Matrix algebra: how to be human in a digital economy

Ray and Charles Working on a Conceptual Model for the Exhibition Mathematica, 1960, photograph. Prints & Photographs Division, Library of Congress (A-22a). Click here to see original image.


“Certainly the cost of living has increased, but the cost of everything else has likewise increased,”[1] H.G. Burt, the President of the Union Pacific Railroad, asserted to railroad company machinists and boilermakers.  For Burt, the “cost of everything else” included the cost of labor. His remedy: place “each workman on his [own] merit.”  In 1902, “workman merit” to a tycoon like H.G. Burt squarely meant equating the value of labor, or the worth of a person, to the amount of output each individual produced.  Union Pacific Railroad eventually made use of this logic by replacing the hourly wages of workers with a piece rate system.  Employers switched to piecework systems around the turn of the 19th century largely to reduce labor costs by weeding out lower skilled workers, and cutting the wages of workers unable to keep apace with the “speeding up” of factory production.

Employers historically leveraged piecework as a managerial tool, reconfiguring labor markets to the employers’ advantage by allowing production rates, rather than time on the job, to measure productivity.  Whatever a person produced that was not quantifiable as a commodity, in other words, did not constitute work.  We’ve seen other examples of discounted labor in spaces outside the factory.  Feminist economists fight to this day, for example, for the work of caregivers and housewives, largely ignored by mainstream economic theory, to gain recognition as “real” forms of labor.  Real benefits and income are lost to those whose work goes unaccounted.

As the historical record shows, workers do not typically accept arbitrary changes to their terms of employment handed down by management.  In fact, the Union Pacific Railroad machinists protested Burt’s decision to set their wages through a piecework system.  H.G. Burt met their resistance with this question: is it “right for any man to ask for more money than he is actually worth or can earn?”

But what is a person truly worth in terms of earning power?  And what societal, cultural, and economic factors limit a person from earning more?

In 2014, the question of a person’s worth in relation to their work, or the value of labor itself, is no less prescient.  The rhetoric surrounding workers’ rights compared to those of business differs little whether one browses the archives of a twentieth century newspaper or scrolls Facebook posts.  Ironically enough though, in the age of social media and citizen reporting, the utter lack of visibility and adequate representation of today’s workers stands in stark contrast to the piece rate workers of H.G. Burt’s day.  Few soundbites or talking points, let alone byline articles, focus on the invisible labor foundational to today’s information economies.  Nowhere is this more clearly illustrated than with crowdwork.

Legal scholar Alek L. Felstiner defines crowdworking as, “the process of taking tasks that would normally be delegated to an employee and distributing them to a large pool of online workers, the ‘crowd’” (2011).  Hundreds of thousands of people regularly do piecework tasks online for commercial, crowdsourcing sites like’s Mechanical Turk (“AMT”).

Over the last year, we’ve worked with Dr. Siddharth Suri and an international team of researchers, to uncover the invisible forms of labor online, and people who rely upon digital piecework for a significant portion of their income.  Crowdwork is, arguably, the most economically valuable, yet invisible, form of labor that the Internet has ever produced.  Take Google’s search engine for instance.  Each time you search for an image online (to create the next most hilarious meme, or find a infograph for a conference presentation) you’re benefitting from the labor of thousands of crowdworkers who have identified or ranked the image your search populates. While this service may be valuable to you, the workers doing it, only receive a few cents for their contributions to your meme or slideshow presentation.  Additionally, a typical crowdworker living in the United States makes, on average, 2 to 3 dollars an hour.  We need to ask ourselves: what is fair compensation for the value that workers bring to our lives?  How would you feel if tomorrow, all your favorite, seemingly free, online services that depend on these digital pieceworkers, disappeared?

Last fall, we spent four months in South India talking with crowdworkers and learning about their motivations for doing this type of work.  In the process we met people with far ranging life experiences, but a common story to tell – perhaps familiar to all of us who’ve earned a wage for our keep: work is not all we are, but most of what we do is work.  And increasingly, the capacity to maintain a living above the poverty line is elusive, and complicated by what “being poor” means in a global economy. Our hopes for finding more satisfying work, a life valued for what it is rather than what it is not — is no less, even as we confront the realities of today.

Moshe Marvit spoke to the complexities of crowdwork as a form of viable employment in a compelling account of U.S. workers’ experience with Amazon Mechanical Turk. He describes this popular crowdsourcing platform as “one of the most exploited workforces no one has ever seen.” Marvit emphasizes how crowdwork remains a thing universally unacknowledged, in that more and more tasks, from researchers’ web-based surveys and to Twitter’s real-time deciphering of trending topics, depend on crowdwork.  However, most people still don’t know that behind their screen is an army of click workers.  Anyone, who has ever browsed an online catalogue or searched the web for a restaurant’s physical address, has benefited from a person completing small, crowdworked task online.  Pointedly, our web experience is better because of the thousands of unknown workers who labor to optimize the online spaces we employ.

As Marvit points out, and our research also notes, people only earn pennies at a time for doing the small crowd tasks not yet fully automatable by computer algorithms. These crowd tasks, however, add up to global systems whose monetary worth sometimes trumps that of small nations.  Yet, when we ask our peers and colleagues, “do you know who the thousands of low income workers are behind your web browser?”  We receive mystified stares, and many reply “I don’t know.”

The hundreds of thousands of people who regularly work in your web browser are not the youth of Silicon Valley’s tech industry.  They likely cannot afford Google glass, or ride to work in corporate buses.  Some are college educated, but, like people today – they are stuck in careers that undervalue their real worth, in addition to discounting the investments they’ve already made in their education, skills, and the unique set of values they’ve gained from their own life experiences.

Yet, the more our research team learns about crowdworkers’ lives, the more we realized how little we know about the economic value of crowdwork and the makeup of the crowdworking labor force. And as Marvit notes, we still don’t have a good grasp of what someone is doing, legally speaking, when they do crowdwork. Should we categorize crowdwork as freelance work? Contract labor? Temporary or part-time work?

In the absence of answers to these questions, some have called for policy solutions to mitigate the noted and sometimes glaring inequities in power distributed between those posting tasks (or, jobs) to crowdwork platforms, and those seeking to do crowdwork online.  But, we argue, good labor policy that makes sense of crowdwork, from a legal or technical point of view, can’t be adequately drafted until we understand what people expect and experience doing task-based work online. Who does crowdwork? Where, how, and why do they do it? And how does crowdworking fit into the rest of their lives, not to mention our global workflows? When we can answer these questions, we’ll be ready to talk about how to define crowdwork in more meaningful ways. Until then, we resist dubbing crowdwork “exploitative” or “ideal,” because doing so is meaningless to the millions of people who crowdwork, and ignores the builders and programmers out there trying to improve these technologies.

We are all implicated in the environments we rely on and utilize in our daily lives, including the Internet.  Those who mindlessly request and outsource tasks to the crowd without regard to crowdworkers’ rights, are perhaps, no more at fault than the rest of us who expect instant, high quality web services every time we search or do other activities online.  An important lesson from Union Pacific Railroad still holds true: workers are not expendable.

This is cross-posted from the Social Media Collective Blog at Microsoft Research, New England and the Center for Popular Economics at the University of Massachusetts, Amherst.

[1]Omaha daily bee. (Omaha [Neb.]), 01 July 1902. Chronicling America: Historic American Newspapers. Lib. of Congress. <>