Tech Won't Save Us - The Poorly Paid Workers Powering Automation w/ Phil Jones

Episode Date: February 10, 2022

Paris Marx is joined by Phil Jones to discuss the hidden microworkers behind supposedly AI-powered automation from major tech companies, how it differs in the Global North and South, and what it means for how we think about the future.Phil Jones is the author of Work Without the Worker: Labour in the Age of Platform Capitalism and a researcher at Autonomy. Follow Phil on Twitter at @philjones7771.Tech Won’t Save Us offers a critical perspective on tech, its worldview, and wider society with the goal of inspiring people to demand better tech and a better world. Follow the podcast (@techwontsaveus) and host Paris Marx (@parismarx) on Twitter, and support the show on Patreon.Find out more about Harbinger Media Network at harbingermedianetwork.com.Also mentioned in this episode:Phil wrote about digital piecework for The Guardian and had an excerpt about refugee labor in Rest of World.Turkopticon helps workers on Amazon’s Mechanical Turk gain some information on the contractors offering tasks on the platform.In 2020, Gizmodo did surveys to find out about workers’ experiences on Mechanical Turk. There were a lot of horror stories.In 2014, workers on the platform sent emails to Jeff Bezos to ask for better conditions.Support the show

Automation is not simply a technological process. It often involves hybrid forms of task fulfillment, which rely on machine learning and poorly paid workers. Hello and welcome to Tech Won't Save Us. I'm your host, Paris Marks, and this week my guest is Phil Jones. Phil is the author of Work Without the Worker, Labor in the Age of Platform Capitalism, and he's also a researcher at the Autonomy Think Tank. Phil's book is an essential contribution to our understanding of work in the 21st century under the internet, these platforms, and the

digital technologies that we are all used to and surrounded with. In recent years, more and more of our society is said to be or promised to be automated or taken over by computers thanks to AI and automation. However, Phil's book shows that so many of those tasks that we think are being done by computers are actually being outsourced to very poorly paid human workers who are working through platforms where the tasks that they do can be as short as 30 seconds to check things for a so-called automated system, whereas when we see it, we think, oh, okay, that's just all being done by the computer. Wow, look at how advanced this is. Well, that's not always the case. And if you thought that the gig economy was bad,

these workers on what Phil calls micro work platforms are often in even worse conditions. The pay is terrible. There's constant precarity because it's easy for contractors not to pay the workers even when they complete a task in the way that the contractor wanted it. And they spend a whole lot of time looking for and competing for tasks instead of actually doing the work.

And companies and organizations that promote this form of work take advantage of people who are in incredibly desperate situations, including in refugee camps, to do this kind of work that then goes to benefit major corporations like Amazon, Google, Uber, Tesla, and many other of the names that you will be familiar with in Silicon Valley. I want to draw attention to one thing that Phil says in the conversation, and he notes that NGOs have been some of the groups pushing these micro work projects in the global south as a way to provide work to people who are often working in informal economies. And he notes that these NGOs are promoting this as this really positive, hopeful thing that can help people in the global south. And he

compares it to previous promotion of things like microloans and microfinance that didn't work out in the way that the narratives that were being promoted in the global north made it seem. And I would also note that if you think about NGOs pushing these micro work projects, think back to the conversation that I had last week with Jacob Silverman and my previous conversation with Olivier Jutel on how NGOs are also pushing these cryptocurrencies and blockchain solutions that have serious problems with them. But once again, they're jumping on the hype bandwagon and helping out these companies that are not really working in the interest of these people in the global south to ensure that they are able

to actually better themselves and things like that. So I would say keep that in mind. But there is so much more in this conversation that is important to understand. And I was so, so happy to talk to Phil about such an important issue. And I think you're really going to like this conversation. Tech Won't Save Us is part of the Harbinger Media Network,

a group of left-wing podcasts that are made in Canada. And you can find out more about that at harbingermedianetwork.com. If you like this conversation, make sure to leave a five-star review on Apple Podcasts or Spotify. You can also share the show on social media or with any friends or colleagues who you think would learn from it. And this episode of Tech Won't Save Us, like every episode, is free for everybody because listeners like you support the work that goes into making it every single week. So if you like this conversation, you can join patrons like Megan Rose from the United States and Marlon from Vancouver by going to patreon.com slash tech won't save us and becoming a supporter. Thanks for listening and enjoy this week's conversation. Phil, welcome to Tech Won't Save Us. Thank you for having me. It's a pleasure to be on my favorite tech podcast.

