Amazon Mechanical Turk


A few years back I heard about Amazon’s new “Mechanical Turk” service. Basically you post a task, an amount you’re willing to pay to have it completed, and then people from all over the world work on it.  In my opinion it is best suited for super repetitive simple tasks that you can’ have a computer program do.  For example:

  • Visit URL xxxyyy and copy the article author’s name, and date of the article
  • Look at this video and tag it with 5 relevant / descriptive keywords
  • Choose the best category for this product
  • Flag offensive content images

It seemed like a great idea, but in a cursory view it also felt like an overly complicated process, and at the time I didn’t have thousands of simple and similar tasks that I needed cranked out.

Fast-forward to this evening (yes, it is 1:00 am right now) and just as I was about to send off a mind numbing task to my new virtual assistant I paused… then the image of the Mechanical Turk came into my mind.  Not only did I feel this project was a perfect match for what I understood about Amazon’s groupsourced system, I was excited about the opportunity to finally test it out.

While Amazon could do a better job with documentation and training videos,  the process honestly wasn’t as painful as I thought it would be.

My task:  Visit all the URLs in a spreadsheet and copy specific information from each one and fill in the blanks.  Basic data mining / data entry.

Here are the steps (simplified) that I went through: Continue reading “Amazon Mechanical Turk”