Media scholars like Siva Vaidhyanathan and Shoshana Zuboff have argued convincingly that ordinary citizens and regulators should be concerned about the immense power Google has amassed over many parts of our lives. Yet Amazon, nearly as ubiquitous, and also a frequent target of critical press, maintains a much less troubled public profile than Google.
This should hardly be the case. Amazon’s role in developing disturbing new workplace trends, especially for non-white-collar workers, should be of central concern for labor advocates. While both Amazon and Google famously maintain resort-like “campuses” to recruit and retain top IT workers, Amazon relies on a workforce three times the size of Google, not including its army of contingent workers, for critical aspects of its business.
As a 2011 story about the company’s Allentown, PA warehouse reported, many of Amazon’s warehouse workers are temps employed by a third-party staffing firm, and it manages its warehouse workers using the same web-centric, piecemeal, “just-in-time” methods it uses for other aspects of its supply chain, enterprise planning, and customer relationship management.
According to a recent report in the Harvard Business Review,
Amazon will release a change [to its enterprise software management systems] about once every 11 seconds, adding up to about 8,000 changes per day. In the time it takes Staples to make one new [software] release, Amazon has made 300,000 changes.
Its software-heavy strategy relies on the use of enterprise-wide tools, similar to companies like Walmart. These tools, which Amazon is beginning to market (or help its partners sell) through its Web Services cloud-based platform, track and manage every aspect of the company’s business — from the location of products in the warehouse to the time it takes for warehouse workers to pick them out of their bins to the seconds spent by customers on a product’s web page (and the number of products left unpurchased in their shopping carts).
Amazon’s development of a wide range of labor and what might be called “quasi-labor” practices (including what Tiziana Terranova and others refer to as “free” or “immaterial” labor activities such as producing customer-based rating and referral systems) meshes uncomfortably with how Amazon micro-manages its customers as producers of value for its commercial enterprises. This is similar to Facebook and Google, though Amazon is rarely mentioned in the same breath as those paragons of immaterial labor.
A number of recent announcements and actions by the company put these problems into focus. All of them in one way or another show the company contributing to the casualization of labor, and to blurring the lines between private and public spaces and between labor and leisure.
The company has long been aggressive in its pursuit of alternative delivery strategies for its products, including perhaps most infamously its proposed use of delivery drones. In June, Amazon announced that it may incentivize people to deliver packages to their neighbors, whose schedules make face-to-face delivery with mail carriers difficult. There’s nothing wrong with some neighborly assistance, but, like Uber in its early days, Amazon is proposing to pay people for an activity that is typically thought of as an ordinary part of community life.
Most people are happy to take care of packages for their neighbors once in a while, and don’t feel the need to be paid for these things or even necessarily want to be paid for them. But it’s not a stretch to imagine capitalizing on this willingness by transforming people into entrepreneurial mini-distribution hubs, and depriving us of one of the few remaining opportunities for serendipitous, selfless interaction.
Another of the company’s recent innovations, Echo the “personal assistant” (also known as “Alexa,” the name given to the Siri-like voice that responds to user queries), raises similar concerns. Echo is an always-on listening device consumers willingly place in central areas of their homes. By saying the name “Alexa” (or “Amazon”), users activate the device, which is then capable of executing any number of tasks related to Amazon’s business, information generally available on the internet, or a limited range of media devices connected to the user’s home network.
Alexa will answer the kinds of questions whose answers might be found on Wikipedia (“Alexa, what is the capital of North Dakota?”), can play music, can manage devices to which it has access, and most importantly, of course, can initiate product purchases on Amazon itself.
In order to do any of these things, Echo must be listening all the time, to everything that goes on in its environment, waiting for the keyword to tell it to take an action. That means it isn’t just listening to and processing things users say after the keyword — it is always listening, always processing. Amazon has been evasive, at best, about whether or not it is collecting and analyzing that data and what it is doing with it.
The potential for data gathering has been noted by several writers, including Alex Hern, Todd Wasserman, and Chris Davies, and raises serious questions. Is Echo recording and analyzing, for example, the number of times members of the household mention a given book, movie, or video game; other brand name; or product category? Is that part of how Amazon plans to ship products to us before we have actually bought them? Will it use Echo to track our neighbors as they deliver products to us? Will it use audio data to determine how long members of the house sleep, how often they have sex, and how many non-family members come into the house?
