Sadowski’s book based on the This Machine Kills podcast is an incisive and important demolition of illusions around technology and AI, argues Kevin Crane

The big turnaround of the socialist left’s ability actually to confront and criticise the tech sector came frustratingly late to the game. It was basically not until the Covid-19 pandemic was firmly upon us that there was really much out there. Huge developments like the rise of social media were already old-hat and international capitalism had seen even the fossil-fuel giants largely replaced by electronics and software companies as the biggest in the world, yet there was a real dearth of serious left-wing political writing or journalism about it all.

For well over a decade, commentary around new technologies was absolutely dominated by sycophantic nonsense from rabidly pro-business ‘thought leaders’, who frantically talked up every dubious claim and press release that was coming from tech companies, eager to both believe the dream and get invited to lavish big-money junkets. For many people, the only counter to these narratives they got to hear was simply a different type of bullshit: confused conspiracy-theory driven nonsense that simply spread superstitious paranoia about technology.

An infamous expression of this was the idea that 5G mobile phone masts are a super weapon … or a mind-control device … or something even more vague and menacing. This stuff was all gibberish, but it gained traction because people had a sense that technology was simply in a constant state of change and transforming society around it, with no real discussion taking place about what was happening and whether it was right or wrong.

So, in the summer of 2020, when lockdowns had imposed even greater dependence on internet-based platforms in to so many aspects of our lives, the podcast This Machine Kills launched as part of a new wave of left tech analysis, and it was thoroughly refreshing. Jointly hosted by the Marxist academic Jathan Sadowski and the anarchist journalist Edward Ongweso, the series has tackled a wide range of topics on and around the tech sector, the business interests that run it, the workers that produce it and the consumers (willing or not) who have to live with it. Certainly, I can recommend the podcast to readers of this website who have an even passing interest in technology, but another option is to check out this book.

The Mechanic and The Luddite: A Ruthless Criticism of Technology and Capitalism is Sadowski’s distillation of many of the key ideas he has been developing over the years on the podcast. Although he would doubtless be able to write much longer deep dives into many of these concepts, he has instead opted to keep things short, accessible and easy to dip in and out of. Each of the chapters is written to function as a separate essay about a particular talking point: like the role of data, how big tech specifically exploits labour and how the sector’s domination has been controlling our very ability to discuss the future itself.

Disentanglement and demystification

One of the things the author excels at is taking things that have been deliberately obfuscated and making them understandable, which is really useful when we’re talking about the digital economy. Alienation, in the Marxist sense of treating human-made products and processes as if they are supernatural phenomena beyond our control or understanding, has abounded in the internet era. The extremes of this are expressed in the way that many Big Tech leaders, notably OpenAI’s Sam Altman, often refer to their products as literally god-like in their potential power.

A more prosaic but important example of this is the concept of data, which is discussed in an almost entirely inane way in the mainstream of our society. At some point in the past couple of decades, the idea that data was metaphorically a resource entered the popular imagination, only for the point that this is a metaphor to get conspicuously lost somewhere. A cliché is sometimes defined as a turn of phrase that terminates thoughts, and ‘data is the new oil’ has certainly filled that function. This leads to endless discussion about accumulating, trading and using data that takes no account of the fact that it is a human-made product, consisting of discrete and uninterchangeable base units. In turn, the value of data is constantly calculated incorrectly.

On one hand inaccurate, biased, spurious, out-of-date or otherwise invalid data is often packaged up into massive structures without being identified as such, becoming overvalued data. On the other hand, big companies ‘forget’ to check that they aren’t stealing protected personal data or even copyrighted information belonging to other businesses, essentially undervaluing it. Ultimately, all this stuff is getting force-fed into bloated ‘artificial intelligence’ and machine-learning models, with the current consensus among tech-company bosses being that if they just keep going bigger and bigger that they will somehow cheat-code their way past the iron ‘garbage-in, garbage-out’ law of information systems. So data we are told to believe is ‘precious’ ends up being treated as absolute junk. Increasingly, the data being thrown into AI models is itself produced by those models, creating what Sadowski calls ‘Hapsburg AI’, which is an effective metaphor I prefer not to think about for too long.

