There's two services here that need to be separated. Doing this as customer acquisition seems to be what most of the commenters here are focused on and does se

There's two services here that need to be separated. Doing this as customer acquisition seems to be what most of the commenters here are focused on and does seem like a poor deal. On the other hand, 1000 software testers for the price of an iPad are just a couple of quarters each. Plus, here's where their unreliability becomes an advantage. Bob's drunk uncle might be a lousy customer, but the aspects that make him a terrible customer make him a great QA agent. He'll forget his password and create fifty new accounts, then demand that they be merged. He'll post comments in EBCDIC. He'll declare he lives in the state of Oregon in the nation of Uganda. He's never heard of Paypal. After launch, he's a support nightmare and you'll be happy that he disappears the instant that the service is no longer free. However, during the beta, a thousand drunk uncles will find all the edge cases that you never even thought to unit test. They'll force you to tighten down your UI to the simplest possible level of readability. They'll stress test your support infrastructure. And they'll do it for half a buck.