What are the conditions of workers in the platform economy, what happens beyond the interface? Sarrah Kassem, lecturer at the Tubingen University who will join Scuola Normale Superiore in the coming months has researched and written on the issue. Kassem was a guest at the Istituto Ciampi’s conference “The impact of AI on society”, we have asked her a few questions on her research.
In your book you talk of three different generations of platforms. What are the changes and what changes for workers?
Many things have changed in just a few years. The first phase is when people started the internet became a mass phenomenon and platforms offered services that already existed by bringing them online. In the mid90s Amazon took the sale of books from stores to the online, EBay did the same with goods in general, while Google is a bit different because it was trying to categorize the internet. The organization of labor was also traditional: you went to an office, did your work, had a salary. This is before the dot.com bubble burst, after which some became much bigger and many did not survive. The second phase is that of targeted advertising when platforms were monetizing on the data they were collecting. Google went from being a search engine to rebranding and focusing on targeted advertising, Facebook came to the market and also built on targeted advertising. Facebook was able to grow because of the exponential growth of people connected to the internet. In this phase you have the birth of social media and the expansion of Amazon, which went from an online store to rolling out its Amazon work services, its cloud that has almost one third of the market and the Mechanical Turk (MTurk), the online microtask platform. Compared to the 1990s when work was organized in a traditional way, with MTurk workers could be located anywhere in the world. It was a tool that was ahead of its time, as the gig economy started later on. At that moment outsourcing and offshoring were already the new normal but in this case, you were an independent contractor, not a worker anymore. As long as you could connect, you could do tasks online from anywhere and got paid if the task was approved by the client. This is a break from the traditional work and it picked up massively after the financial crisis of 2008, the moment in which the gig economy was born, even if it was at first disguised as the sharing and on demand economy. This flexibility looks very different because workers are not protected, receive a gig wage (pay per ride) but are expected to pay all the expenses. It is a form of hyper-outsourcing. So the history of the platform economy is a story of making labor more precarious and not protected, even if at national level, here and there (and at the European Union level) there are attempts to try to reclassify these workers as such in order for them to get labor rights.
You write of Amazon’s warehouses as a digital controlled taylorist factory. What do you mean?
Amazon has successfully implemented a way to organize work in a way as efficient as possible to make sure to be the most customer centered company in the world. The customer makes an order and it must receive it in the most efficient and fastest way possible, even in the highest of seasons like Christmas. And the way they are able to do this it’s a kind of Taylorist factory of the 21st Century. An Amazon warehouse is divided in “Inbound” and “Outbound”. Inbound is where the products arrive, is scanned and labeled by a worker, then another would store it. Here the efficiency is related to the chaotic way of storing: similar products don’t go together, but you store wherever there is space: a teddy bear next to a book next to a hat. When in the “Outbound” phase I pick up a product, I go search for it and, in the slot where I am looking, I will just one so that I don’t need to be sure if that is the right brand, size or color. The Outbound is the part of the process in which you receive an order, you go pick it before it moves to packing and then to the shipping. Workers have a unit per hour wage, so the workers I have interviewed spoke picking around 120 items per hour, but in other reports and research we know that elsewhere it is many more (double in Poland, 400 in some US warehouses). This means that as a worker you have to perform to keep this rate up for the whole working day. When you work in the outbound the performance pressure is higher as the customer is already waiting. The division of labor is hugely tayloristic and at the same time you have the surveillance by the manager plus the technological one, which is embedded in the device you are using. The social surveillance would deny you a bathroom break – but this depends on the laws and rules that change from country to country. As a picker, a hand scanner can see if I am keeping up with the performance that is expected from me. Workers told me that the warehouse would haunt them in their sleep, nightmares that their ID would not work because they have been dismissed – although this could not happen in Germany, where you cannot dismiss from one day to the other. Amazon has cracked on how to make the process as efficient as possible by keeping the managers and at the same time incorporating the technology.
Interesting enough, with some exceptions Amazon tends to build its big warehouses where unemployment is high. In Bessemer Alanama, the anti union message was: “if you unionize, we move and there’ll be nothing lefty nothing here”.
Amazon warehouses are mainly built in places close airport or highways and often placed in areas with high unemployment rate. This guarantees access to labor market: depending of where that is in the world the turnout of workers can be really high, so you want to guarantee that there will always be workers coming in to keep up with the demand of the customers/consumers. One the warehouse is built in a place with high unemployment rate Amazon offers a salary which is higher than the minimum wage, insurance and the possibility to join a Union (this depends on the country); from the Amazon point of having your warehouse where there is high unemployment allows pushing workers to their limit with no risk of having labor shortages: you’ll continue to have labor offer even if this is an exhausting and extremely high performance pressure job.
Let’s talk about the digital work, which is increasing but invisible. How is it different from the one in warehouses?
