Sunday, December 27, 2009

Crowdsourcing for unlimited spying and more...

KeelyNet"When an ad hoc team of 5000 people who assembled in just two hours found 10 weather balloons hidden across the US by the Pentagon's research agency earlier this month, it was just another demonstration of the power of crowdsourcing – solving a task by appealing to a large undefined group of web users to each do a small chunk of it.

So far crowdsourcing has been associated with well-meaning altruism, such as the creation and maintenance of Wikipedia or searching for lost aviators. But crowdsourcing of a different flavour has started to emerge. Law enforcement officials in Texas have installed a network of CCTV cameras to monitor key areas along that state's 1900-kilometre-long border with Mexico.

To help screen the footage, a website lets anyone log in to watch a live feed from a border camera and report suspicious activity. A similar system called Internet Eyes, which pays online viewers to spot shoplifters from in-store camera feeds, is set to launch in the UK in 2010.

An Iranian website is offering rewards for identifying people in photos taken during protests over June's elections. Some people have declared those examples chilling. Now Jonathan Zittrain, a Harvard University law professor and co-founder of the Berkman Center for Internet and Society, says the next step may be for such efforts to get web users to help out covertly.

In a recent talk, "Minds for Sale", at the Computer History Museum in Mountain View, California, he pointed out that this could be done right away, using Amazon's Mechanical Turk, a service that provides a platform for anyone to farm out simple tasks.

In a speculative example, Zittrain has calculated that, assuming a population in Iran of around 72 million people, it would cost around $17,000 for the government to use Mechanical Turk to identify any arbitrary person's picture, without the users that are doing it realising the cause they have enlisted in.

The scheme would show "Turkers" a photo of a protest, or just faces extracted from one, along with five randomly chosen photos from the country's ID card database, and asked to say whether or not there is any match. Users would receive a few cents each time they contribute.

Furthermore, Zittrain says that such a task might be made into an addictive game, similar to Google's image labeller. "The people making the identifications in India or the US, idly doing this on their lunch hour instead of Minesweeper, would have no idea of the implications of what they are doing," Zittrain said in the talk. "I think people ought to know how their work is being used," he told New Scientist.

Crowdsourcing's power to compartmentalise and abstract away the true meaning of tasks turns human intelligence into a commodity. Zittrain's thought experiment shows how it could potentially entice people into participating in a project that they otherwise wouldn't support." - Full Article Source

No comments:

Post a Comment

Followers