Our respondents don't get paid unless they actually explain their answers - which both reduces false responses and provides unique value and insight to the service.
I still think doesn't help. I liked the idea and did the turk for one of your questions (this was my first time doing the turk, btw). I was asked to visit a myspace webpage, listen entirely to 2 songs and say which one I like better and why. All this for 3 cents? I did it, cause I am a good fellow but I have hard time thinking of people who are willing to spend 10 minutes of their time for 3 cents.
the FAQ example is a bad example: I can always answer "yes" and then supply as reason "every questions I had was answered". No need to read the FAQ page, of course.
I think if you want to insert any kind of validation system you would have to count the time it took for the turk to answer your question and make sure we are in that range for every hit.
As noted before, it's tough to know that 100% of the responses are truthful on any survey. With PickFu, you can read the explanations to get a feel as to which respondents put some thought in.
I'm sorry you had a rough experience with that specific music-listening job - we've never had a more time-consuming question submitted. Most are in the order of "which site do you prefer?", like this current one: http://pickfu.com/AGSIBT. These get answered far faster by the MTurk workers.
For example, we used this feature to dogfood our own FAQ page: http://pickfu.com/4UQM7U