- Hosts in the Machine
When capitalism imagines itself it dreams of machinery. It sees the spinning jenny and not the highland clearances. It sees the laptop and not the plant in China. And now, in the era of cloud computing and the internet of things, we are invited to believe that all that is, is air, and to talk of solids melting is an anachronism. But the internet isn't air, it is made of metal and plastic, and the work that it does is most often done by people. This basic comprehension of how things really work is the starting point of Trebor Scholz's latest book. Uberworked and Underpaid is a survey of the twenty-first century economy with the labour left in, and, from this foundation, it provides some of the most important insights that the field of media scholarship has produced for a long while.
Throughout the work Scholz emphasises how far technological mysticism distorts labour, law and life. The Amazon Mechanical Turk (AMT) is a key example here; a virtual labour exchange which recruits thousands of workers from the US, India and Brazil to perform minor tasks which users on the receiving end of the transaction would consider automatic. For instance, the 'Amazon Remembers' service invites a user to upload a photo which is then identified and located for them online. The work is done by a worker, or 'Turker', who is then 'paid ten cents' (p22), while the user believes the search to be the work of a machine. The software Soylent works the same way only for spelling and grammar checks. As someone who has worked for an online copywriting company, this is the machine that I, as a reviewer, recognise. Everything on the internet is done by someone, but usually not the person you expect. Exploitative labour practices become slightly more visible when human interaction is part of the service; see Uber, TaskRabbit or - in the UK - Deliveroo. The sleight of hand, however, is the same: hide the worker from the customer through an automated app, and hide them from the government by calling them 'independent contractors' rather than employees. As Scholz astutely comments, this is not happening now because of new technology but because of new social conditions: these business models are 'reliant [both] on the availability of an abundance of cheap labour and a permissive regulatory environment' (p44). Amazon Mechanical Turk, notably, was not only deemed too exploitative to operate in Europe but also in China where workers are required at the very least to be paid in money, rather than Amazon gift vouchers.
Rather than present an anti-exploitation polemic, however, Scholz is [End Page 118] also careful to present the voices of the workers themselves. In the case of AMT Turkers these voices reveal as many positive responses as negative ones. Working from home and flexible hours are cited as benefits, plus the inclusion of gamification principles into interface design makes it possible to transform some mundane tasks into compulsive, Candy Crush-esque puzzles. Scholz mentions a number of these game-like softwares beyond the AMT which incorporate user feedback to hone their system's overall accuracy. Some of these gamified systems are well-designed enough that users contribute to them for free. One example given is the Google Image Labeller, which pits two players against each other listing keywords when presented with images, the aim being to get a match. If both players input the same keyword they both win; the players get points and Google gets a verified keyword to feed back into Google Image Search. Another was the reCAPCHA human verification system which, as you may have noticed, has recently developed from text identification to image identification and utilises the same user feedback principles. Although it was released too late for inclusion in Scholz's work, his analysis of how play is turned into labour is typified by 2016's Pokemon Go phenomenon. A re-skinned version of the game Ingress, Pokemon Go is a way for Google to track and map pedestrian...