Thank you so much. I'm thrilled to hear that. You know, you have this new book, Work Without the Worker, that I think looks at like an element of work since the arrival of these digital technologies and the internet and stuff that doesn't get nearly as much attention as it should. And, you know, hopefully, this is the beginning of us focusing a bit more on what's going on with this kind of micro work, click work and things like that. And so, you know, I think to start, because there might be people who don't even

realize that this is a thing, we should start by getting an idea of how this came to be and what it looks like. So, you know, a narrative that we still hear from time to time, but was particularly popular in the second half of the 2010s is that automation is going to wipe out a ton of jobs and leave people destitute, right? There were a bunch of books, you know, on this premise saying that this was what we had to expect for the future. All the stories about how truck drivers were all going to lose their jobs. That seems kind of silly right now in the midst of the supply chain crisis that we're

having right now. But in your book, you know, you look at an important transformation to work that companies have pioneered by taking advantage of digital technologies, which call that automation narrative into question. So what has actually happened to work over this period? The automation story has definitely lost traction in recent years, as you say. Partly, I suspect this has gone hand in hand with a greater loss of faith in big tech as a whole. The sense that Amazon, Google, Facebook, et cetera, are not so good for us. And perhaps more fundamentally, they rarely stick to their kind of bombastic promises about self-driving cars within five

years, et cetera. In my book, I tell a very different story. The real driver of unemployment, underemployment, precarious work, informal work, et cetera, has been a dip in global labor demand set in motion by the crisis of overproduction that began in the 1970s. And in many respects is still with us today as widespread economic stagnancy. This context has created lots of badly paid service jobs, but it was also the context from which big data and machine learning emerged. After 2008, capital continued to struggle to profit from manufacturing, as it had done since the 1970s, so turned itself towards speculative investments in AI futures.

In the early 2010s, we started to hear lots about automation, the third industrial revolution, et cetera, all made possible by developments in artificial intelligence. But what we didn't hear about, and you alluded to in your question there, are the kinds of jobs that were needed to support artificial intelligence and indeed still are needed to support artificial intelligence. indeed still are needed to support artificial intelligence. So there are programmers, all manner of tech entrepreneurs, but there are also lots of poorly paid workers doing short data tasks on sites like Amazon Mechanical Turk, Clickworker, and Appen. These sites act as intermediaries between contractors who use the sites to outsource tasks. They're usually big tech companies like Google,

Amazon, or maybe a smaller startup. And then workers will jostle to complete the tasks. So a standard task might involve labeling images of city streets with labels like car, pedestrian, traffic lights, to show an autonomous vehicle how to navigate an urban center. Or it might involve recording specific accents to train chatbots or annotating images of faces to train facial recognition technology. The tasks are super short and for this reason are often referred to as micro work. They might last anywhere between 30 seconds to 20 minutes and are paid very poorly. You know pricing of a task will be 30 cents for 20 minutes, 15 minutes. Because of the brevity of the tasks, because of how short they are, workers might do 30 to 40 different tasks over the course of a single day, perhaps more than that,

for 30 to 40 different contractors. And it should also be added that like with gig economy companies such as Deliveroo or Uber, the workers are not classed as workers or employees, but independent contractors. So they have no rights. There's no regulations around working conditions. And this means that contractors can often get away with murder, effectively just not paying workers for the work that is done, often without any kind of justification. Yeah, especially on that pay piece. It was really shocking to read in the book, like the degree to which non-payment is facilitated by these platforms, by the way that they are put together, like just the way that they are constructed. It leaves a lot of room for the people who are putting these tasks on the platforms to not pay these workers at all. One of the ways I sort of think about it in the book is that micro work moves us

from a wage to wager so that the wage becomes more like a gamble. The worker doesn't experience their pay as a contract, as most workers should and do, but instead will experience it more like tokens or rewards. So if you do the task well, if you do the task to the standard that the contractor has set, then you will get your pay. One of the largest surveys carried out across micro work sites found that up to 30% of workers will regularly go unpaid. And on Clickworker, one of the main sites, they found as much as 15% of all tasks go unpaid, which is quite extraordinary, really, when you think about the amount of workers that are regularly doing work, but getting no remuneration for it. Yeah, it's wild to think that if I was working a shift job or something,

and 30% of my shifts, I just didn't get paid for them. That would be wild and unacceptable. But because it's this micro work that a lot of people aren't paying attention to, where it's people who are very disempowered, very small tasks, it's a lot easier for these companies to get away with these things and not be held to task for it. Absolutely. Yeah, you've hit the nail on the head there. Part of the reason

for this is that these platforms are incredibly opaque. So it's very difficult to see what's going on on these platforms as an outsider. So it's been very difficult for journalists to make contact with workers. It's very difficult for workers themselves to often get a sense of what's really going on on the platform in terms of the wider projects that their tasks are going towards, in terms of the kinds of companies that they might even be working for. Hence why it's been very easy for these platforms and the contractors that use them to facilitate a scenario in which workers are regularly unpaid. Now, obviously, we could look at micro work as this thing that is the same everywhere. It's just these micro work platforms. But I feel like in the book, you describe how this works a bit differently in different parts of the world, right? There's a distinction between micro work in the global north and the global south. So I wanted to ask separately about what

