There would be a trade off here. The more you try to mitigate ignorant voting the more risk you have of corruption. It would probably be best to set the fee pretty low because in most cases ignorance cancels with ignorance on the opposite side of the issue.
I agree that the amount to be paid should not need to be more than 1 NXT but I also think that if the total amounts are to be redistributed evenly then someone throwing in a bunch of NXT basically is redistributing their stake as much as influencing the outcome (and if the outcome has only a % chance of directly benefiting them then unlikely they are going to pump that much into it).
This would be a very good thing IF you could have some reasonable guarantee of this part
(and if the outcome has only a % chance of directly benefiting them then unlikely they are going to pump that much into it).
So take as an abstract example, just for the sake of thought experiment, this question up for voting. Should SuperScammer1337 be payed 1 million nxt out of the unclaimed 9million nxt fund for "promotional" use. Send nxt to address 1 to vote yes, and address 2 to vote no. In this example its highly unlikely that the community would outvote SuperScammer1337 in this issue. In economics this is called the problem of "concentrated benefits and diffuse costs".
http://en.wikipedia.org/wiki/Public_choice_theory Now it would be wrong to say that this problem couldnt be avoided. If you have 10 different causes on the ballot than it is unlikely that someone would take the risk of trying to purchase his own victory. The point is that we would need someone with a solid understanding of economics, someone who knows how to think like an economist, to run the voting process.
My basic argument is just that, With the idea of using tokens + a flat fee there is little to no risk and no need for the creator of the ballot to understand economic thought processes.
