Today's WSJ contained an article on Human Computation - that right! Human Computation. This is the label that computer scientists give to system that incorporate low-level human judgments on a massive scale.
Prof. Luis van Ahn of Carnegie Mellon create a game called the ESP game. The apparent objective seems to be to guess what another person is thinking. It works by pairing two random persons and showing the same photograph to both. You are supposed to enter the phrase that you think the other person is thinking. If they match, then both are awarded points. You get 3 minutes to describe 15 photos.
Over 130,000 people have spent hours playing this game! The kicker is that the real objective of Prof. van Ahn is to categorize photographs with meaningful tags...all for free! As the WSJ article points out, he is pulling a 'Tom Sawyer' by enticing others to paint the fence of image recognition.
A more generalized effort is the Mechanical Turk by Amazon.com to be the eBay for services. It is named after a fake chess-playing robot popular from 1770 to 1880. Read the Wikipedia entry; it is fascinating! Amazon coined the term Human Intelligence Task or HIT. A requester defines the HIT, such as transcribing a podcast or finding a company on the web. SOmeone can accept the HIT, perform the task. When accepted by the requester, the fee (usually less than an dollar) is deposited in your Amazon account.
Another example is Rent-A-Coder, a marketplace with small but serious coding projects. The fees are typically around $100.
So, what? How does this relate to BI/DW in corporations?
There are a lot of HITs needed to maintain any BI system. The objective is to construct game that will leverage the intelligence of employees and even customers to categorize information, detect trends, and surface issues. Instead of limiting these tasks to the few having responsibility, open the task to a wider audience.
Posted June 27, 2007 5:04 PM
Permalink | No Comments |