When it comes to both workers and consumers, data plays an absolutely key role for capitalists that is both economic and ideological: ‘datafying’ things reduces them to objects that are merely collections of signifiers, and this is a fantastic way of depicting systems as containing no actual subject that is actively doing things or making choices. They do in fact have subjects, of course, and the objective for people like us should be to point this reality out. Sadowski’s arguments really help with this.

On the most basic level, he breaks down a lot of the nonsense that’s been building up over the years around algorithm-driven work and the ‘gig economy’, which he describes more accurately as ‘servant apps’: a technological veneer to disguise the fact that you are simply buying people’s labour from a gang-master. There is then a slightly more advanced concealment of labour that he calls ‘Potemkin AI’, which is where the technological veneer intentionally creates ambiguity about where it ends, and the human worker begins. Working within one of these systems is both absurd and draining, such as the now-widespread practice of employing a team of people to train a chatbot. The robot should be mimicking the staff, but it goes awry and forces them to mimic it: doing an impression of themselves. This is objectively stupid, but it reflects the contradictions involved in so many of these systems as commercial operations.

Pointless struggles that never end

The above leads us to a recurring theme throughout The Mechanic and The Luddite: the way that capitalists try to use technology in erratic ways because they are so often struggling for objectives that are either unachievable, or that would be catastrophic if they achieved them, or both. In the specific case of Potemkin AI, the bosses are straining towards the fantasy of producing capital without labour. This can’t be done for material reasons, as even the most expensive and highly-advanced models and platforms need some human input because you can’t rely on them otherwise. This means they are in the end just vessels for labour – highly advanced tools, in other words – but also this is fundamental to how the whole process generates wealth at all. Capital is ultimately only accumulated surplus labour: if labour were rendered worthless, so would it be!

Fantasies are a powerful motivating force for today’s capitalists, however ridiculous they might be. In the chapter on risk and insurance – which Sadowski powerfully argues is the most unwisely overlooked aspect of modern capitalism – he points to the dynamics of a business sector that seems to hate the purpose of its own existence. An insurance scheme is, ultimately, resources that have been pooled to help members recover from a shock situation. A perverse incentive of sorts has arguably always existed for insurance providers to exclude (or simply penalise) people and groups with increased risk of negative outcomes, but this has become a mania in the era of datafication.

Hyper-estimating who and what is a risk is surface-level feasible with today’s computational power, but it isn’t strictly speaking insurance, and even senior figures within the industry have publicly speculated that they are in danger of creating a situation in which you can only afford insurance if you don’t need it and cannot afford it if you do. This is all on top of very serious concerns that we could and should raise about the use of data as a source of discrimination with potentially catastrophic impacts on affected people.

Probably the most emotionally irksome example of tech capitalists wanting to have it all ways, however, is the question of investment. Just as insurance companies want to live in a crazy world where they never pay out, but still exist for some reason, venture capitalists (VCs) in the sector believe they are entitled to live in a world where every horse that they back turns out to be a magical unicorn. Listeners to This Machine Kills will be familiar with their demolition of the mythos around VCs, but either way it’s both entertaining and instructive to pick apart who this section of the bourgeoisie actually are.

Far from being extremely smart, or even lucky, investors, VCs are a caste of businessmen that the United States government intentionally fostered in the wake of World War Two to take the technology sector – born as it was out of state-investment during the war – into private ownership and direction. The VCs’ constant pumping of vast quantities of dollars into this and that project is enabled by a faith that the American state, via legislation and tax breaks, will only let them fall so far if things go wrong. So, we have investment without risk: it’s all reward, for nothing.

Since the pandemic, the VCs have ploughed money into a series of failed projects – the metaverse, cryptocurrencies, non-fungible tokens, now AI – in a desperate bid for a thermodynamically impossible perpetual value machine. This has allowed them to maintain the persistent and huge overvaluation of tech-sector businesses, but of course that is looking very unsustainable right now.

