Amazon Turk versus Wikipedia

Since I have been back in Cambridge, I have seen no less than three presentations by researchers at Harvard and MIT on the use of Amazon Turk in social science experiments. I don't think it's a stretch to say that the ability to easily and efficiently recruit experimental subjects online is a significant shift in the landscape.

David Rand, a fellow Berkman Fellow, gave a particularly interesting talk, which was blogged by Ethan Zuckerman. During the discussion I asked if involvement in such (minutely) compensated tasks might depress peoples' willingness to participate in voluntary activities such as Wikipedia. Fortunately, David did ask about Wikipedia participation in one of his surveys, and shared the data with me. Of David's Turkers who read Wikipedia daily, 61% never contribute and 39% contribute at least once a month. If we compare this to a survey of Wikipedians, and we assume that self-selection leads to them reading Wikipedia about every day, we see that 65.92% are only readers and 30.67% are contributors ("occasional" and "regular", p. 5).

Those figures are surprisingly close, in the rough. One could investigate this further by asking other questions about gender balance, whether Wikipedia administrators "turk," and how much time Wikipedia contributors estimate they spend on Wikipedia and/or Turk?

Comments !