I'd like to see it attempted (or am interested in being the guinea pig if need be) of building almost an entire service the way I described. It works for the project I'm thinking of because I work with the law and it's jurisdictional and categorical. To devote finite resources to a different area of the law or different jurisdiction might not be worth it if there is such demonstrable demand to expand features within an area/jx already present. For example, I could build a product for attorneys and law students in my home state that focuses on a small area of law and never expand outside of that area/jx - if those customers/clients are gracious enough that they can effectively block any attempt to expand elsewhere, why wouldn't I reward them? Conversely, if another area or jx can steal my attention with their efforts...
Sure, you could use surveys and contact forms but even if the amount you're raising is not correlated to the development cost, you're awarding your most valuable customers first - those that are willing to put their money where their mouth is. For example, if every major feature, regardless of actual or estimated cost, had a $2000 Backer page - the order in which the Backer succeeds would determine priorities.
Obviously, I'm just spit-balling b/c I just saw this but it's got me thinking. I wouldn't want it to look like a nickel-and-dime operation but one where it actually makes sense to use this method. I think the law lends itself well in that situation.