With that kind of expected ROI, you'd better have an idea you believe is worth tens of thousands of dollars, in which case why would you leave the chances of it actually being developed to a popularity contest...
Outside of that, we actually give them back all of the data from the 'popularity contest' in the form of pretty detailed analytics package, so they can refine their idea and take it elsewhere or give us a shot again at a reduced price.
Have you tested the price point? Assuming your business model is ultimately based on selling viable products rather than submission fees, removing the entry fee or lowering it to a nominal amount might actually benefit you in terms of better submissions, particularly since you already have the community to filter for crap.
Our main argument for the current fee has actually been to discourage a huge amount of submissions so that we could take time to evaluate the viability of each submission. However, as we've grown (we're just over a year old now), the fee certainly seems to have become more of a hinderance to good ideas and the community has actually become a lot better at actually judging and filtering out products based on their viability.
Because it's terrible.
More specifically, it's poorly implemented and too high. It's "hidden" until after you finish filling out the 'submit idea' page with a big button on the home page, and it's poorly phrased... "Also, your payment information is required to complete your submission." Oh yeah, like you almost forgot you were gonna do that. Be more up front about the fee (change the layout so it doesn't hide below 'the fold'), and make the fee smaller. You want to avoid people filling out their idea submission, getting pissed that it costs $100, and then refusing to come back.
You should charge something (rudimentary spam-filter), but $100 is too high. The tinkerer has already identified a problem, come up with a solution, drawn their idea on a napkin or even Google Sketchup, is also going to have to weigh that $100 against material costs. As an example - cream cheese that comes in bagel-shaped slices. I could gamble $100 on it at Quirky, or spend $50 in cream cheese and some shop-time to build a prototype. If I'm really convinced that my shelf-shaped cereal-bowls-so-your-cereal-doesn't-get-soggy idea is going to take over the world, I would have a rough idea of the costs involved to fully bring it to market (shapeways.com + etsy.com + reddit/digg/etc, and if it really pans out - use proceeds to pay for injection molding).
If I'm looking to spend $100 at once on an idea, well... it costs $100 for a provisional patent, and tinkerers are well aware of that.
(And that half-baked idea that I'm too lazy to implement? Not worth $100.)
How about sliding scale - pay $20 to submit your idea, but you get less of the revenue?
I hope you guys succeed, if only so I'll have somewhere to buy my dustpan-with-fingers.