[yummyfajitas]

This is a non-starter for several reasons, mostly relating to spam.

First, you can't just give people cash. They will immediately turn to spamming, and you'll get crap data. You need to do statistics, comparing turks to other turks, and only pay the ones who don't spam. It's highly likely you can't do this in realtime.

Second, you need to track the identity of your turks. If worker X is known not to be a spammer, then you want to assign work to X preferentially. Similarly, if worker Y is a known spammer, you want to refuse him work.

Third, most poor people are not working and are not looking for work. Only 30% or so of poor adults are in the labor force at all. Why would they decide to start working at a mechanical turk station when they seem to have little desire to work anyplace else?

[notatoad]

"most poor people are not working and are not looking for work. Only 30% or so of poor adults are in the labor force at all. Why would they decide to start working at a mechanical turk station when they seem to have little desire to work anyplace else?"

just because somebody is not looking for a structured job, doesn't mean they won't do tasks for money. think about the homeless guy who wanders around with his shopping cart collecting discarded cans all day. the only social issue i see with a turk kiosk is that it would essentially be a "do things in exchange for drugs" terminal.

[wilfra]

It is not a non-starter for those reasons, it may just require some modification if they become problems or if it is deemed they are obstacles too big to even try to overcome.

When getting instant, reliable cash is the result, people will jump through an enormous number of hoops - and even pay money themselves the for opportunity.

For instance, you could require users to create an account on mturk and/or w/ the mturk shop so they can be tracked. If it is simple to do and prompted it should be fast an easy. If they can't even figure this out, they probably aren't qualified to be doing the work anyway.