According to Ong, labor arbitrage is one of the strategies that informs the conditions of governing and disciplining by way of deterritorializing labor. Labor arbitrage breaks apart the traditional relationship between the national labor legislations and the worker as citizen. Ong describes labor arbitrage as “the latest technique to exploit time-spacecoordinates in order to accumulate profits, putting into play a new kind of flexibility" (Ong 2006: 174). Cognitive labor is particularly susceptible to labor arbitrage technologies because computerized division of labor enables the fragmentation oftasks into smaller and standardizable units, allowing their completion by an assembly of workers across the globe (Ong 2006: 161). I believe crowdsourcingis an apparatus of a neoliberal system of exception that signifies a novel instance of labor arbitrage, where online cognitive labor markets are established as aggregation platforms that simultaneously act as a techno-immigration system.
The exploitative aspects of cognitive labor arbitrage are clearly exemplified by Amazon.com's Mechanical Turk crowdsourcing system. The Turker community seems to have varied responses to the claims of exploitation. Some U.S.-based Turkers oppose such claims and state that their interest in Mechanical Turk is solely motivated by the novelty of the experience. This fact could be explained by the seemingly negligible amount of income that canbe earned through AMT for à U.S.-based worker.15
On the other hand, workers from countries such as India or China appear to be mostly interested in Mechanical Turk as a primary income source, though some of them find that AMT undervalues their labor. For example, Rajesh Mago, a computer freelancer from New Delhi, criticizes Mechanical Turk in his blog as follows:
They call the assignments posted by their requester as HITs (Human Intelligence Tasks). So, is the human intelligence worth cents only? LOL! I know no one is forcing anyone to do these assignments but yet it doesn't justify the usage word “intelligence"—a mockery of human brain.16
Mago states that he completed more than 10,000 HITs working for a few hours a day for Mechanical Turk through 2008. He earned \$572.62. His HIT approval rate was98.2%; in other words, the requesters he worked for rejected approximately 2% of his completed tasks, for which he was not compensated at all. According to Mago, requesters do not give any credible reason for their rejection. Moreover, even the payments for accepted works are most of the time delayed, a matter that appears to affect many other Indian Turkers. Rajesh Mago does not work for Mechanical Turk anymore, and, in retrospect, he concludes that:
Mechanical Turking was kind ofaddictive as I always challenged myself to test and experiment and work for low-paying HITs thinking that I will be able to make decent money. But, Mechanical Turk requesters are pretty smart; they had done more R&D than me andwere sure that they would get the work done at the lowest rates or for free!17
Returnof theCrowds
91