Wednesday, July 8th, 2009...4:45 pm

Distributed Labor and Amazon Mechanical Turk

Jump to Comments

Yesterday, one of my favorite Berkman fellows, Aaron Shaw (we share a love of North Oakland), gave a brilliant talk at the Berkman Luncheon Series on the research he’s doing on Amazon’s Mechanical Turk. Since you can watch the video yourself (and I highly recommend that you do) I won’t spend too much time repeating it, but from what I gathered Aaron is basically trying to see if it’s possible to use Mechanical Turk as a survey platform for academic (or other?) research. Intrinsic in his research is a study of how Mechanical Turk itself works and what issues this new labor market raises. (Apologies to Aaron if I totally butchered his project).

For me, the most interesting part of the talk/Q&A revolved around these various issues. Namely:

*Are there any implications for labor policy here? When you’re paying someone 2 cents for 5 minutes of work, that’s a steep devaluation of people’s time. If sites like AMT take off and become a real labor force, will we see government getting involved? Legislation/law suits? Labor organizing?

*How does this differ from volunteer work? I’ve heard from a lot of people who work on various crowdsourced projects that most people would rather work for free and see themselves as volunteering their time to a cause than get paid less than market value for their skills, which is insulting. There’s something about inserting money into the equation, even if the difference is only a penny, that changes the motivation. I know people have done a lot of thinking and writing about this and I’m just really ignorant to it (Wikinomics and whatnot), but I find that dynamic really interesting. I was curious to know whether there are any examples of sites that “pay” people in some non-monetary but still tangible form. Like rewards points or something similar. Aaron didn’t know of any, but I wonder how that would change the dynamic? You’re still getting paid for your labor, it’s not an altruistic act, but is there something about taking the actual cash payment out of the equation that makes a difference in motivations?

*What’s the impact of disassociating a task from the project? This one was covered mostly in an off-the-record meeting of the Berkman Fellows directly after the lunch, so there isn’t much on the video. Jonathan Zittrain attended that meeting and pushed back with a much less rosy view of distributed labor. His critiques centered around what happens when a task is broken down into such small pieces that all meaning is lost. The laborer has no view into what he or she is building or contributing to? What impact does that have? He pondered the spectrum of that impact from the loss of craftsmanship to the potential for bad guys to engage a mass labor market to help build towards some nefarious cause (ie: having people circle all the hospitals in a satellite photo in order to identify bomb targets). The discussion touched on potential barriers to bad guys, like putting more of an onus (normative or legal) on sites like AMT to police their job requesters.

I’d say that JZ’s fears are probably several years, if not decades, out. At this point, AMT is a really small site and it’d have to get much bigger and competitors would have to join the game for the threats to become really relevant. And who knows what can happen in the meantime? As for the uneasiness of the devaluation of workers’ time and what this means for the labor market, I find it hard to get too worked up on that front. All of these people are joining AMT voluntarily, and, as one luncheon guest pointed out, they’re probably performing tasks while they’re at work making a real wage. If they’re offended at the going rate, they don’t have to participate. i do think it’s fascinating to think about how this might change labor markets or be reflected in union organizing, but I’m skeptical that these sites will get big enough to make a difference in the real market. In the end, as afraid as we are of robots taking over the world, I really don’t think there’s any replacement for human labor. There are only so many projects you can break down into miniscule tasks; it doesn’t seem to me that AMT is going to save us from our desk jobs any time soon.

UPDATE: Please Note: This talk incorporates research-in-progress from the Berkman Center’s Online Cooperation Research in collaboration with Daniel Chen and John Horton. After the event was over, Aaron realized that he neglected to explicitly acknowledge Chen and Horton’s invaluable role in the project during the presentation. Aaron feels terrible about this and sincerely apologizes. He also hopes that you’ll visit their websites (links above) and read at least one of their papers. Daniel and John’s contributions to the field of experimental research on online labor markets include (a) recognizing that AMT could serve as a venue for experimental studies; (b) conducting the earliest labor market experiments on AMT; (c) solving a bunch of difficult problems so that they could make valid causal inference based on the results of these experiments.

1 Comment