect on the world is no longer an institution in the sense of social theories, and institutions can be formal laws, they can be informal social norms, they can be technological architectures, they can be any rules that structure or interactions, but they have to be effective. Once they stop being effective they no longer are institutions.
DIGILABOUR: Another important institution is the State. How to solve the overtaking of the platform over the State, or what is the future of this conflict?
VILI LEHDONVIRTA: I think there are two levels or areas in which platforms are competing with States in terms of being the leading institutions to leading rules and one is in territorially bounded markets. I’m an economic sociologist, so I always approached this question from the question of markets and economic exchange. But I think the same applies to the regulation of speech on social media, for instance. So as I used this example before, that gig platform enters the city and then suddenly the previously, the city already like Uber, enters the city which has a formally regulated taxi system and suddenly the rules of the taxi system are now kind of being circumvented by this technological rule set, which is ignoring the rules of the taxi system. And there’s a competition there. So that’s one area. But then the other area is in these transnational markets, in which people interact with other people across national boundaries. I’ve studied these transnational online marketplaces and more than the local markets in the city and the thing with any market is that we know from economic history and economic sociology that markets need some institutions to work, basic institutions like contract enforcement. Somehow you have to ensure that if people make an agreement that then they will stick to the agreement, for instance. I na national economy, it’s ultimately the job of the State to provide these institutions and to enforce these rules, but States are by definition almost bad at enforcing contracts that are cross-border in nature. It becomes very expensive.
For instance, in my research, I encountered a Filipino online freelance worker and he had a client in London. Then the client was refusing to pay and he says, there is no way I can sue the client in court in London, it would just be, it might not be legally possible, and even if it’s legally possible, it will just be far too expensive for a Filipino worker to do that. Instead, what the Filipino worker can do is complain to the platform under which he made this transaction with the London client and then the platform acts as a sort of judge and applies the rules of the platform, not the laws of the United Kingdom or laws of Philippines, but this transnational rules of the platform, and so this is another area in which um platforms are overtaking the State in the sense that they are providing these basic services like contract enforcement and also digital identity is very important. They’re providing digital identity in these transnational settings in which territorial states are not really doing it very well.
DIGILABOUR: Let’s talk about the global online labor market, you have led the iLabour project in the last years. I would like to hear from you about the job politics of this global online market in terms of North-South and also statistical discrimination.
VILI LEHDONVIRTA: We were very lucky to get funding from the European Research Commission (ERC) to do a long 5-6 year project in the end on online labor platforms. So as you know well, there are websites in which people buy and sell work such as graphic design, software development, data entry, virtual assistant services, and so on. So we started to research these marketplaces and one of the very basic things we did was to just start to develop statistics on this online labor market because it’s not really covered by National Labour markets statistics. One of the things we found as you mentioned is that there is a strong global North global South pattern where most of the buyers are in global North countries, especially the United States and certain European countries, and then most of the workers are in global South countries. And now, this is not universally true. There are some categories, such as translation, for instance, in which European countries actually have more labor supply than global South countries, but overall, if you average across the whole market, this tends to be the case.
