Xerox researchers present study on “turking” and other crowdsourced work.

By David B. Martin, research scientist, Xerox Research Centre Europe

dave_martin_2

“Relationships are key: Turkers like anonymity and flexibility.” — David B. Martin

Turking is a term that refers to crowds of people who perform tasks that computers don’t do well, such as picking details out of images. If a city government wants to count the number of parking meters covered in graffiti, for example, it can pay “Turkers” nominal amounts of money to click through thousands of photographs and tag the meters that need to be cleaned up.

I led a team of researchers who studied this online marketplace. We explored who Turkers are, how they carry out human intelligence tasks (HITs) and how crowds are designed and controlled to get this “invisible” work done.

Below are just five of the many findings the research unearthed:

  1. Relationships are key. Turkers like anonymity and flexibility, but want decent working relationships with courteous communication. They want fair pay for fair work (such as decent wages, fairness in judging work, and timely payment). Respect works both ways: Good requesters are prized.
  2. Members on Amazon Mechanical Turk see Turking as work and are primarily motivated by earning.
  3. Earnings vary but Turking is low wage work: High earners on Turker Nation make approximately $15,000 to 16,000 per year.
  4. Workers aspire to earn at least $7 to $10 per hour, but (newbies especially) do lower paid HITs to increase their reputation and HIT count.
  5. Many Turkers choose Amazon Mechanical Turk because they cannot find a good ‘regular’ job or need other income. Some are housebound; others are in circumstances where Turking is one of the few options to earn.

I posted these findings on Follow the Crowd, a crowd research blog, which drew some interesting input from Turkers. I’d love to know what you think.

David will present “Being a Turker” at the Association for Computing Machinery conference on computer supported cooperative work and social computing (CSCW), which runs Feb. 15-17 in Baltimore. The paper was written by him and fellow Xerox Research Centre Europe researchers Benjamin Hanrahan, Neha Gupta from Nottingham University currently visiting XRCE, and a former colleague Jacki O’Neill.