[anigbrowl]

That's why many projects use a bunch of non-product rewards where the costs are unpredictable, like t-shirts, stickers, and other stuff whose costs are completely predictable and which serves a sort-of marketing function, without being dependent on the core technology.

Then every dollar from those rewards can go into development before getting to the manufacturing stage. Those lower-value rewards can easily be fulfilled, and if you realize there's some sort of fundamental problem with the project you have the possibility to shut it down and refund device payments before that pledge money is spent.

I realize that this is a very idealized approach and that in many cases the amount raised from ancillaries isn't that much, the bulk of funds come from potential buyers for the device. But perhaps that's part of the problem. I'm considering doing a crowdfund project myself later this year (for a film) and although I'm months away from making any decisions I am trying to think through contingency situations now.