First of all, we found some discrimination. So, people, clients would pay less for the same work to someone who is coming from a lower income, lower middle-income country than they would pay someone from a high-income country. This is, of course, we know that this sort of discrimination is common. We see it in other markets, we see it in the immigrant labor markets, we see it when businesses from low and medium income countries, when they offer services to the global market, they get essentially paid less for the same services or same good. This is not unique to global online labor markets. But an interesting thing that we found was that when people gain experience on the platform. When they get positive feedback ratings from clients and when the platform metrics of how many jobs they have done and so on start to grow, then the workers’ income goes up, but the workers’ income goes up faster for workers from global South countries. For workers from global North countries, they’re sort of catching up and we explain this through the notion of statistical discrimination that the clients are assuming someone from a global South country to be less skilled than someone from a global North country, not because they know anything about them, but they’re just making that assumption, maybe based on their country image, and so it’s just an assumption. Once they start getting some actual evidence about the competence of that worker, they start to see that “oh, this worker is actually getting good reviews and has a lot of experience”. Then they increasingly set aside the assumption about the whole class of people, about everyone from that country. They don’t apply this assumption, and they look at the individual because now they have information about the individual, the platform provides very detailed information about the individual and that individual space, that individual space starts going up, but for the Global North workers, the pay doesn’t go up so much because they were assumed, based on the country image, to be more competent and more skilled and more trustworthy to begin with. So the new information actually doesn’t change the situation for them. And this is really interesting because it means that actually the online labor platform is to some extent helping freelance workers from global South countries to actually overcome that statistical discrimination to some extent and earn better rates than they could earn if they were working directly outside the platform, trying to reach some clients. But the important thing to note is that this earnings gap between global North and global South is not fully closed, even with additional experience and additional positive feedback. So there is still a big gap. Not all of the assumptions go away and it takes a lot of, you have to do a lot of projects to increase your pay.
DIGILABOUR: What is the real role of network effects and what really matters in terms of competition and their limitations too?
VILI LEHDONVIRTA: we’re just talking about a handful of very large companies, they have used also some anti-competitive tactics to become large, right? We know that it’s not all just network effects, it’s also some business practices like Amazon has sold some goods at a loss until competitors had to give up and then they could own the whole market, but there are also network effects and I think what the institutional approach helps us to do is to break down a little bit what do network effects actually mean on a more detailed level, in the case of a marketplace, because the theory of network effects is very abstract. It just says that the value of the platform grows, the more users the platform has the value of a platform to a user is higher the more other users there are on the platform and this leads to a positive feedback loop and kind of winner take all dynamics in platform competition, but this is quite an abstract picture. What, why does it become more valuable and so on. And what the institutional approach allows us to do is to put a little bit more detail into it and say, well, the platform is a set of rules. It’s an institutional framework for doing business or for engaging in economic activity? If you want to do business with someone or interact with someone economically you have to follow the same rules as they do. It’s not possible that you follow different contract law than me. Because if we do a contract together, we have to agree, ok, what are the rules that govern this contract? We can’t have two different sets of rules in mind, because then when we have a conflict, we don’t know who is right, so we have to agree to follow the same set of rules if we are doing something together, and that means we have to agree to use the same platform and this places limits on competition between platforms as a means to make platforms accountable to the users. Because the problem, as you know, we have is that these very large platforms, they’re very, very powerful towards the users and they abuse that power in many ways. And so then we ask how do we make these platforms more accountable? And a lot of people are saying, well, let’s make them compete against each other. So instead of having one big platform, let’s make two platforms compete against each other and then users can vote with their feet and that forces the platform owners to be nicer towards the users.
But if we, once we understand that the platform is actually an institutional framework, then we understand that a single user cannot themselves decide to just change to a different platform because they have to be in the same platform where the people who they want to interact with are. And this means that actually in order to change platforms, all those people who are interacting together on that network, have to switch together, not one by one. The switching cost of switching alone is too high because you’re leaving behind all your contacts. You’re leaving behind your reputation, your cultural knowledge, your name recognition and so on. So actually the whole community has to switch together and this is a collective action problem, right, to getting everyone to switch at the same time, that’s a collective action problem and so competition is based on the idea that individuals should choose, but instead of competition, we have to think about political institutions, because political institutions are ways for collective decision making.
And if that community of the users of a platform can make a collective decision, then they can make the owners of the platform accountable. They can tell that the owners of the platform look, unless you change the rules where we’re going to leave you or we’re going to go on a strike against you, or so on and so on.
DIGILABOUR: What are the biggest challenges regarding platform workers organizing and the role of “digital middle class” in terms of potential for collective action?