it looks like in those two areas of the world. And I know that's obviously generalizing in its own way, but we'll start with the global north. Can you explain how micro work and, you know, I guess the gig economy, for example, as well, have actually brought back an older form of work and what implications that has for workers and our society as a whole. In the US and Europe, surveys over the last decade would suggest that this, you know, that micro work is very much on the rise and that it's mostly done by students, disabled people, and women, not least because the work can be done from home and can fit around the second shift of care and housework that women will often undertake. It also tends to act as an adjunct to other part-time or insecure work. So rather than

representing a full-time job in itself, it seems that there are sort of growing numbers in countries like the US using micro work as a way of topping up the stagnant wages that they've got in their more full-time employment. Equally, there do seem to be increasing numbers, at least in the US, who are using this as full-time employment as well. As you mentioned in your question, in the North, microwork represents a return essentially to what we might call Victorian capitalism, where employment came with very few rights, benefits, or wage minimums. There are significant similarities actually between digital piecework on micro work sites and the Victorian sweating system,

where there would be a middleman hired by a capitalist who would watch over piece workers, pushing them to work harder, in other words, sweating them for greater profits. Micro work sites fulfill a kind of similar function for tech corporations today. On most of these sites, workers are paid only for tasks that are completed to a sufficiently high standard. So, you know, like the sweating system, as I just mentioned a minute ago, wage theft is incredibly common. The platforms basically seem to be set up so that contractors can readily avoid paying workers in the same way that the sweating system in Victorian England was. Contractors need not offer a reason for refusing work, which effectively means that even if the

task is done to a decent standard, they can decide not to pay. As I said a minute ago, this means that a large percentage of tasks will go unpaid. Where micro work differs, I'd say significantly from the Victorian sweating system, are the forms of management that it uses. So punitive rating systems, which look more like aspects of a Taylor mismanagement, kind of like Taylor mismanagement on steroids. Through these systems, platforms effectively hand contractors the power to entirely marginalize workers from the platforms. So contractors can review workers, but workers can't review contractors. And the contractors can give basically any review that they feel to the worker, and the worker has very little recourse to protest or take action against

the contractor. The workers will get a composite score determined by their past ratings. And if their score drops too low, it will be impossible to find more work on the platform because contractors won't want to work with them. Equally, if they act in ways deemed unacceptable by the platform and contractors, they can be kicked off without warning or reason, which as you can imagine, is financially devastating for workers, considering that for many, this is their main source of income or is a key source of income. It's worth emphasising as well that while there are great many similarities between the Victorian peace worker and today's micro worker, the labour conditions on these sites are actually far more similar to those of surplus populations in

countries such as India, Kenya, Venezuela, where the great majority of this work is completed. I think it's fascinating. And we'll make that shift to the global south in just a second. But I think one of the things that stood out to me, and you give a ton of examples of how this works in the book, you talk about how with the gig economy, and with this larger transformation of work that's happened with the gig companies in recent years, you know, we've talked a lot about algorithmic management and things like that. One of the examples that stood out to me was you talked about how Uber actually farms out a number of the tasks to happen for, you know, these micro workers to then say, approve workers photos of themselves and things like that to make sure that they align with, I guess, the expectations of the

platform. And so then that kind of indicated to me that, okay, yes, we're talking about all this algorithmic management, but also how much does that algorithmic management rely on these micro workers that people don't know are behind the scenes? Yeah, so this is an example I give in the book. Uber is famously or infamously automated taxi driver management, replacing taxi managers with algorithms, which will organize the workflow, dictate task length, wage prices, etc. But behind the scenes, much of this work is not done by algorithms. It's quietly being completed by workers on micro work sites. So the manager of a taxi firm would usually have to supervise a team of drivers,

making sure, for instance, that they're safe to be on the roads, that they are who they say they are. As numerous scandals have suggested over the last few years, this is not something that Uber has done a very good job of. Part of the problem is that its identification software is really prone to errors and has been more so than other taxi platforms bad at judging whether the driver is who they say they are. The way that the company has dealt with this is by sending a validation task to a micro work platform. So a worker on a platform like Clickworker or Appen will be given, say, 30 seconds to decide whether the face of an Uber driver on shift matches their photograph on record. If the worker decides yes, then the ride will go ahead. If the worker decides no, then the ride will be cancelled. You can see here how a task

that would have once been done by a manager or supervisor is not actually automated, but simply outsourced as micro work. So in effect, the function of the manager is dispersed among algorithms and a crowd of precarious workers. This might lead to the driver being kicked off Uber. So in that sense, the micro worker has also, in this very perverse sense, sacked the driver. It's so shocking to see that example and to understand the way that that actually works, right?