Given Amazon’s profound interest in the behavior of its consumers, it is beyond question that that data would be useful, and given Amazon’s secrecy regarding its business practices, one can’t help wondering what the company is doing with it.
In June, just after describing its proposed neighbor delivery system, Amazon announced that it would be instituting a new payment system for authors who use its Kindle Select publishing platform (in other words, authors for whom Amazon functions as both publisher and distributor), according to which authors will be paid each time readers turn the page of a book.
As Peter Wayner notes, “instead of paying the most ambitious, long-winded authors for each page written, Amazon will pay them for each page read.” An analysis of the details by Hern concluded that “an author will have to write a 220-page book — and have every page read by every person downloading it — to make the same $1.30 they currently get from a book being downloaded.” Casey Lucas estimates that author royalties could decline by between 60 and 80 percent.
After attempting to collaborate in 2010 with Amazon on a publishing venture, well-known literary agent Andrew Wylie compared Bezos and the company to Napoleon. Wylie called the company “megalomaniacal,” and argued that its publishing business is not “an effort marked by sincerity” but is instead designed so that it “can be misperceived by the Department of Justice and the publishing industry in a way that is convenient for Amazon’s bottom line.”
Indeed, the Department of Justice investigated Amazon as part of an inquiry into ebook pricing that ended in a 2012 settlement with three major publishers; some allege that the investigation showed Amazon engaged in predatory pricing.
This intervention into the system of compensation for creative work recreates the sort of vertical-monopoly business model that resulted in the infamous 1948 US v Paramount Pictures Supreme Court decision, which interpreted the 1890 Sherman Antitrust Act to mean that one company could not own the entire production and distribution channels for a product category.
One of the features of recent digital capitalism is the tendency for firms to build companies that appear to skirt around the spirit, and perhaps the letter, of the law regarding vertical monopolies (for example, some have argued that Facebook’s “internet.org” initiative might have vertical-monopolistic consequences). Unsurprisingly, the Koch-funded Mercatus Center insists that regulators should keep their hands off.
Also in June, Amazon announced that it would increase its commissions from 10 percent to 20 percent on the Human Intelligence Tasks (or HITs) on its Mechanical Turk platform. Mechanical Turk (or MTurk) has long been considered the ne plus ultra of the gig economy. Its name is a deeply unfortunate, though revealing, reference to the chess-playing automatons of eighteenth-century Europe, a parlor trick concealing small human beings who actually did the work purportedly done by machines.
As historian Ayhan Aytes notes, these automatons were dressed in “Oriental” garb in part because everyone to the east of Europe was understood to be “docile” and “soulless.” MTurk allows employers to design tasks that require large amounts of data entry and analysis that, for whatever reason, currently remain more efficiently or more accurately done by human beings than by computers. MTurk allows task designers to set the prices for these services, but in general they are almost unbelievably inexpensive: in fact the cheapness is a major part of their appeal. Amazon claims that more than five hundred thousand workers are currently registered with MTurk.
MTurk is the target of criticism both from Turkers themselves and from outside observers. While independent MTurk support sites like Turker Nation argue that with enough time investment it is possible to earn at least $8–$14, and possibly $25 or more an hour on MTurk tasks, it is nevertheless the case that most tasks listed on the site pay an extremely low wage, averaging below $5 per hour.
That number also assumes workers actually get paid. Media scholar Trebor Scholz writes that “wage theft, while not explicitly tolerated by Amazon, is a daily occurrence” and that “wage theft is a feature, not a bug.” In the past, surveys have indicated that the majority of US Turkers are women (though that may be changing), and that the two countries with the most MTurk workers are the US and India.
In response to the widespread exploitation and abuse of Turkers, scholar (and former Googler) Lilly Irani and information scientist and trade unionist Six Silberman built a browser plug-in called Turkopticon, which provides panoptic surveillance of projects and those offering them, with an eye toward making it easier for Turkers to find projects that pay a reasonable wage and whose sponsors have a track record of paying their bills.
Innovation in Exploitation
Amazon’s core asset is its mastery of distribution channels, for which it relies on a network of enormous warehouses located all over the world. Conditions for workers in these warehouses are much worse than those of its software developers and business managers, and as a result, the warehouses have repeatedly been the target of exposés, protests, and unionization attempts.