VILI LEHDONVIRTA: most of the focus at the moment, is on collective action among the sort of working class of the platform economy. And I think that if we’re talking about those local platforms in the city, then gig workers can organize and they can go on strike and so on. But I’m mostly focused on studying the online platform economy. And the online context, of course, presents some difficulties for collective action, because although like let’s say in a platform like Mechanical Turk, these platforms are very important today because data labeling is a big job in the AI-powered platform economy. There are a lot of data labelers and quality checkers are needed, but the workforce, this kind of a digital labor force is spread around the world, not evenly. Of course, some people that’s more concentrated in some places than other places. But still, two people might be working together on Mechanical Turk or the same tasks in Mechanical Turk, and they might be living next to each other, be neighbors, but they wouldn’t know if they meet each other on the street they wouldn’t know that we are actually working together and that makes it harder for them to organize.
Now, of course, we have some online communities like Kristy Milland. She was the lead admin of Turker Nation, a very important online community of Mechanical Turker workers. And she used the Turker Nation to try to organize the workers and organize a collective action against the directed at Jeff Bezos to try to get basils to listen to them and give them a voice in the rulemaking of the Mechanical Turk platform because it was not treating the workers well that there were constant problems. And, in the end, this collective action, unfortunately in terms of being able to mobilize a lot of workers, it was not successful, and there are similar actions that have been attempted and they have not been successful in mobilizing a lot of workers and I this is a very important question. Why?
The answer to the question is not because the workers are not interested, it’s not because they don’t need collective action. They have some, in some cases, the data labelers very bad working conditions and the pay is very low. So there is a motivation and there is a need, but the problem is they are in many cases there are two deprived they can’t afford to take the risk of demonstrating because they are afraid of losing access to the platform and some mechanical that we know cases where people have been thrown out of Mechanical Turk for some violating the terms, so they can’t take the risk that protesting against the rules and demanding a voice could result in them being thrown out, and they’re also because they’re working so hard. The Mechanical Turkers are working incredibly hard. They’re working very long hours in many cases. They don’t have so much time for reading about them, thinking about politics. They’re struggling to make ends meet.
Another collective action we studied with my colleague Alex Wood was on Upwork, an online freelancing platform where you have more highly paid online work. So you have software development, graphic design and so on. That was much more successful in terms of attracting a lot, they had got several 1000, the workers joined the protest and, interestingly, the workers who joined the protests. They were not the ones who most needed the protest, they were not from global South countries. Actually, it was the global North countries people who were protesting mostly, the better paid online freelancers and my explanation for this is that, because they are in a better position in a more sort of middle-class position, they have more time and they’ll have more time to participate in politics and they have less risk from participating, because even they’re less afraid of being shut off the platform because they have skills and connection, they have alternative opportunities, they are in global North country, they can find some other work. So they are less afraid to protest.
But they have solidarity with the less well-off workers because they are on the same platform and they can see people struggling there. So they have solidarity. They’re not personally so threatened by the platform. They’re quite well off. But they have solidarity with those who are threatened by the platform. And those who are threatened are in a precarious position to fight for themselves. So some people were better off who are outraged by what the platform is doing. They step in and protest. So this is quite interesting.
Even Kristy Milland herself, while she was still working, she worked incredibly hard on Mechanical Turk because she had to actually support her whole family. She was based in Canada, in Toronto, and the living cost is quite high. Her husband lost his job and they have a child, so she has to support the whole family on Mechanical Turk income and she’s working so hard and she develops like repetitive stress injury and all kinds of she some very disturbing images and so on. But actually, while she is still working herself, she doesn’t start campaigning because she’s just too overwhelmed by the work. It’s only after her husband finds a job and she doesn’t have to work on Mechanical Turk anymore. That’s when she starts to campaign, because she still has solidarity for those people who are left, who still have to work on the platform. She doesn’t have to work on the platform anymore, but she has solidarity with those who do. And now she also has the resources to actually do it.
DIGILABOUR: What are the differences between institutional and Marxist approach to collective action?