Because it's not the way that we're told that it works and how these systems are supposed to function because the micro worker behind it is hidden away. So we don't know about that aspect of it. And now I want to shift our lens to the global south as well, because, you know, obviously, naturally, things are working a little bit different there because of how labor works differently, how the economy is set up differently. And again, you know, that obviously varies by country, it's not to say that it's a monolith. But the expansion of internet connectivity and the platform economy have often been promoted on an international scale as a way to extend economic opportunity to other parts of the world. But given the work that you've done on this book, what does that opportunity

amount to for people in the global south? Well, it should first be said that most micro work is done in the global south, the majority of it is, and it's actually developed to the point of effectively creating new labor markets in many countries, but not in any way that we should applaud. The veneer of respectability micro work has managed to garner the sort of the ways in which it's managed to sort of garner applause from the establishment of some countries, at least I argue in the book, is the result of efforts by institutions such as the World Bank. So back in the early 2010s, it started to push micro work as a kind of like an all-purpose solution to employment in poor countries around the globe. And this phenomenon wasn't singular, it happened in tandem with the rising number of NGOs starting micro work programs in spaces like refugee camps and slums, often with ominous mottos like give work, not aid.

In a spirit, again, I argue in the is a kind of contradiction in terms, considering outsourcing is always supposed to disproportionately benefit the contracting firm. I'm not sure what kind of positive impact that could really have. The first of the companies, to my knowledge, that started doing this work is a company called Sama, which recruits workers in Haiti, Pakistan, and I think also Ghana as well, often refugees or those who dwell in slums. And they'll complete data training projects for large tech companies such as Google and Facebook. The company started off like a lot of these impact sourcing organizations as a not-for-profit, but it's now for-profit. And despite its dubious activities, it's had a lot of positive

coverage in the business and technology press for redefining global development. As I earlier mentioned, its motto is give work, not aid. Now, while many of these companies and the World Bank and many of the impact sourcing organizations make claims to pulling people out of unemployment and the informal economy. Microwork, in many respects, differs little from the most abject kinds of informal work. This is something that I'm very keen to emphasize and is a key part of the book. I'd use in the book the work of the sociologist, Jan Bremer, who's written wonderfully and very lucidly on the informal sector in India. And he has this term, wage hunters and gatherers, which describes people in cities like Mumbai and Kinshasa, who work very, very much at the margins of the capitalist system,

hawking goods and shining shoes, moving from petty task to petty task as the market dictates. Microwork, I argue, is in many respects a kind of digital equivalent to this kind of work, in that workers often spend more of their day hunting for tasks than actually doing paid work. The majority of these workers coming from the Indian subcontinent, from East Asia, Latin America, have few formal protections, will perform multiple services for multiple contractors over the course of a single day, and will remain continually uncertain whether they'll find enough income that day to survive. So this idea pushed by the World Bank and by NGOs and some of these tech

platforms that micro work is kind of the phoenix of the South is this sort of this way of saving workers from destitution. You know, it's patently false, as you can see from the sort of the conditions that the workers labor under. For countries in the global South who have kind of a growing micro work industry, what does that actually mean for countries where a lot of workers are doing this kind of work? Because I guess the companies that they're working for won't necessarily be based in the country. And I guess what does that mean for paying taxation and things like that? I guess, is that a beneficial form of labor or work for them? Or is that just another way to extract the labor and wealth of these countries for the benefit of these large Western firms?

This is something that I've thought about and I've kind of struggled to find a decent answer for. And it's a question that kind of hovers over outsourcing under capitalism more generally. I mean, part of the reason I would imagine why these programs and why micro work platforms more generally have been so popular in the global South is that they are partly a way of massaging down unemployment figures and also their evidence that governments are making an effort to bring workers out of the informal sector and into the formal, I say formal here in quotation marks, into the formal economy because like work often doesn't really have formal protections and it certainly doesn't deserve to be called employment as such. So that's one reason. I think it's difficult to make a case for these companies really benefiting

the national economies of countries such as Kenya, because of course, as you say, all of the value is being extracted by companies that exist in Silicon Valley. So I'd hesitate to say that there's an economic benefit here. I would probably emphasize the political benefits in terms of parties being able to make the claim that they are finding employment opportunities for their workers. This is certainly part of the World Bank's big argument is that micro work offers micro employment opportunities. I think that's fascinating. And I appreciate you outlining that. And just to provide the listeners one example, I guess, you talked about in the book how Tesla has actually benefited a lot from using Venezuelan workers for its systems was pushing something like a million percent,

millions and millions of the Venezuelan middle classes were effectively made unemployment very, very rapidly during that period. Microwork is not obviously well known in a lot of countries. What tends to happen is that it will become popular amongst a few people and then it will spread by word of mouth over WhatsApp. So what happened during the crisis in Venezuela is that lots of people suddenly started to get these messages in their WhatsApp saying new employment opportunities, et cetera. And the companies that were behind this were Scale, Hive, and Mighty AI, three companies which have a significant client base among autonomous vehicle developers. So that would be like Uber and Tesla. So large numbers of the Venezuelan middle class ended up doing incredibly poorly paid

