Ask HN: Why don't we solve the micro payments problem?

There have been countless times where I've thought up some little idea that would involve paying thousands of people a dime for a second of their time, the problem though is that it isn't worth it because there is a fixed cost associated with transferring money. So I got to thinking that if 1000 people could know, for certain, that they each had a 1/1000 chance of getting $100 bucks (minus, say 50 cents in paypal fees) they would be willing to do "dime work".

Right now I'm thinking of a one way function that could take in 1000 different concatenated seeds supplied by the workers themselves and, even in theory (not just in "there isn't enough computing power on the planet"), would map to a 1 in a 1000 chance to awarding someone $100 that nobody could game. Not even me, the employer.

Ideas?