VILI LEHDONVIRTA: It’s a great question and to be honest, I’m not an expert on Marxist theory, but what I kind of discuss in the book is that Marx talked about how the very nature of capitalists work organization generates the possibility of collective action because he was observing early 19th-century factory towns. And in the factory town, everyone work in the same factory they go to work when the factory whistle goes off in the morning, the same time they work in with the machines next to each other, after work, they all get out of work at the same time. They all go back to the tenement houses, where they live together in small, cramped conditions and they eat the same food. They really can see each other as belonging to the same class that has the same interests. And they can also easily communicate with each other because they are, well, physically next to each other. And then in the digital context, all of this is different because people work in these platforms, they work very different hours, they come from very different kinds of backgrounds. Some are students doing it as a part-time job, some are trying to support a family like Miland, they come from different countries, different cultures, different languages. And this presents a lot of difficulties for collective action that I think to Marx, maybe these workers would have appeared on organizable.
Marx is a kind of institutional process also broadly understood, but the approach that I then take is to kind of look at the Mancur Olson, “The logic of collective action”, which is a theory that seeks to explain through people’s personal incentives whether collective action succeeds or fails and it has quite a pessimistic conclusion, because Olson concludes that because of the free rider problem, it’s very difficult or even impossible for a large group of workers to successfully engage in collective action, because always there is somebody who thinks I will not participate, that way I don’t have to bear the risk and the cost of participating, but I still get the benefit if the action is successful. All the workers get the benefit of better working conditions, but if the action is unsuccessful, only those who participated in the action paid the cost. So workers who think selfishly. We’ll choose not to participate in the action and hope that other people do. When they can benefit, they can ride for free on the back of other people’s efforts and if enough people think this way or if enough people think that other people, they may think this way, then the whole action fails. It doesn’t. People don’t participate. The only way also said that large groups of workers can get around is if or one of the only ways is if the workers can somehow or worse, each other into participating. It has been traditionally in protest in manufacturing for the workers to set up picket lines to physically stop strikebreakers from entering the factory, so you cannot choose not to participate. You have to participate. If you try to participate, if you try to work and not participate in the strike, you will be stopped physically.
But this is very difficult or maybe impossible to do in digital labor platforms. The workers cannot set up a picket around the platform and they cannot see whether other workers are working or not, so they don’t know if somebody’s breaking the strike or not. So this makes things even more difficult, but then there are others who have later commented on Olson’s theory and said that ok, but this focuses too much on the idea of this rational, selfish individual and there are other motivations besides immediate personal benefit that motivate action and one such motivation is solidarity towards others, and the other is perceived outrage or outrage at the perceived breaking of a psychological contract or a social norm.
And, in the case of the Upwork protest that I told you about, both of these things happened because Upwork suddenly essentially doubled the taxes, doubled the fees for most workers, especially for the poorest workers. It was a regressive tax hike where the workers were having the biggest, longest contracts. They actually pay less tax as a percentage, only pay 5%, and the ones who do short term very precarious gigs, they have to pay 20%. This was perceived as a breaking of a longstanding kind of institution, informal institution, informal rule that there is a flat 10% tax and so this cost outrage. What was one factor. And also solidarity between the better off towards the worse off was another factor. So I have people saying that, well, this tax hike we have somebody from London saying, well, this tax hike actually doesn’t affect me personally. But I feel outraged that the platform is attacking the least well-off and using, abusing their power and that’s why I’m protesting.
In my theory, I take all these factors into account to explain collective action, but still, so there are three factors. There are the resources, which means do the participants have enough wealth and security that they can afford to pay the cost of participating in the action and take the risk of participating in action as one factor and then the solidarity and then outrage at the perceived breaking of the social norm and when we take these three factors into account, the most deprived people in the platform economy the working class, they don’t usually have enough resources to participate.
DIGILABOUR: What is the role of middle class on that?
VILI LEHDONVIRTA: The middle class, people who are earning more, may have more resources or I would say actually the Upwork Online free losses there, maybe not in my model of the class structure of the platform economy, there maybe not quite middle class, maybe there are like upper working class or lower middle class and so they have just about enough resources. But they are all very importantly, they also have solidarity with the working class because they work on the same platforms, so they use their resources to fight, but they still don’t have enough resources to mount a successful campaign. In the book, in this action, they managed to get thousands of workers to protest in very short notice, but it was not enough to actually change the platform’s decision. They still changed the tax structure. They ignored the work as it was not enough, it’s not successful. It was successful in mobilizing more people than the working lower class protest, but it was unsuccessful in actually changing the platform decision.
