Slashdot Mirror
Amazon's Mechanical Turk
rscoggin writes "Amazon.com has a new program that wants you to 'Complete simple tasks that people do better than computers. And, get paid for it.' (example: 'Is there a pizza parlour in this photograph?'). For each task you complete you get a small payment, usually ranging from a few cents to a little under a dollar. It's named the Amazon Mechanical Turk after a famous hoax from the 19th century. Kill time and get paid in tiny increments to boot!" Similar to Google Answers, there seems to be a reliability ratings system and some incentives.
Great... Another way for /.'ers to waste time at work.
GOOD JOB AMAZON
[Fuck Beta]
o0t!
- can you see boobs in the picture ?
- Is there a donkey in the picture ?
- Can you see the can of whipped cream ?
- is there chocolate paint involved..
Advanced indexing of Pr0n, humanity is moving forward, no doubt.
Actually, yes. The whole motivation from this came from the same person who invented the CAPTCHA, and was explained in his thesis defense on Wednesday. Abstract for those who care:
Subject: Thesis Oral - Luis von Ahn
November 2, 2005
Luis von Ahn
12:00 PM, 3305 Newell-Simon Hall
Thesis Oral
Title: Human Computation
Abstract:
Tasks like image recognition are trivial for humans, but continue to
challenge even the most sophisticated computer programs. This thesis
introduces a paradigm for utilizing human processing power to solve
problems that computers cannot yet solve. Traditional approaches to
solving such problems focus on improving software. I advocate a novel
approach: constructively channel human brainpower using computer games.
For example, the ESP Game, introduced in this thesis, is an enjoyable
online game -- many people play over 40 hours a week -- and when people
play, they help label images on the Web with descriptive keywords. These
keywords can be used to significantly improve the accuracy of image
search. People play the game not because they want to help, but because
they enjoy it.
I introduce three other examples of games with a purpose: Peekaboom,
which helps determine the location of objects in images, Phetch, which
collects paragraph descriptions of arbitrary images to help
accessibility of the Web, and Verbosity, which collects common-sense
knowledge. I also show that, in principle, every problem that could be
solved by a computer, today or in the future, could be solved using
enjoyable computer games.
In addition, I introduce CAPTCHAs, automated tests that humans can pass
but computer programs cannot. CAPTCHAs take advantage of human
processing power in order to differentiate humans from computers, an
ability that has important applications in practice.
The results of this thesis are currently in use by hundreds of Web sites
and companies around the world, and some of the games presented here
have been played by over 100,000 people. Practical applications of this
work include improvements in problems such as: image search,
adult-content filtering, spam, common-sense reasoning, computer vision,
accessibility, and security in general.
Thesis Committee:
Manuel Blum, Chair
Takeo Kanade
Michael Reiter
Josh Benaloh, Microsoft Research
Jitendra Malik, University of California, Berkeley
Pepsi pays Amazon 3 cents for product placement. You are shown an image of a Pepsi can. "What kind of soda is this?" "pepsi", you answer. You get paid 2 cents.
Maybe it's just me, but it seems that it's not really worth it. Consider the following task, for example:
Guess how much you get paid for that. 2 dollars? 3? That wouldn't be unreasonable, I think, considering that you're supposed to write an entire product description from scratch for which additional "research" is required. The actual amount paid is only 65 cents, though.
Maybe it's just me, but if I check to see how much I need to work in my regular job to make 65 cents, then it does not make any sense to invest more than a few minutes into a task like this, and it seems that it would take more than that to actually complete it. The fact that there's a review required afterwards doesn't exactly make things better, either - if what you did gets rejected, then you've essentially worked for nothing (I wonder if there's anything that keeps amazon from still using your description in this case, too...).
In other words, the whole thing seems like a good idea in theory, but it won't really take off until they're willing to actually pay you a reasonable amount.
quidquid latine dictum sit altum videtur.
following HIT: "Is there a goat in that picture?"
Well, mturk.amazon.com redirects you to www.mturk.com... seems to imply something.
$nice = $webHosting + $domainNames + $sslCerts
Since this is all web-services driven, it seems to me you could create an interesting cycle with a simple program:
1) Use the API to find a HIT, and sign up to complete it.
2) Create a new HIT that basically asks someone to complete the first HIT,
only for $0.01 less than the original HIT was offering.
3) Do this for every existing HIT.
4) Profit?
Keep your friends close.
Keep your enemies in a little jar on your desk.
Amazon has a page on their site about MTurk
Every post I make begins with the assumption P=~P.
The reminds me of the Philip K. Dick novel in which the main character thinks he lives an ordinary life, and who solves the daily puzzle in the newspaper every day for cheap entertainment. In reality, though, the whole town he lives in is a front, and the fun puzzles he's solving in the newspaper are actually cleverly disguised military strategy problems of some sort.
Quick -- someone patent that storyline and sue Amazon for infringement!
Don't say I didn't warn you.
Soylent Green is peoplicious!
We're grateful to have been Slashdotted! Our beta site, mturk.amazon.com, is experiencing the Slashdot effect. You can still read about Amazon Mechanical Turk and its web services APIs at www.amazon.com/webservices. Also, send a blank email to aws@amazon.com if you want us to email you when page load times recover. The Amazon Mechanical Turk Team