I recently read Good Work, a collection of economist/philosopher E. F. Schumacher’s musings on meaningful job creation and appropriate technology for the world’s poor. Schumacher’s ideas were revolutionary for the 1970s — he was one of few development thinkers to argue that economic growth shouldn’t come at the expense of fulfilling work. The book’s prologue quotes Camus: “Without work, all life goes rotten, but when work is soulless, life stifles and dies.”
Over at Samasource, we’ve been following the development of Amazon’s Mechanical Turk closely. The Turk, which derives its name from a 19th century chess-playing machine that was secretly powered by a human, helps engineers find people to complete rote “human intelligence tasks” (HITs) such as photo tagging to complement work done by computers. HITs are completed for as little as $0.02 each by people in over 100 countries. When it launched, Amazon was widely criticized for providing a “virtual sweatshop” platform to US-based companies — the primary unit of labor on the site is a task (rather than an hour, or a project), so posters needn’t worry about minimum wage laws, overtime, health insurance, or other policies designed to protect workers.
However, a brief perusal through one of several forums devoted to “Turkers” shows that worker perceptions of the platform, at least here in the US, aren’t so negative. Some use the Turk to earn money while procrastinating, while others have been paid to participate in creative projects, such as the much-publicized Sheep Market art installation designed by Aaron Koblin. People I’ve spoken to in the business process outsourcing industry, which often engages humans to perform routine tasks requiring little creativity, think the Turk might be a more efficient, flexible way to distribute work across a pool of willing workers.
What does all this mean for companies that aim to create “good work” for their employees? Some outsourcing firms compensate lower-level employees for performing monotonous tasks by providing good recreational facilities, benefits, and training programs designed to help them transition into other careers (Digital Divide Data, a data entry firm based in Cambodia, is a fine example). This is costlier than simply parsing out work in a system like the Turk; people aren’t fungible resources, whereas tasks are, and fungibility seems to make things cheaper.
Right now, it seems like most Turk work is fairly low-level — for a while, at least, business process outsourcing companies that handle high-value services like integrated digitization or customer support won’t be threatened by task-oriented marketplaces focused on highly distributed data projects. But in the long term, what might Turk-style automation do to companies trying to create meaningful jobs for people?