Then we get to the upper middle class, and in my model, these are people like app developers on Apple App Store and successful online merchants on Amazon or eBay or successful influencers and streamers. These are people who are making a lot of money from online platforms, they’re not the owners of the platforms, so they’re not mega-rich billionaires, but they are still financially doing very well. These people, if they decide to protest and campaign and engage in collective action, they have enough resources to actually make changes happen. In my book, I tell the story of this app developer called Andrew Gazdecki, who was running, he was from a very modest background. He was from a kind of very low learning background in the United States. All throughout his childhood he had to kind of do all kinds of jobs and entrepreneurship to make some money. In college, he got to college and he founded this app development company and it became a moderate success. Finally, it was starting to look like he’s going to be wealthy in the future. He’s really dreaming. He was really dreaming of being wealthy and ensuring that his children don’t have to go through the same deprivation that he did when he was young and then he was already negotiating the sale of his company to towards or he was getting offers from companies that wanted to buy his company so he could exit and become rich and call it a wonderful life.
Just as this was happening, then suddenly Apple changed the rules of the App Store and said your business is not allowed anymore, your app is now illegal, completely destroying the value of his business. What he actually did was then he started to organize a campaign. So he got in touch with other app developer companies in the same position, in the same situation and they started together this big campaign where they started an online petition. They started calling all the journalists because they had some connections in the media. They pulled their money together and they hired a lobbyist to start talking to lawmakers in Washington, DC, and attention and they got thousands of other app developers kind of joining in their online petition and writing these messages to Apple and protesting and so on. They managed to put together such a formidable campaign that in the end, Apple had to say, “ok, you win” and remember, this is the world’s first trillion dollar company, the biggest company by stock market valuation in the world ever, and very, very powerful. And they actually had to say, ok, you win.
This is a really interesting example. Of course, it’s only a single case, but there are other similar examples so then others their app developers protesting against Apple and Google kind of forced Apple to lower their fees, App Store fees for smaller developers from 30% to 15%. So this shows that it is possible actually for the users of a platform through collective action to influence the rules of the platform. But at the moment it really takes a lot of resources, so those who succeed in it, they’re the upper middle class. They’re not the working class, unfortunately.
In the middle class, they are starting to push back against the platform power. Of course, we see the same happening in the Middle Ages in Europe and that’s how democracy kind of first started in the Middle Ages in Europe. Of course, we have Greek democracy before that, but we have the peasants or the lowest social class in the feudal society. There are too poor to succeed in protesting against the Lords. But the middle class is growing and prospering in the markets of the High Middle Ages, so merchants and artisans. And they start pushing back against the feudal Lord’s power and they start saying, hey, we want to have a say in how these market towns where we live, how they are governed.
And ultimately they actually managed to win for themselves the self-governance. We have many autonomous market towns emerging in Europe in the Middle Ages which are governed by the burgers. The townsmen and not by the feudal Lord. And so this is kind of analogous now to what we see maybe happening in the platform economy, but then when we get to the 18th and 19th century, then the working class becomes very important, because in the Middle Ages it was still only the rich middle-class property-owning men who really get to participate in governance, and so in the 18th and 19th century, the working class movement expands those political rights to everybody to first to all wealth classes and, then also to women. At the moment, in my analysis of the platform economy, we’re still in the early Middle Ages. And we haven’t even got to the point where we have autonomous market towns run by the burgers, the middle class. But once we get there, then the working class has to come and expand those rights to everyone. And in Europe, this process took hundreds of years to unfold, but so far on the Internet, this institutional change has happened at lightning speed. I kind of claim in the book that the Internet has recapitulated the past 3000 years of economic history in just 30 years.