Who the crowdworkers are: six surveys over twenty months
In April 2010 five researchers at the University of California, Irvine pulled together six surveys of Mechanical Turk workers spanning twenty months. The share of workers from India rose from 5 per cent in November 2008 to 36 per cent a year later.
Why it matters
Until then the people who label data were discussed by guesswork: either bored Americans earning pocket money or cheap labour in poor countries. This work turned it into numbers and showed a third thing - young, well-educated people, some of whom do not treat the marketplace as pocket money. Arguments about conditions in data work could now be made with figures.
Six samples: March 2008 (n=300) and November 2008 (n=1010) from Ipeirotis's earlier surveys, then the authors' own - February 2009 (n=878), May 2009 (n=512), August 2009 (n=578) and November 2009 (n=733). Reliance on the earnings: around 27 per cent of Indian workers said they sometimes or always need Mechanical Turk "to make basic ends meet", against around 14 per cent of US workers; about 18 per cent said so overall. The authors put the average wage of Indian workers at roughly 75 per cent of the Indian minimum wage. Average pay on the platform at the time was about $2 an hour. The record does not claim that most Indian workers earn under $10,000 a year: the text says something else, that nearly a third of all respondents in November 2009 reported under $10,000. There is no February 2010 survey in this work; the last one is November 2009.