In 2013 the BBC program Panorama sent an undercover reporter into Amazon’s Swansea, UK warehouse. The journalist wrote that he and the other “pickers” who collected items from inventory “are machines, we are robots, we plug our scanner in, we’re holding it, but we might as well be plugging it into ourselves,” and that “we don’t think for ourselves, maybe they don’t trust us to think for ourselves as human beings, I don’t know.”
The BBC submitted its findings, including films of the work done at the warehouse, to a leading UK expert on stress at work, who found that “the working conditions at the warehouse are ‘all the bad stuff at once,’ and that ‘the characteristics of this type of job, the evidence shows increased risk of mental illness and physical illness.’”
Guardian reporter Carole Cadwalladr, who also went undercover for a week at the Swansea warehouse, writes that Amazon “deliberately sites its distribution centers” “in places of high unemployment and low economic opportunities” and that it nevertheless “received £8.8m in grants from the Welsh government” for bringing the warehouse there.
Similar stories proliferate about bad labor practices at Amazon, from its Allentown, PA warehouse to its warehouses in Seattle and Germany (where reports of poor working conditions prompted calls for a consumer boycott). Yet Amazon remains the world’s fourth most admired company, according to Fortune magazine’s 2015 ratings.
Journalist and scholar Simon Head, in his 2014 book Mindless: Why Smarter Machines Are Making Dumber Humans, treats Amazon at length, and pays particular attention to the resemblance between Amazon and Walmart. Head writes:
Amazon equals Walmart in the use of monitoring technologies to track the minute-by-minute movements and performance of employees . . . Amazon’s shop-floor processes are an extreme variant of Taylorism that Frederick Winslow Taylor himself, a near century after his death, would have no trouble recognizing.
Amazon is a master of micromanaging the lives of its employees and contractors. “If an employee is behind schedule,” Head writes, “she will receive a text message pointing this out and telling her to reach her targets or suffer the consequences.” The longer employees work in the warehouses, the higher their “output targets” go, and Head suggests that these targets allow Amazon to fire employees as they gain the seniority that might entitle them to better pay and benefits.
Like Walmart, Amazon stalks the leading edge of purposefully introduced precarity, using all the legal means at its disposal to prevent the classification of its workers as full-time employees entitled to benefits, and to block workers from unionizing. Like Walmart, it employs thousands of people in near-sweatshop conditions, carefully skirting the edges of labor laws, and playing geographic domains off each other to exploit its employees to the fullest extent possible.
Like Uber, Amazon threatens governments (for example, when airspace safety regulations prohibit Amazon from testing drones the way it wants) and plays them against each other as if they were business entities when they fail to do what the company wants. Like Facebook, it develops remarkable, fine-grained tools to track the wants, needs, desires, and hopes of everyday people, perfected to the extent that it claims to know what you want before you want it, as it stated in its 2014 patent for “anticipatory shipping.”
Like nobody else, it is an expert at breaking down problems into tiny pieces, parceling them out to people all over the world willing to do them nearly for free, while pitting these low-wage workers in a “competition” against each other.
As Scholz writes, the company’s business plan appears to be something like “wage theft and total workplace surveillance”; initiatives like Echo and Amazon’s neighbor-delivery program suggest that the “workplace” in that formulation is becoming indistinguishable from the rest of our lives. Again and again, Amazon demonstrates that its goal is to apply computational analytics to every part of the social sphere, and to wring whatever profits it can from low-margin, no-margin, and as-yet un-commoditized parts of the social world.
Like Walmart, Amazon offers buyers (and investors) a deal that is literally too good to be true, an offer we can’t or won’t refuse. As the Canadian writer Tom Slee writes, “individual choice has turned out to be on the side of the powerful. And somehow we have ended up making choices that make us worse off.”
Amazon today offers us “choices” that go far beyond what consumers usually understand that term to mean, extending all the way from how we sell our labor to what parts of our lives count as labor. The company goes out of its way to hide how it does this, selling us products and services that, if examined more closely, would appear deeply troubling — extensions of market logic into places we have not asked for it to go.
Low prices and convenience don’t justify Amazon’s practices — they should be challenged and transformed. Our success or failure to do so may shape many of the labor struggles to come.