We are accustomed to interact with riders or drivers delivering packages, not to the digital work is even though we are very much connected to the work that is put up by these workers. In this field there is micro-tasking and macro-tasking. It could be to label data or copy receipt numbers, or it could have a big project such as a media campaign, coding, video editing. Research has shown that predominantly those who post tasks online are in the global north and those who do tasks are in the South. This is important because offering a few dollars in the US or Europe is different than offering the same amount in the South. It could be an opportunity in places where there is not much work, or in patriarchal contexts where it is easier for a woman to work from home. Web based work has allowed people who owns a device to connect to access these labor markets. The barrier to enter these markets is very low. The macro task platforms, such as MTurk can vary very much in terms of pay, from few cents to several dollars. As you do this kind of work you can see and feel that the only reason you are doing this work is because you are a human being. This is why also Amazon calls these tasks human intelligence tasks. You can feel how you are contributing to feeding the data that will be used by machine learning algorithms to train AI. So just like we are on our phones or laptops and we have to prove that we are not robots clicking on traffic lights. This is a big portion of the tasks that are out there, they are important giving how central AI is becoming and is being incorporated across or society and economies. The kind of work they do, as I mentioned, is extremely precarious. On the one hand it means people can access a labor market in places that give no chances, or they can come from a context in which most people work in informal economy and don’t have a contract. Because with MTurk they make sure to remind you you are not a worker of Amazon and at the same time you don’t work for the person or company posting the task. The request can come from a massive corporation, a start-up or even academics. We actually started noticing MTurk because it was overwhelmingly used by academics, which is not okay. These workers are paid by task but only if it is approved. You could finish your job, submit it and the requester can decide it is not good enough and reject it. They also allowed to use your work and still not pay you. This is wage theft. Some tasks can be completed by thousands of workers and the approval or rejection is carried out by an algorithm, which can end up in cases of mass rejection. Who do you communicate with in these cases of algorithmic management? In an Amazon warehouse you know who the management is, while with Uber you can be hired and fired by the application. Also in MTurk you have an algorithmic management that calculates your time, approval rating, etc. And depending on your approval rating you access (or not) a certain market or you lose that access. Workers don’t have insurance, they cannot be represented by Unions, the traditional labor rights are disappearing in the gig economy and are pushed to an extreme with the online work. If the company is in another part of the world and can even anonymize itself when posting tasks, who do you turn to? Amazon will tell you that is only there to mediate and provide the platform. According to whom you would have a minimum wage? The place where the worker is doing the job? The place where the company has its headquarters?
Amazon sells flexibility as freedom, what does not work in this discourse.
My first instinct is to ask flexibility for who? If you have to take care of another person or children, you can squeeze in doing some tasks in between in a way that would not be possible if you had a traditional job. The question is what kind of reality does this translate into? You have space flexibility, but if you dive more, you find that if you have tasks coming from the global north and I am in the global south it means there is time difference and maybe my working day needs to adjust to the European time even if I am in India. The other question is the distinction between paid and unpaid time: Uber driver is paid once the ride is completed and the in between time is unpaid. I get paid per task, but I spend time to search and apply for tasks, so I get chunks of wages based on completed tasks. The other question is: is this online work a secondary or primary income? On MTurk you have everything, those chained to their laptop to students who would tell you “I do it in between”. You have the whole spectrum and the more dependent you are on the gig economy for your income, the less flexible you are. You have to fully function every day hoping to catch those higher paying tasks and this is the same with drivers and food delivery.
Labour organization and transnationalization. What are the strategies? What works? There are transnational issues and local issues.
This is something that I have heard warehouse workers emphasize in many occasions: this is a transnational company and we have to organize on a transnational level. The perfect example can be the case of Germany, which outside of the US is the largest market and where the first strike in Amazon history has happened. They have organizing for 10 years asking for a collective bargaining agreement with no success. So, if you don’t have obstacles as in the US to join a Union, they did not manage to get what they fought for. One of the main reasons here is how to coordinate the local and the national level and what do you do when Amazon has also warehouses in Poland that cater to the German market. So, even if you can unify the campaign at the national level you have warehouses just across the border sending products to the customer and this weakens tremendously the organizing. The whole system is very decentralized, in most warehouses you have a bit of everything. Amazon is very smart in order to weaken Union organization. When it comes to the local level it depends on different factors: older warehouses can be more organized, different conditions (worker on a Visa or a national citizen), different languages, which also complicates coordination and organization. In a country like Germany you are not going to loose your job if you join a Union, but that is something that you need to clarify, you need to reassure. But organizers are also punished: many told me that those who are more active are more often employed to be pickers, which is the most exhausting of the tasks, you walk around 20 km per day. There are challenges to organize workers in general and you have these undermining conditions and on top of that you have a company that has the reputation to be a Union buster. With Tesla Amazon are on the top of the list of companies who would do anything to weaken labour organizing.
You end by talking about the power of workers…. Tell us about it.
Just because the reality looks like this, we know that workers of the past are the reason we have rights that we can enjoy today. When it comes to Amazon’s warehouses, I would say that the leverage of workers we can perceive and conceptualize in those traditional terms. Meaning: the ability to disrupt work, that in theory exists if workers are able to organize they can in theory bring the circulation to a halt or come up with campaign to coordinate that mobilization. Has this ever happened? It does but we don’t always hear about it. It has happened that managers had to go and pack items. So, the bottom line is that there are instances where we can see that there is this potential. Why does that not happen is related to the obstacles I have mentioned. There is a global campaign called Make Amazon Pay that around Black Friday brings strikes where this is possible and solidarity of some form elsewhere. The rules for unionization or collective bargaining agreement are crucial and this is why I stress that the context is so important. We live in a time where Amazon is synonymous of shopping across countries and generations (which was not the case 10 years ago). When it comes to MTurk workers which is a category that is beyond the gig workers who are location based as riders or drivers, that were able to organize, you will not find traditional organizing. How can you even think of a global strike online when you are working in your individual online cublicle. If I decide not to work, no one will notice, there will just be one less task delivered. Plus, gig online workers are not even classified as workers. But on a societal level these workers may appear as invisible, but their work is very much integrated in our lives and they show solidarity in digital communities. On Reddit you’ll find lists of tips and advice on how to find good tasks or you have a community. TurkOpticon for instance started with researchers conducting surveys asking workers what issues bother you the most. Asking workers is a way to understand. Today on TurkOpticon workers rate customers, ask questions and share information, warning on who to avoid because it will not pay you and who will. There are less traditional ways to organize and we can learn from those ways and incorporate also digital tools into the wider labour struggle.