work on these platforms, as in it would amount to around about a dollar an hour. And the income that you could make on these sites got worse over time. So the more workers that arrived on the platforms looking for employment opportunities, the more these platforms could lower wages. the more workers that arrived on the platforms looking for employment opportunities, the more these platforms could lower wages. And so it got to the point where actually lots of these workers just left the platforms because they presumably could make more money in the informal sector. Wow. You know, these autonomous vehicle companies that have not achieved the goals that they set out that they said they were going to achieve with these vehicles everywhere, you know, are still

benefiting from these really poorly paid workers who are in a really difficult situation to develop their systems that we're told are just the computer doing everything, right? So it's just yet another example of this. But now you write about a potential future where work exists less to facilitate labor than to feed these machine learning systems, as you're kind of talking about, right? And you make the case that these systems will push monetization and commercialism into every waking hour of our lives, which sounds like a total dystopia, even though it also seems like we're totally headed that way, and we're already kind of on that track. Can you expand on this and what it might mean for our lives, given we seem to be in the midst of another push along these lines

with companies championing things like Web3 and the metaverse? So that's a multi-pronged question. So to speak to the first point, micro-workers are a clear example where a worker's main role is to feed machine learning systems. So micro-workers are not only showing machines how to do the jobs of other people. So an example here would be showing a computer how to do a taxi driver's job, but they're also showing machines how to do their own jobs. That is directly showing machine learning how to do the specific task they are currently undertaking. So part of the micro worker's role, we can sort of glean from this, is to directly automate their

own job away. So often the data about how the task is done is as important as the product of the task. So micro work sites collect this data, which can then be readily sold to larger tech companies, presumably companies like Google and Facebook, who it should be emphasized have been pretty explicit about their desire to automate, say, content moderation tasks. So it's a strange scenario where in some cases, what we're seeing is workers might be effectively sort of simulating their own jobs to show machines how to do it. Amazon Mechanical Turk is pretty explicit about this. That is, if you take the time to take a look over their terms and conditions, they effectively say that any tasks completed on the platform, the data from

that task can be used by Amazon for its own machine learning purposes. So it makes perfect sense, to my mind at least, that the real reason why Amazon took the Mechanical Turk platform public, rather than keep it as an internal service, was not to take a cut from transactions between workers, which in the grand scheme of Amazon's operations does not seem like a particularly profitable venture. The real reason was to gain wider access to a range of once unavailable data about the ways in which particular tasks are done. We can take this a bit further and see that the vision these companies have, you know, whether this vision ever actually comes true is another thing, but the vision these

companies have is one where labor is entirely sidelined. But just because capital cannot create the conditions for the social reproduction of labor, or refuses to create the conditions for the social reproduction of labour, it does not mean that labour no longer needs a wage. To live under capitalism as a worker is to need a wage. Under pains of survival, you must search for one. So as work becomes increasingly marginal to a system no longer creating proper jobs, and this is something we're already seeing, I don't think this is necessarily just something in the future, we find that work has come to permeate the entire social fabric as workers sort of desperate for income are forced to turn all of their time and activities into monetizable activity. So this links in then to your final point about the metaverse,

whatever that turns out to be, probably not what tech tycoons imagine. But one thing we can be certain about is that it will undoubtedly provide all manner of opportunities to turn our lives into monetizable activity, activity which is fundamentally profitable for tech companies. So this might take all manner of forms, which I could see being based on the microwork model, but applied to other jobs and sectors. You can see, for instance, how VR might allow for what I've elsewhere called migrant labor without migration.

We've seen kind of crude renditions of this already, where poorly paid workers living in Colombia remotely control delivery bots in campuses in California. VR carries the potential to untether all kinds of once physically tethered work from its geographical situation. And I think it's worth adding that this could be combined with a micro work model where you can simply strap into VR and go and do a short task, which might involve remotely controlling a robot that exists 2,000 miles away. Absolutely. What you're describing makes me think of, I don't know if you've seen it, but Alex Rivera's science fiction film, Sleep Dealer, where that is effectively the premise,

right? Workers in Mexico are wearing these VR rigs and things that they can control with their hands to control robots over the border in the United States. So they don't need to actually have the migrant labor anymore. They can stay in Mexico and still do the work that is needed to keep the United States running. So I think that's kind of a dystopian vision of something that could potentially be in the future. But as you were mentioning earlier, following the automation narrative in the 2010s, many people made bold predictions about how automation would enable us to live lives of unparalleled leisure, because robots would start doing all the work for us, right? Like automation would enable this society where we don't need to work at all. This view was even championed from the left in visions for fully automated luxury communism,

