Much of the recent news coverage on Amazon.com focuses on their emerging program to let consumers purchase online access to pages or chapters of books. Although it’s a worthwhile initiative, I don’t find it very exciting. I just prefer paper to monitor when reading a lot of material (I might feel differently if electronic paper were more advanced). That said, there are useful applications for Amazon’s new program, especially for reference material: imagine, for example, being able to print a single recipe from your favorite cookbook for $0.25. There are times when I’d prefer to do that than buy the whole book.
Amazon’s Web Services (AWS) initiative, on the other hand, is much more interesting. The program lets developers build new applications based on data provided by Amazon. The data typically comes from Amazon’s current product catalog, though Amazon also sells web traffic information (collected via Alexa) and historic product pricing data.
For example, an independent developer made a dynamic web page showing up to 1000 Amazon products in a single page view. It’s worth a look. He also made a tool to let users create customized, interactive banners for their own sites.
Others have created many more programs–for advertising, e-commerce, and reference, among other things–and all use Amazon’s data in one way or another. Obviously, Amazon could have created web pages and banner creation tools like those shown above. Thanks to their Web Services platform, though, they didn’t have to. An independent developer did it for them. In the process, Amazon generates revenue (from people who visit the page, click through, and buy something) and, at the same time, the developer and shares a cut of the proceeds through Amazon’s affiliate revenue split deal. Win-win.
Similar programs at Google (particularly related to Maps) probably get more buzz, but AWS makes real money. As of last quarter, Amazon had about 1,000,000 third-parties hawking various products (many, if not most, used Amazon’s Web Services); collectively, these third parties generated sales of almost $500 million (via InformationWeek).
Amazon’s most recent Web Service, the Mechanical Turk, is even more intriguing. From Amazon’s website:
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 Intelligence directly into their processing by making requests of humans. Developers use the Amazon Mechanical Turk web services API 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. In reality, 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.
In other words, Amazon’s Mechanical Turk web site, which is now live, is a market for human tasks. Since the market is still new, most of the available tasks were posted by Amazon (over time, others will post jobs, also). For example, Amazon is using the Mechanical Turk to improve the quality of photos shown in their search engine, A9. Specifically, there are tasks on the Mechanical Turk asking participants to select the most attractive photo from a collection of images; presumably, the selected image will be displayed on the A9 site. For completing the job, the participant gets paid $0.03. In another task, the participant is asked to create a new description for a product in the Amazon.com Automotive catalogue. Since that job is more complex, the pay is higher–$0.65 for each description. In theory, the market could list just about any task that can be completed by using a computer connected to the Internet.
So, why am I interested in the Mechanical Turk concept? For developers, the Turk simplifies the creation of cost-effective applications that require human intelligence. And, for certain problems–like identifying objects in photos or evaluating beauty–humans are much more effective than machines. For people who have time and want money, the Mechanical Turk provides a market where work can be easily found.
I have no idea if the market for Mechanical Turks will take off. However, the concept is a wonderful innovation–and much more interesting than paging through an online textbook.
Great post there! Very informative.
Left by Daniel Nerezov on November 5th, 2005