Machine learnings

Welcome to the inaugural edition of Metal Machine Music. This one is mostly focused on automation, but it also includes a roundup of some recent pieces and a little bit of history at the end. I’m hoping to do one of these every week, and I expect the subject matter to vary a fair bit depending on what I’m reading.

Anyway, I hope you enjoy. And if you do, subscribe.

Robots aren’t the reason you don’t have a job

Automation is one of those subjects that tends to attract a lot of hype and a lot of misinformation. The mainstream conversation about tech has gotten more sophisticated in recent years, but the mainstream conversation about automation remains surprisingly undeveloped. To put it less politely, there’s an incredible amount of nonsense out there. And now we have Andrew Yang in the Democratic debates talking about how self-driving trucks are going to bring mass unemployment—a beloved (and inaccurate) talking point of the mainstream automation discourse.

Enter Aaron Benanav, whose brilliant piece in the latest issue of the New Left Review, “Automation and the Future of Work,” demolishes the popular myths around automation. It’s a long and meaty article, so I won’t try to summarize all of its points here. But one takeaway is that the mainstream automation discourse is responding to a real phenomenon—the fact that we have too few jobs for too many people—but placing the blame where it doesn’t belong. Persistently low demand for labor isn’t being caused by automation, Benanav argues. Rather, the actual culprit is decades of deepening economic stagnation—what the historian Robert Brenner calls the Long Downturn—which has its roots in global industrial overcapacity.

There are also two new reports on automation that are very much worth reading alongside Benanav:

  • How Robots Are Beginning to Affect Workers and Their Wagesby William M. Rodgers III and Richard Freeman of the Century Foundation. This looks at the effect of robots on wages and employment in the US from 2009 to 2017. Among the takeaways:

    • Manufacturing is by far the sector with the most robots, and their use has grown sharply since 2009.

    • So far, growth in roboticization is not producing mass unemployment. Workers displaced by robots tend to find other jobs (even if they pay less).

    • Even so, certain regions and certain kinds of workers are hit harder by roboticization than others: in the Midwest, for example, robots have significantly hurt the wages and employment of young, less-educated men and women in manufacturing occupations. (Young, less-educated black men and women are the most adversely affected.)

    • The (weak) economic recovery since 2009 may have “masked” some robot-induced job losses, because the economy has been capable of absorbing the displaced workers—even if their new jobs were probably worse paid and more precarious. (Most of the new jobs created post-recession have been “alternative work”: temp, on call, contract, freelance, etc.) This suggests that if another big recession hits, technological unemployment may become more visible.

  • “The Future of Warehouse Work: Technological Change in the U.S. Logistics Industry” by Beth Gutelius and Nik Theodore of the UC Berkeley Labor Center. This looks at how the adoption of new technology may transform warehouse jobs over the next five to ten years. Among the takeaways:

    • Big job losses are unlikely, but jobs are likely to get worse:

      • Employers may use technology to deskill roles rather than eliminate them, so that they can pay workers less.

      • Technologies may also be used to intensify the pace of work, a phenomenon already very evident in Amazon warehouses. (See Sam Adler-Bell’s “Surviving Amazon” for a detailed look at how this works.)

      • More invasive forms of labor discipline through electronic monitoring and algorithmic management may also increase.

    • Technology uptake will be uneven. This is an important point, and one too often neglected in our conversation about how companies use technology: even firms that introduce the most advanced forms of automation generally use them alongside older, legacy systems.

Other readings

Some of the other things I’ve been reading and thinking about recently:

  • More fun at Google. Google has been the epicenter of the wave of tech worker organizing that has swept the industry in recent years. A couple of new reports help give us a sense of where things are at.

    • Google management tried to shut down an employee-organized talk with two union representatives at its Zurich office entitled “Unions in Switzerland.” According to Recode’s Shirin Ghaffary, “Google leadership wrote in the note that it would host its own talks about labor laws and employee rights that bring ‘a diverse set of presenters and perspectives.’” The Google employees went ahead with the talk anyway.

    • Google has developed a Chrome extension that alerts management if any employee tries to create a calendar event with more than ten rooms or 100 participants. According to an internal memo being circulated by a Google employee, this tool was designed explicitly to monitor (and discourage) worker organizing.

  • The business of regulation: Tom Krazit over at Mostly Cloudy has a fascinating post on why Compliance-as-a-Service (CaaS) is likely to become a lucrative business if tech regulation increases, and why Facebook would be well-positioned to dominate it. “Facebook will likely have to build an enormous compliance operation to handle even half of the proposals being kicked around across the planet,” Krazit writes, “and it will develop unique insights into how to introduce compliance at companies that rely on user-generated content that other cloud giants and startups can’t easily duplicate.” Yet another example of how the emerging tech regulatory landscape may end up benefiting the big platforms.

  • Senator, we sell ads: Nick Serpe in Dissent explores the “new culture war” around online speech in the platform era, and the different strains of anti-tech sentiment that have taken root among Democrats and Republicans. It’s helpful context for the recent Zuckerberg hearings, and more broadly for the ongoing platform regulation debates on Capitol Hill.

History corner

Lastly, a little bit of history to close us out. Reading Ruha Benjamin’s Race After Technology, I learned about a group called the Polaroid Revolutionary Workers Movement. Formed in 1970, this was an organization of African-American employees at Polaroid who demanded that the company cut its ties with the apartheid regime in South Africa.

At the time, the South African government was using Polaroid cameras to take photos of Black citizens for the “passbooks” that controlled their movements around the country. (In particular, they used Polaroid’s ID2 camera, which included a “boost button” that created a brighter flash in order to make Black faces more visible—which was needed because of film stock’s bias against darker skin.)

Polaroid ended up firing the core organizers of the Polaroid Revolutionary Workers Movement. But by 1977, thanks to ongoing pressure, the company halted all direct sales to South Africa.