Landlords and propagandists

One of the most novel topics of discussion in the book by far is the chapter on rent and landlordism in the tech sector, which is actually two parallel topics: using tech for rent extraction and extracting rent from tech itself.

On the latter, it feels counterintuitive, but ‘the cloud’ is actually a physical thing owned by landlords, an arrangement intentionally created by the same set of conscious policies that gave us VCs. On the former, we get the gradual slide towards the ever-expanding pervasiveness of everything-as-a-service.

Turning things that were at one time products into services enables capitalists to establish themselves as rentiers, and it is very much the way everything is going as cheap credit and overvalued advertising sharply reduce incomes for Big Tech. It’s at this point that the author introduces one of the biggest points of difference between his thinking and a significant number of other writers on the left: Sadowski does not endorse the concept of ‘tech feudalism’ that has been suggested by people like the socialist economist Yannis Varoufakis. His reasoning is simple: rentiers are not actually foreign to capitalism; this is just something that is ideologically argued to make capitalism sound better than it is.

The Ur-example of a tech rentier, Bill Gates, is a case in point: he is not some overhang or remnant of a past order that glommed onto the emergent personal computer industry from the outside or the past. He successfully sold a disk-operating system to another business (the now mostly forgotten IBM), which he had bought rather than written, and managed to get them to pay him royalties because they just didn’t realise how much it was going to be worth. These are normal property deals and relations in capitalism; the fact that Gates got unusually rich doesn’t change that.

Sadowski argues that saying ‘tech is feudal’ is actually getting trapped in a position where you are simply responding to the right-wing libertarian argument that ‘tech is freedom’ with a counter that has reversed rather than beaten your opponent’s logic. The big danger for anti-capitalists accepting the existence of tech feudalism is that you actually end up reproducing libertarian arguments about ‘the real capitalism’ being good, because unlike the ‘feudal’ rentiers, it is dynamic, enabling and promotes efficiency. We can just look at the creaking overblown mess of AI and see that that’s got to be wrong.

Generally, the author feels that where the left needs to raise its game on these issues is to avoid ‘criti-hype’, which is what he says is an unfortunate tendency to accept the claims and promises made by tech-sector boosters and simply add the caveat ‘but it’s bad’. What he proposes as the alternative and antidote to such mistakes leads us back to the book’s title.

An alternative to all the failed utopias

Sadowski calls on readers to be both a mechanic, and a Luddite, in what he regards as the more proper original meanings of both words. A mechanic was not, until recent times, someone with a fairly specific job relating to maintaining motor vehicles. It was rather a person with a much broader understanding of machines, systems we might say, in a wider sense. The Luddite, meanwhile, is a figure that is much more commonly slandered than honestly talked about: capitalists pretend Luddites were mad barbarians who simply hated the march of progress. In reality, they were skilled and astute workers who became an armed resistance movement because they knew very well that the technologies being imposed upon them were not delivering progress that they would enjoy.

The title of the book is essentially a call for both good theory and good practice: understand how technology works, but also how it relates to capitalism and when to oppose it. The contents set out an excellent starting point for anyone who wants to do this, even if all you want to get out of it is solid arguments about relatively straightforward points of debate. A good example would be why AI, in the form it’s being imposed on us, is bad and we shouldn’t be afraid to say it’s bad.

The book is subtitled as ‘a criticism’ and not a manifesto, which is probably to do with the fact that the author did not wish to weigh his conclusion down with more concrete, but also, maybe, overly-specific, calls to action. There’s also been a bit of a debate on the left recently about teleology – the study of things via their ultimate ends – and how useful this really is. Sadowski is very critical of the way that the tech capitalists have been selling us all their own failed, and often idiotic, utopias to justify their actions. That’s given him pause for thought before potentially putting forward a utopia of his own, so the book ends in quite an open way. Nevertheless, this is emphatically a book about looking forward so we can present our alternatives to what capitalism has to offer. I am sure it will give the reader enough of their own ideas to do so.