for example. But you outline how there are often poorly paid people living very precarious lives behind what looks like artificial intelligence and automation. So what do you make of these future visions? In many respects, I don't disagree with the principles behind these visions. In a socialist or post-capitalist world, whatever way you want to describe it, you would surely want, if possible, to automate away some mundane rote labour. This, of course, would need to be democratically determined. What is bad and unenjoyable for one person provides satisfaction to someone else. But there's certainly a great deal of wage labour that I'd never want to do, and I'm sure many other people would agree. It might also be good, for instance, to automate

some work outside of the wage, such as housework. And while I agree with this in principle, the problem is that automation as it stands does not mean the total eradication of work from a particular activity or task, but rather the removal of one set of tasks replaced by another set. Full automation is a useful demand in that it inspires the utopian imaginary. But in reality, it's not often possible. Certainly in the context of AI, it doesn't seem possible. It seems like the promise of AI on both the left and the right has been this fully automated future, which it strikes me now is evidently kind of false. So just to return to the example that I give in the book of Uber, which we spoke about earlier, we can see here how an automated role and an automated role that we would probably want in a post-capitalist or socialist society, which is a taxi service, still relies on a significant amount of low-skill labor.

It still needs somebody to check photographs of who's on shift. It needs someone to make decisions about the workflow if an algorithm kind of trips up in that respect. So the idea that we could have this sort of fully automated luxury society where taxi drivers are smoothly taking passengers from their home to a nightclub on a Saturday night without any workers involved, and that we all just live this sort of life of leisure,

it's a useful and I think a decent demand, but I think in reality, it forgets the fact that automation is not simply a technological process. It often involves hybrid forms of task fulfillment, which rely on machine learning and poorly paid workers. That's such an essential point, right? To recognize, as we've been talking about through this conversation, how there are a lot of poorly paid precarious workers who are behind many of these technologies that we assume are just this new advancement is able to perform this task in a way that we never expected technology would be able to do. But as you say, you know, there's Venezuelans powering Tesla's self-driving

software. There are Appen workers who are approving the management tasks of Uber. There are Colombians and Filipinos driving the little delivery robots that we're increasingly seeing on the sidewalks. So it appears that there is this like growing degree of automation within our society. But actually, behind the scenes of that, there are all these poorly paid workers and then building a future on top of that, assuming or not taking into account the workers who are actually behind that seems like, I guess, really the wrong way to approach things. And of course, into capitalism, there is a seriously dark side to this. Many of the

technologies that workers are in the background powering, we wouldn't want in a socialist society. So for instance, I can't really see the use of a facial recognition technology under socialism. But under capitalism, in the present world that we live in, this has some really seriously bad implications for workers, particularly workers on micro work sites. So let me just give you a bit of an example here. Many of the face tagging tasks on micro work sites are used to train facial recognition technologies. And we know that the technology contains eugenicist logics and produces highly racist results, often targeting black and brown people as criminal suspects. The technology is increasingly popular as a police strategy.

We've seen it rolled out over the pandemic in the US and across Europe and indeed in China. So the LAPD, for instance, has used the software, I think I've read somewhere, I could be wrong on this, it's 20,000 to 30,000 times since 2010. And during COVID, we've seen the use of the technology grow rapidly across the globe. But the tasks aren't labeled with any information about how their products will be used. So we have facial recognition technology growing, and we know that there are these serious kind of ethical problems with it. Yet workers on micro work sites who are making this technology operational do not know that they are working on these specifically

oppressive projects. So for instance, the worker doesn't know whether they're helping an AI algorithm used by the LAPD. The tasks are entirely opaque and the institutions and bodies that use the technology will often secretly contract it from companies like Amazon and Google. And then of course, Amazon and Google will struggle basically to produce the kinds of machine learning technology that can do this work automatically and will require workers to label images of faces to show the machine learning technology how to do the work. So that then means that the tasks are outsourced onto platforms like Mechanical Turk, App and Click Worker. And the thing is, unless a contractor, usually one of these big tech contractors, decides to give a worker the details about the task they're working on,

the worker has no way of knowing whether they're helping to produce an oppressive technology. Yeah, there's maybe a solid argument to say that this has always been the case under capitalism, that workers are, by their very role in the system, unwittingly creating projects that subjugate to the profit motive or racist technologies or whatever. But I would argue that microwork actually represents a significant shift in the knowledge available to workers, a kind of significant diminishment in the knowledge available. So you can think of plenty of theorists in the Marxist tradition. They've shown that to some degree, the worker's exploitation must remain unknown to the worker. It's not something that's

visible or that most people are conscious of day to day. And this is a good thing for the capitalist owner, otherwise the workers might rise up. With microwork, though, not only is the exploitation hidden, but also the nature of the work. So workers are robbed of the tiny bit of agency they have to decide whether the work is something they wish to do on a basic moral level. I think it's probably fair to say that even those at the furthest reaches of global supply chains will have some idea of what they're working on, even if they don't know who they are working for. So I guess the point I'm making is that people who use micro work platforms, they're very often robbed of the capacity to know precisely what it is that they're even working on. As you explain there, and as you write in the book,

