Coding Horror

programming and human factors

Is Amazon's Mechanical Turk a Failure?

Amazon's Mechanical Turk Service is a clever reference to the famous chess-playing hoax device, The Mechanical Turk. The Mechanical Turk dates back to 1770, and has quite a storied history. Read through the Wikipedia article if you have time; it's fascinating stuff.

The Mechanical Turk

The secret of the Turk, of course, was that it wasn't a chess-playing machine at all. There was a small person inside, controlling the machine.

Similarly, Amazon's Mechanical Turk is a machine that harnesses the work of hidden humans. It's a service that attempts to match people to small, bite-size units of work that are unsuitable for machines.

As of this writing, there are 128 Human Intelligence Tasks available via the Mechanical Turk task page. The reward for these tasks ranges from $1.00 to $0.10, skewing heavily toward the bottom of that range. Almost 100 of the 128 tasks are $0.10 each. Here's a quick sampling of the available tasks:

  • Transcribe a 9 minute, 2 second podcast ($2.31 w/bonus)
  • Write a review of a blog ($1.00)
  • Make ten 2-3 sentence posts in a fansite forum ($0.50)
  • Write a 2-3 paragraph blog entry ($0.50)
  • Provide 3-D and 4-D ultrasound pictures of your baby ($0.40)
  • Send unsolicited junk faxes from California companies ($0.25)
  • Say 6 phrases in Turkish ($0.10)
  • Write a short plot description of the movie "Black Snake Moan" ($0.10)

Read through some of the available HITs yourself. Be sure to click on the HIT to get the details on the job and any rules. You're at the mercy of the requester; it's up to them to judge your work worthy of payment.

Based on the quantity, quality, and type of tasks available, I think Amazon's Mechanical Turk may be a failure. It's been almost two years, and almost all the tasks have one or more of these problems:

  1. obviously and suspiciously spammy
  2. require a lot of subjective human intervention and effort for "grading"
  3. the rates make working in a sweatshop seem lucrative

What I find ironic about Amazon's Mechanical Turk service is that Amazon built an entire business around the value of user reviews. The strength of the user reviews is one of the main reasons I frequent Amazon. That's user-submitted content that people invested countless thousands of man-hours on. And Amazon didn't pay anyone a dime to do it.

I think the secret to running a viable Mechanical Turk service is, paradoxically, to do away with payment. Instead, they should have chosen a reward system based on intrinsic motivation. Intrinsic motivation is the reason why..

Nobody's paid to do any of the above. And yet each item I listed is easily equivalent to multiple Turk HITs. The best explanation I've found for this behavior is in Mary Poppendieck's Team Compensation (pdf).

There are two approaches to giving children allowances. Theory A says that children should earn their allowances; money is exchanged for work. Theory B says that children should contribute to the household without being paid, so allowances are not considered exchange for work. I know one father who was raised with Theory B but switched to Theory A for his children. He put a price on each job and paid the children weekly for the jobs they had done. This worked for a while, but then the kids discovered that they could choose among the jobs and avoid doing the ones they disliked. When the children were old enough to earn their own paychecks, they stopped doing household chores altogether, and the father found himself mowing the lawn alongside his neighbors' teenage children.

Were he to do it again, this father says he would not tie allowance to work.

In the same way, once employees get used to receiving financial rewards for meeting goals, they begin to work for the rewards, not the intrinsic motivation that comes from doing a good job and helping their company be successful. Many studies have shown that extrinsic rewards like grades and pay will, over time, destroy the intrinsic reward that comes from the work itself.

The theory of intrinsic motivation goes a long way toward explaining why Amazon's unpaid user reviews are so popular and effective, and yet the paid Mechanical Turk service appears to be withering on the vine.

Poppendieck notes that choosing a payment reward model can be an irreversible decision: once you go down the path of monetary rewards, you may never be able to go back, even when they cease to be effective, as they inevitably will. I think that's clearly the case for Amazon's Mechanical Turk.

Written by Jeff Atwood

Indoor enthusiast. Co-founder of Stack Overflow and Discourse. Disclaimer: I have no idea what I'm talking about. Find me here: