The Mechanical Turk

First, some history of the term "The Mechanical Turk":

The Turk was a famous hoax which purported to be a chess-playing machine. Constructed and unveiled in 1770 by the Hungarian baron Wolfgang von Kempelen (1734–1804), the mechanism appeared to be able to play a strong game of chess against a human opponent, as well as perform the knight's tour, a puzzle which requires the player to move a knight to occupy every square of a chess board once and only once.

Publicly promoted as an automaton, the Turk was given its common name based on its appearance, and was a mechanical illusion that allowed a human chess master to hide inside and operate the machine. With a skilled operator, the Turk won most of the games it played. The apparatus was demonstrated around Europe and the United States of America for over 80 years until its destruction in 1854, playing and defeating many challengers including statesmen such as Napoleon Bonaparte and Benjamin Franklin.

There's quite a bit more in Wikipedia.

The problem for The Mechanical Turk, of course, was that the technology that existed at the time - unlike now - was unable to create a chess playing machine. The idea that there are tasks too difficult for machines to perform (so far anyway) is the basis for this web site. The term Mechanical Turk and the idea behind the site are described here:

Today, we build complex software applications based on the things computers do well, such as storing and retrieving large amounts of information or rapidly performing calculations. However, humans still significantly outperform the most powerful computers at completing such simple tasks as identifying objects in photographs—something children can do even before they learn to speak.

When we think of interfaces between human beings and computers, we usually assume that the human being is the one requesting that a task be completed, and the computer is completing the task and providing the results. What if this process were reversed and a computer program could ask a human being to perform a task and return the results? What if it could coordinate many human beings to perform a task?

Amazon Mechanical Turk provides a web services API for computers to integrate "artificial artificial intelligence" directly into their processing by making requests of humans. Developers use the Amazon Mechanical Turk web service to submit tasks to the Amazon Mechanical Turk web site, approve completed tasks, and incorporate the answers into their software applications. To the application, the transaction looks very much like any remote procedure call: the application sends the request, and the service returns the results. Behind the scenes, a network of humans fuels this artificial artificial intelligence by coming to the web site, searching for and completing tasks, and receiving payment for their work.

All software developers need to do is write normal code. The pseudo-code below illustrates how simple this can be.

 read (photo);
 photoContainsHuman = callMechanicalTurk(photo);
 if (photoContainsHuman == TRUE) {
   acceptPhoto;
 }
 else {
   rejectPhoto;
 }

The site lists the current "Human Intelligence Tasks," the tasks that are ill-suited for computers. Here's a description of a HIT from the site:

What is a HIT? HIT stands for Human Intelligence Task. These are tasks that people are willing to pay you to complete. For example a HIT might ask: "Is there a pizza parlour in this photograph?" Typically these tasks are extraordinarily difficult for computers, but simple for humans to answer.

Many of these tasks pay a nickel, a dime, 69 cents. Here are a few samples. This is the first entry on the default list (click on Get Started Now in the link above to view more HITS):

Classify URL
Requester: Idearc Media Inc.
HIT Expiration Date: Mar 1, 2008 (49 weeks 5 days)
Time Allotted: 15 minutes
Reward: $0.01
HITs Available: 2262
Description: Classify URL as one of several business categories
Keywords:
Qualifications Required: None

Another example:

GIS Image Tagging - Sidewalks
Requester: Geospatial Vision
HIT Expiration Date: Mar 24, 2007 (5 days 13 hours)
Time Allotted: 2 hours
Reward: $0.05
HITs Available: 1512
Description: In this assignment you will be given 50 images and asked to go through each image clicking on the edges of the sidewalks.
Keywords:
Qualifications Required: None

Most pay peanuts. Here's one of the higher paying HITS, though it's not much for the work involved:

Make my editor happy -- write a better draft
Requester: Matthew P Stadler
HIT Expiration Date: Apr 1, 2007 (1 week 6 days)
Time Allotted: 10 hours
Reward: $20.00
HITs Available: 1
Description: Rewrite my draft and satisfy my editor
Keywords:
Qualifications Required: None

And finally, someone needs comments on their blog:

Write a comment on a personal finance blog
Requester: Adam Powell
HIT Expiration Date: Mar 25, 2007 (6 days 13 hours)
Time Allotted: 60 minutes
Reward: $0.02 HITs Available: 1
Description: Visit my blog and write a meaningful comment about the latest entry.
Keywords:
Qualifications Required: None

Interesting technological development in "outsourcing" to humans.

Comments

[maria]

I wonder if there is a website where students solicit someone to write a term paper for them, for pay, of course. There are sites with already written term papers, but professors have found ways to discover when these are used. But a new custom-written and thus unique term paper could not be easily detected. I would think there are students rich enough to pay, say $250-$500, maybe more, for a good term paper, and writers willing to do the job. Any opinions?

[Nicolas]

People are willing to identify items on pictures for free at http://www.espgame.org/.
More details :
http://video.google.ca/videoplay?docid=-8246463980976635143&q=human+computing

[yartrebo]

maria: Rich students paying to have custom term papers written in nothing new. Still, it can be detected by an astute professor based on the writing style and quality of the paper not matching the quality of the student's in class work.

[cm]

Compare this to the "lifelong retraining" propaganda. Symbol processors, literally. Perhaps that's the new Taylorism. Automation of high-level "business logic" and "plugging in" human processing units" (HPUs) to perform "functional modules".

[Andy Dent]

I noticed that it hasn't taken long for the spammers/google hackers to pick up on the possibilities - see the job requests for example from ContentSpooling.net to get people to generate unique content.

[David Eisenhardt]

Amazon was just getting cheap beta testing for Powerset, the search engine that's supposed to compete with Google. More details. I think they got it right.