like the workers on these platforms might be helping to power the very technologies that are being used to oppress, deport, or police them in various different ways. In many instances, unknowingly, they won't know who they're actually doing the work for. But I wonder, do you see opportunities for resistance to these technologies by these

workers? And are there particular forms of that resistance that seem most promising to you? So far, micro workers have struggled to organize on these sites for all of the reasons that we've discussed so far. They are a disempowered workforce. Often the contractors that they're working for are opaque, so they don't know who they're working for. So it's very difficult to oppose a contractor that you can't even see. It's also quite common for

workers to receive or to be forced to sign non-disclosure agreements for their work, which means that it's difficult to talk to others, particularly other workers, which is an essential part of creating sort of collective action. And the instances of collective action that have emerged over the last sort of decade have tended to be shut down quite quickly. So there was a letter writing action, which was basically workers on Mechanical Turk to try and get themselves greater visibility, wrote public letters to Jeff Bezos. This was actually quite a successful action in that it got some journalistic eyes on Mike for work. But the forum which organized this action closed down soon after

because it relied on Mechanical Turk to validate for the forum who were real workers and who were not workers, so who were effectively Pinkertons going on there to sabotage organization. So they relied on Mechanical Turk for these validation tasks. As soon as Mechanical Turk realized what was going on, it basically just shut down the accounts that were doing these validation tasks on there. So it was impossible then for the forum to validate who were real workers and who weren't. The most successful form of collective action that we've seen again on Mechanical Turk is TurkOpticon. So just to reiterate a point from earlier,, on Mechanical Turk is TurkOpticon. So just to reiterate a point from earlier, on the Mechanical Turk platform, workers are given no way to review

contractors. So contractors can review workers, but workers within the platform can't review contractors. TurkOpticon is effectively a browser plugin, which allows workers in real time to look at worker reviews posted already about contractors who use the platform. So basically they can spot dodgy contractors in real time, warn other workers about dodgy contractors that are operating on the platform at that time. But it should be emphasized that for all of the reasons to discuss throughout this podcast, it's super, super hard to organize these platforms. I think that greater possibilities for collective action might require forms of solidarity with other sections of the workforce. So in the book, there are basically two potential axes of alliance

that micro workers might pursue. The first is with other tech workers. Those who have sort of full-time positions in companies such as Google and Facebook, who might be sympathetic to their cause or recognize shared interests or the potential of sort of strategic alliances. We've already seen hints of this emerging between Facebook employees and its outsourced content moderators. There's certainly nothing solid. So when Facebook employees walked out because Facebook hadn't taken sufficient action against Donald Trump, content moderators wrote a letter of solidarity to Facebook's employees, basically saying that we are with you, we would also walk out, but for the fact that it's impossible for us to do so, partly due to NDAs, but also because there was the risk that they would be simplified

on the spot. It would be nice, of course, to see solidarity extended in the opposite direction, from those more privileged employees at Google and Facebook to the outsourced workers on micro-work sites or content moderators. That's what I'd like to see. Since writing the book, I've started to think that this might become more common, particularly as employees at tech firms are becoming increasingly politicized. And I do think it is part of the writer's or podcaster's job is to give micro workers the visibility that puts them on the radar of employees at these companies. So I've had a few emails over the last few months since writing the book from people working at

these companies who are really astonished at what is going on. Anyway, that's a bit of a tangent. The second alliance that I think that micro workers might pursue, and the one I discuss in more detail in my book, is with other precarious and surplus populations. So historically, there have been unemployed movements, and there still are some, which organise those who are either out of work or intermittently out of work or precariously employed. Most of these workers have historically met in physical space. And as we can see with Deliveroo strikes over the last few years in the UK and in the US with Uber Eats, these strikes often rely on workers meeting each other face

to face. The problem with micro work is that it happens remotely. So workers rarely meet each other. One thing that might make such action more likely, I argue in the book, is providing facilities so that micro workers can meet in physical space. So one of the solutions I suggest is the worker centre model, which exists mainly in the US as a way of bringing together migrant workers who would otherwise be fragmented and isolated, doing bits of day labour here and there, but rarely actually meeting for long enough to organise action against employers. So these centres would be set up in urban centres where there are great numbers of active

micro-workers and would provide a space for them not only to work so that they're surrounded by other workers in the day as opposed to just remotely working in their homes, but also a space to meet other micro-, to discuss the kinds of problems that they have, to talk about any gripes that they have, which is the sort of the first steps towards collective organizing. Hopefully this podcast will wake some tech workers up to what is going on. And as you're talking about the worker centers there, one possible example that came to mind to me was when I spoke to Rila Khadri about gig workers in Indonesia. And in Jakarta, they have like little workers centers, I guess, or like sites that they set up that are controlled by the workers to all meet together and like have that collective

