What’s behind YouTube and Mechanical Turk?

This is the second provocation on the theme of digital labor from me and Ramesh Srinivasan. To warm up, check out Saskia Sassen at last year’s Internet as Playground and Factory as she warns us about how financial logicians uses networked technologies to manipulate human ingenuity:

Free Use as Free Labor on YouTube

YouTube, subsidiary of Google, serves as a notable example of how a company creates value through free, user-contributed labor. User-producers upload content to YouTube for free and are given the opportunity to freely use Google’s immense, proprietary data centers (commonly called the “cloud”). Adding content, commenting, tagging, and even browsing all add value to the corporate product, though the amount of user investment and creative immersion differs in each of these cases. In the process, content creators facilitate Google’s ability to place targeted advertisements. These advertising schemes are monetized via the billion+ views YouTube receives per week. Commenting, tagging, and browsing are more passive forms of labor, as each adds to YouTube’s ability to build a social space that users will continuously return to, and optimize algorithms that allow for more efficient retrieval and browsing.

While a number of select YouTube partners are being selected for revenue sharing agreements, the vast majority of contributors receive no revenue from the advertising profits generated around the content they produce. As several anthropologists have pointed out, YouTube is also a social space (Patricia Lange) and an educational tool (Michael Wesch, Alexandra Juashz) but it is first and foremost a business that until the YouTube partner program got going in earnest this year was loosing $100,000,000s. YouTube demonstrates the importance of free digital labor in creating profit-making value for a major corporation. But is the free labor users contribute to such sites exploitative? Let’s assess a second example, one in which users are given pennies for networked work.

Amazon’s Mechanical turk (mTurk)

YouTube’s ability to profit from free labor relies on its openness as a system – users can browse, upload, and comment easily. Amazon’s mechnical turk (mTurk), differs from YouTube in that it functions as a more targeted site, linking jobs/employers with potential laborers worldwide willing to work for the price specified. In contrast to free labor sites such as YouTube, mTurk is built around the relationship of an employer assigning a compensated task to an employee who bids on it. This and other digital labor sites (such as odesk.com) connect employers who post simple digital labor jobs (such as spellchecking, color correction, and basic software development) with any laborer worldwide who accedes to the employer’s compensation terms. These tasks are often ones which automated computational systems struggle with, such as image recognition. mTurk laborers can thus be seen as humans working to complement and augment algorithms and systems.

Interestingly, mTurk’s name is inspired by the Englightenment-era chess-playing automaton, “the turk,” which purportedly would play chess against intellectuals and aristocracy in the 18th century. The chess machine magically engaged the chess player or employer. Yet the turk was actually a clever ruse, a machine where a small Turkish man would stand under and move chess pieces that were magnetically linked to the bottom of the table, and often defeating the “machine’s” opponent.

Is mTurk an innovation that positively impacts access to employment for the traditionally excluded? Or can it also be seen as exploitative through the corporate use of this technology to access cheap labor?

Digital labor functions to generate corporate value, yet also can be seen as empowering individual agency by allowing for a variety of uses and interactions by the user/worker. Systems such as YouTube and Amazon’s mechanical Turk, both function in these ambivalent manners based on ethnographic data we have gathered. This blog post traces out some of the details around new media systems and their ambivalent, concurrent invocations of agency/access and exploitation/capitalization.

I am a cultural anthropologist and media studies scholar currently teaching and researching in the Sociology Department at Lancaster University, UK. I investigate media technologies, digital finance, and network activism. @mediacultures

10 thoughts on “What’s behind YouTube and Mechanical Turk?

  1. How do we assess the value I gain from accessing Google and Youtube? Serious question. I feel that both have added hugely to my ability to access useful information in a timely manner. That may be an illusion. How would we go about testing this balance?

    As for the mechanical turk, my suspicion based on the (unbidden) testimony I’ve heard is that it’s more or less only as exploitative as the underlying economic conditions allow. The job market is shitty, it’s harder and takes longer to find work –mTurk lets you continue contributing a few bucks to the household income while you’re on a dry streak. So that’s /exploitative/ in a sense, but unless mTurk is responsible in some way for there being no jobs, I’m not certain the friends I know who’ve used it would be better off without it. On the other hand, mTurk obviously wouldn’t have as large a pool of labour to draw on if there were real jobs for people.

    So, it’s about as exploitative as dollar stores, probably. :) A symptom, not a disease.

  2. Its one hell of a technique, this camera work. I mean, how do you learn to do it? When do you tell yourself as a cinematographer to wander a bit there in that direction, lean in this direction now. Thos are all micro-camera movements–inches. Its very ambulatory and breaks with all formal training but gives it an ambiance that isn’t dissimilar to the music, her mannerisms, the informality of the operation. And its a risky maneuver, to shoot a single camera shot and meander like that. For these reasons, I commend it.

  3. @Andrew Both good points. The question we are posing is not whether firm-based sociality is good for society. There are stacks of tech-pop best sellers for that (Shirky, Howe, Jenkins, Benkler et al). We love and use their tools rabidly and, here is your balance, conceptualize them digital social enterprises. We are worried, as is Sassen in the video, that the logic of the financier trumps the engineer and the civil activist in access, monetizing, and populating these socio-technical tools.

  4. So, provisionally — does it? I like open-ended arguments as much as the next anthropologist, but I feel like I need a little more to go on here.

  5. As someone who works for a company that use Mechanical Turk, I wanted to weigh in on whether mTurk can be considered as “exploitative.” You’re right, the pay isn’t great — although Amazon won’t release detailed data on the subject, we can be fairly certain that the average worker is probably earning about $1-$2/hour.

    The reasons why market rates are so low, however, are complicated. Indeed, there are more tasks available on the service than workers ready to do them — so if anything, its a market that is tilted toward labor. What seems to be the problem is that mTurk’s reputational infrastructure doesn’t allow workers to distinguish themselves in any meaningful way. How much would you pay a worker who hasn’t been vetted, interviewed, and may have no previous experience in the subject? the answer: not very much, and while many companies (like mine) tries to pay good workers more, Amazon does not make this easy. This is a classic market for lemons, where high quality products (e.g. workers willing to do great work for you) are indistinguishable for low quality products (e.g. workers who give you answers quickly, but are usually incorrect).

    Its worth nothing that most workers are only doing 100 or so tasks in a given week, for a grand total of 4-5 hours of total work a week. From what we can tell, very few people are using mTurk as a primary source of income, but are instead using it as a way to supplement their income, spend free time, or even use it out of boredom. The prototypical mTurk worker is either a) an American middle-aged woman who uses the service in her free time or b) a young indian man who uses it to supplement his income. Neither group is making more than $10-$20/week on the service, and about 70% of both groups report being employed elsewhere.

    For more, I suggest you check NYU professor Panos Ipeirotis’ blog on the subject: http://behind-the-enemy-lines.blogspot.com/ .

    Oh, and sorry for the anonymous post — I would post under my name, but I currently work at a company that uses mTurk. obviously my opinions are my own, and not my employer’s.

  6. That camera drives me crazy. But it’s amazing that Sassen can put up with it and continue to talk coherently. That’s an art!

    Sure its a risky maneuver to shoot like that, but anyone can be risky. I think it was a mistake. Better to have experimented in a test shot. The drama of the camera takes away from what is being said, I don’t see how it adds to it.

    I was much more interested in the montage at the beginning and think that that was enough. This reminded me of watching those old Stuart Hall lectures with the crazy ‘cool’ backgrounds of the early 1990s.

    Either way it was a pleasure to listen and good to hear more about digital labour.

Comments are closed.