The encyclopedia · Software & IT · Product decision · 2005–2011
Amazon Mechanical Turk turned microtasks into a marketplace
Requesters post tiny human tasks for pennies; a diverse global workforce completes them fast — and the data proved at least as reliable as lab samples.
Amazon
the move
Amazon launched Mechanical Turk in 2005 as a marketplace where requesters post microtasks — image labeling, transcription, surveys — and a large pool of workers completes them for small payments.
Buhrmester, Kwang and Gosling's 2011 evaluation found MTurk participants are slightly more demographically diverse than standard Internet samples and significantly more diverse than typical American college samples; participation responds to compensation rate and task length, yet people can still be recruited rapidly and inexpensively.
The study concluded that realistic compensation rates do not affect data quality and that data obtained via MTurk are at least as reliable as those obtained through traditional methods, establishing the platform as a legitimate research tool.
why it works
- A global, on-demand pool removes the cost of recruiting and retaining workers.
- Microtask pricing lets requesters pay only for the work they actually need.
- Built-in payment infrastructure handles compensation and incentives.
- Diverse, self-selected participants can match or beat convenience samples in data quality.
what transfers
Break work humans do well but computers can't into paid microtasks and let a marketplace set the price — you get scale and diversity without hiring a workforce.
what came after
MTurk became a standard source of research subjects and microtask labor, spawning a wider gig-economy of human-in-the-loop work and a large literature on crowdsourced data quality.
references
- Amazon's Mechanical Turk: A New Source of Cheap, Yet High-Quality, Data?
- Amazon Mechanical Turk — Worker
spotted an error? The archive wants to know.