Amazon and The New ‘Big Data’ Proletariat

amazon
Reuters

Last week, the New York Times ran a long, viral, and controversial story about the working conditions at online retail giant Amazon.com. Amazon is huge, so its business practices are often a subject of media scrutiny, even though it’s an entity the Left is generally comfortable with.

The Times article went remarkably hard on what it described as a “bruising” workplace, where employees are pitted against each other in endless back-stabbing wars, and grueling sweat-shop hours were demanded even from workers dealing with serious health complications. In fact, some of the interviewees said working at Amazon was giving them health complications.

Amazon views its work environments as a tough response to slack conditions elsewhere in the labor force, encouraging excellence and superior effort from its workers by setting standards it proudly describes as “unreasonably high.” Many horrified readers thought the NYT article described brutally exploitative conditions covered by corporate happy-talk about achievement.

Critics of the piece noted that its assertions were difficult to square with large amounts of feedback data gathered from Amazon employees, who in the aggregate gave the company decent marks, and at worst considered it roughly comparable to the pace set by other big Silicon Valley firms. The New York Times’ own public editor thought the article drew too many broad generalizations from a small sampling of terrible anecdotes.

Perhaps the most interesting critique comes from New York magazine, which argues that the New York Times’ analysis was missing the bigger story about Amazon.com. The degree of surveillance and monitoring Amazon subjects its employees to is more disturbing, and probably a greater harbinger of things to come across the American workforce, than demands for long hours and swift responses to late-night corporate email.

The true menace here, in author Benjamin Wallace-Wells’ view, is data:

In Amazon’s warehouses, we learn, workers “are monitored by sophisticated electronic systems to ensure that they are packing enough boxes.” In the white-collar jobs that are the story’s real subject, the company is exacting in similar ways. “The company is running a continual performance improvement algorithm on its staff,” a former marketer on the Kindle team explains. Before regular performance reviews, Amazon workers are given “printouts, sometimes up to 50 or 60 pages long,” that measure their performance on many different metrics. (It’s a little amazing that Amazon is still printing this stuff out on paper.) The totality of this measurement, the Times suggests, means not that Amazon is unique but merely that the company has been “quicker in responding to changes that the rest of the work world is now experiencing: data that allows individual performance to be measured continuously.”

Which is the line at which the average white-collar Times reader is meant to experience a sense of imminent collapse and dread. Obnoxious as most of the abuse is, as many lines as it crosses, a reader who works at another company can chalk it up to a particular sick corporate culture, located in Seattle and presided over by a megalomaniac. You can reassure yourself that you have a kinder boss and a more decent set of rules. But “continual improvement algorithms” are innovations, the Times explains, the kind that are now arriving in “the rest of the work world” and just happened to come to Amazon first. The real villain of the Times piece isn’t Bezos or his senior executives. Instead, it’s Taylorism for the professional class, in the guise of data. “Data,” a senior Amazon executive tells the Times, “is incredibly liberating.”

Wallace-Wells sees this as a transition from a more idealistic, laid-back era, exemplified by Facebook and its CEO Mark Zuckerberg, to a new level of cubicle hell where workers will essentially be treated like robots… because they’ll be competing with robots.

In fact, the data-driven corporate culture obliquely referenced by the New York Times and picked out by New York Magazine is similar to the internal diagnostic routines a massive computer system runs on itself many times each second. The employees become components, monitored as such by impersonal data analysis. Furthermore, they become part of the data harvesting system, as well as productivity sources, because they’re encouraged to provide feedback on each other in various ways. They provide both hands to work, and eyes to monitor what all the other hands are doing.

It’s an approach likely to go “viral” through competition and mass production. Every company, even fairly small operations, will soon be capable of such data harvesting from its workforce.  It’s fair to say they already have the tools – almost every job requires the use of computerized devices at some point, they’re all networked together, and networked data can be scooped up into big-picture reports. Even the tools in a mechanic’s workshop are, or soon will be, capable of communicating with the company’s master computer system. Even the appliances in a kitchen can put their digital heads together and produce a fairly accurate picture of what the cooks are doing all day… or, more to the point, every minute of the day.

Competition will make workforce surveillance more widespread, as a growing number of companies realize they can automatically monitor performance through various data metrics, and fear to become the last company that loses productivity by failing to do so. The process will be exacerbated by such policies as higher mandatory minimum wages and workplace benefits. As the cost of labor blows sky-high, employers will want to make do with fewer warm bodies, expect more production per minute from the human workers they retain, and become ever more interested in spotting both unsatisfactory and exceptional workers quickly. As value is forcibly decoupled from the price of labor, management naturally becomes more interested in forcing value to meet the price they can no longer control.

This will also naturally lead to discomfort and resentment from employees, who resent being monitored with such creepy Panopticon efficiency, and judged based on impersonal reams of data. It’s always been a matter of contention how much an employer can and should demand – those are two separate questions, divided by the wisdom of finding the point at which the overall productivity and health of the operation declines because the employees are unhappy, over-stressed, or inclined to bolt for better working conditions at the first opportunity, taking valuable experience with them.

It’s always been a human question, often influenced by how closely management works with employees – in a small operation, the hands-on working manager sets conditions he or she is personally comfortable with, and perhaps is more interested in maintaining upbeat personal connections with workers seated five feet away. Data mining on the Amazon scale purges the human element in a large workforce, and creates a coldly rational statistical model. It makes employees feel less trusted to meet reasonable standards with their own methods, having been given a clear idea of what they’re expected to accomplish.

None of that matters a bit to the increasingly intelligent machines that are becoming increasingly affordable alternatives to human employees in a growing number of positions, particularly as consumers grow more comfortable with ordering from machines, and in many cases prefer it. Such is the case at an operation like Amazon.com, where consumers don’t talk to human employees very often, and are primarily interested in the astounding machine-like efficiency and cost-effectiveness of a company that delivers a staggering variety of products at discount prices with remarkable speed.

Is it fair to say that such expectations are growing – that more people expect Amazon’s model of performance in everything from retail shopping, to transportation and pizza delivery? If so, would it also be fair to say the Amazon workplace model will inevitably follow those expectations?

COMMENTS

Please let us know if you're having issues with commenting.