Mechanical Turk

Along with many other people, I’ve been playing around with Amazon’s new tool Mechanical Turk, a mechanism to apply distributed human intelligence to mundane tasks that are still too complicated for comouters. I’m guessing I’m one of many given the poor availability of the service over the weekend – it was ‘unavailable’ for the majority of time I was trying it, stating it was too busy. Amazon seriously underestimated the desire for people to earn ‘free’ money.

However, I have managed to earn the grand total of 9c, which I could now apply as a discount to one of my next purchases (although I’m more likely to just leave it there). I’m waiting for another 60c to come through, waiting for someone to approve my choices. Which does beg the questions – I’m paid “… when your answer is approved by the person that listed the HIT”. So, if someone has to go in and approve all the HITs (‘Human Intelligence Task’ don’t you know), why don’t they just do the job in the first place? How much time is saved through the two stage process than just getting a single team to do it? Although I’m guessing this process may be better for the person’s snaity 😉 To do any serious earning on this, it will take a lot of tiome – and a lot more power on the Amazon side to keep the service up and running.

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Comments are closed.