It doesn’t matter if you think it’s cumbersome, collecting emails is not permitted on mTurk.
“You may not use Amazon Mechanical Turk for illegal or objectionable activities. Here are some examples of prohibited activities:
- collecting personal identifiable information”
Amazon itself considers an email personally identifying information. I was working with a requester who wanted to get email addresses so he could notify workers when more work was posted. His HITs were pulled as violating the TOS. I asked Amazon directly why, and they said it was collection of emails. He then tried to ask people to sign up to follow his Twitter for the same reason, and the HITs were pulled again.
Not asking for emails isn’t about Turkers’ needs per se, it’s about not breaking Amazon’s rules.
Thank you, dark_bird_of_paradise. This is a vital point.
AMT workers are highly individualistic. Most of them, at least occasionally, work HITs which violate ToS .
But every snowflake is different. Not all ToS violations are created equal, either. Different workers disagree on how far they are willing to go. A recent CMU study https://docs.google.com/viewer?url=https%3A%2F%2Fwww.andrew.cmu.edu%2Fuser%2Fnicolasc%2Fpublications%2FCEVG-FC11.pdf paid workers to download software, but more than half wouldn’t do it, no matter how much money was offered. Some workers may download a program because they’re clueless and hungry, others (less than 2% in the CMU study) may set up a virtual machine ‘sandbox’ and believe their safety is ensured. A few reject it as categorical moral hazard.
But the ToS are defined by Amazon, not by us. At the IRB level, we are dealing (or ought to be dealing) with carefully prescribed research protocols. Part of this protocol is to obey the law of the land, and part of the protocol is the ‘law’ of Amazon, which requesters must subscribe to before they can even be called requesters.
I hope the bête noire counterexample for ethical requesters to mock and critique will always be the University of Minnesota experiment, whose “exogenous manipulation” of Turkopticon’s database probably inspired our current effort (and certainly provided a spur). If Minnesota had enforced its own code of ethics on use of its data center, their project could not have occurred as it did; bureaucratic complacency made that code of ethics a dead letter. Minnesota’s IRB was equally negligent of the relevant state laws prohibiting unauthorized use of other people’s computers and databases.
An amorality like that practiced by University of Minnesota does not need additional fig leaves, sanctioned by this project, in the name of “the workers” or anybody else.
Are workers in a Zyklon/B factory responsible for the final use of the product? Perhaps not … from the perspective of atomized individuals feeding their families, without recourse to collective action.
This project is a collective action. No matter what we may say, no ethical researcher can consciously violate Amazon’s ToS. I’m gratified to see attempts at educating requesters. Some researchers will violate ToS, and some workers will perform their HITs. But today we’re in the business of describing ethical conduct for the benefit of IRBs, not providing a bathroom hall pass for offenders.
A introductory reminder of the legal and ethical underpinnings of an IRB’s role might not be out of place.