New Scientist reports that researchers at Carnegie Mellon University have created an online multi-player game called "Phetch" that captions Web images for the blind. It works thus: One player views a randomly-retrieved picture and constructs a pithy paragraph describing it. Three other players (who haven't seen the picture) use the caption as a starting point to search for it. Phetch is an example of the fragmentation, transportation, and repackaging of labor that a fast Internet makes possible. Whoever finds the picture first becomes the next pithy constructor. Repeat. According to New Scientist, 130 people playing this cooperative game managed to generate 1,400 captions in a week, at which rate 5,000 people could generate captions for every Google-indexed image in the space of 10 months. (Not as infeasible as it might sound: Many people who already do recording for the blind could probably be recruited.)
Phetch is interesting because it's a perfect application for Mechanical Turk, Amazon's instant labor market, in which you recruit people for short, relatively low-skill tasks (taking red-eye out of photographs, for example, or humming). To the extent that your recruits are from developing countries, this approach to labor becomes cheaper and hence more attractive. ("Turk" has fallen on hard times of late, suggesting that the idea may be flawed--perhaps the market it creates is a little too fluid: You may have to renegotiate pay, task, quality, metrics, holidays, grievance process, etc. with somebody new each time you want to do something. That could become tedious and might drive you to a simpler source of labor such as yourself.)
Phetch is an example of the fragmentation, transportation, and repackaging of labor that a fast Internet makes possible. McDonalds tested another approach: The drive-through order-taker is actually off-shore. He or she listens to what you want and transmits it to a display above the counter, where it's read and executed by the local food preparation people. The cost savings are apparently significant, and the best part is that you can load-balance: Order-takers have no "loyalty" to a particular restaurant and can fly instantly to busier franchises, resulting in even greater cost-savings.
Another possibility: A "call center" in which airport scanner feeds are monitored by trained personnel who will press a Big Red Button if anything suspicious comes by. Advantages: load balancing (again), which means you'd need fewer employees; a single feed could go to several monitors, increasing the chances of spotting Bad Things; realistic simulation training could be done in the interstices and unbeknownst to the workers receiving it; and, since monitors wouldn't know which airport they were viewing, the chances of collusion between bad people and scanning personnel would be virtually nil. Local employees would still need to be in the airport for security, of course, but they'd no longer need monitoring skills. Needless to say, as a member of a consumer-oriented economy, I'm violently in favor of this sort of thing because it reduces costs and hence the prices I pay. Just don't go overboard and fragment, transport and repackage my job.