relationship and to support one another. And it seems like a really clear example of something like that, where these workers can come together, get to know each other, find their common interests, etc, etc. After discussing so much about how this work works right now, under capitalism, the problems with it, I would like to end by discussing maybe the future or a more positive future. Because at the end of the book, and I found this really intriguing, after breaking out many of the problems with micro work, you called back to the Paris Commune, to William Morris, to Marx and Engels, to argue that it could actually present a path to a better future. So to end our conversation, I was hoping you could give us an idea of what you think that future could look like. So though microwork is, as we've just discussed, fragmented, opaque, it comes without rights,

the pay is often terrible, if indeed workers are paid at all, we can also see in its promise the lineaments of a better world of work. So what I wanted to do in the final section of the book was as a kind of utopian thought experiment, consider what micro work would be like if undertaken outside of the wage relation. Was there, for instance, any what micro work would be like if undertaken outside of the wage relation. Was there, for instance, any qualities that we would like to see in a post-scarcity society? And micro work might, to the listener after this discussion, seem a somewhat strange place to conduct such thinking. You know, the monotony of the tasks, the lack of routine, the insecurity, the opacity, hardly seems to suggest a kind of post-work Eden. But the promise that

Microwork makes of an independent, flexible and leisurely working life, I think should be taken seriously. For one thing, a day spent working on these sites will mean doing a multitude of different tasks. I argue that this offers a kind of perverse rendition of Marx's idea of communism, where you fish in the afternoon, rear cattle in the evening, and criticise after dinner. Microwork, of course, only offers a warped, hyper-capitalist version of this promise, but it does demand that we rethink the rigid boundaries of occupation, particularly when so much work can be decomposed into small tasks. This offers a vision of a world where we each do lots of different jobs and tasks, or maybe we have

a part-time occupation and do lots of other shorter and tasks, or maybe we have a part-time occupation and do lots of other shorter jobs. A post-scarcity world would require people trained in specific vocations, but these specific vocations needn't be their primary occupation, or indeed take up all of their time. So at root, I guess, micro-work can evoke a world where there is greater flexibility and people can do more of what they want to do. I also argue that micro work sites often reduce paid labor to a minimum and that the worker spends more time hunting for jobs than doing them. And I also think it's worth pointing out that this hints at a world where labor is less necessary and we all work less. So hidden below the bad stuff we associate with bike for work,

the precariousness, the low wages, somewhat ironically is a shorter working week. That is less time spent undertaking social labor. And like many have recently argued, machine learning algorithms could be used to calculate and distribute work, not in ways that intensify productivity, but in ways that privilege free time and autonomy. I also argue that machine learning algorithms could be used to distribute work, the ways that privilege free time and autonomy. I also argue that machine learning algorithms could be used to distribute work in ways that suit individual desires and capabilities, instead of simply being used for labor arbitrage and hunting down the most desperate workforces. Similarly, gamification on these sites, so returning to the incredibly draconian ratings and ranking systems that we discussed earlier, we can see here that these rating systems could in some sense, you know, succeed the wage relation

as a means to motivate due effort. Though perhaps not the best model, one that springs to mind immediately is the Soviet Union, which often substituted competitions for the wage system to make sure that workers completed what was socially necessary. Today, you could think about digital games holding a similar potential. I think it's also worth noting here that what I'm trying to do with this thought experiment is also reveal a paucity of imagination in Silicon Valley. I actually think Silicon Valley represents capital's failure of imagination, except in the realm of exploitation. Many of its technologies, forms of organization, you can see how they might be used to support a better

world beyond the wage. And while such technologies can't, of course, make that world happen, and indeed, as the title of this podcast suggests, technology alone is not the answer, but we can see that, you know, used more rationally than under capitalism, they could help support a post-scarcity world where we work less and do more of what we enjoy. Yeah, I really like that vision that you laid out at the end of the book. And I particularly like that criticism of Silicon Valley that you're ending our episode on. Phil, it has been fantastic to speak with you, to learn more about the book and the work that you've been doing. Thank you so much. It's been a wonderful conversation. And I've been asked lots of questions today that I've

not been asked on the podcast. So it's been, yeah, it's been for me a real treat. Phil Jones is the author of Work Without the Worker, Labour in the Age of Platform Capitalism, and a researcher at the Autonomy Think Tank. You can follow him on Twitter at at PhilJones7771. You can follow me at at powersmarks and you can follow the show at at techwontsaveus. Techwontsaveus is part of the Harbinger Media Network and you can find out more about that at harbingermedianetwork.com. And if you want to support the work that goes into making the show every week, you can go to patreon.com slash techwontsaveus to become a supporter. Thanks for listening. Thank you.