I would like to point out that this kind of development is by no means a new thing. Just look at roguelikes, RPGMaker games, GameMaker games, BYOND games, open source games, etc. These games were always community driven, and developed out of love for the game. There would never be a point where the game would be "finished", because developing the game was the goal! Minecraft simply was one of these kinds of games that happened to be picked up by the mainstream, and then the developer decided to charge money.
I think what really needs to happen, is for publishers like Steam to create decent programmes where such "game enthusiast developers" are properly supported. That is, methods for developers to come in touch with communities, *without* having to actually sell a "popular product", as is the case with Steam Greenlight. Steam Greenlight produces crap like Guise of the Wolf, what we need is the ability to try Guise of the Wolf for free, then decide whether the concept is good or not, then decide whether to fund development of the game.
Another thing that needs to happen in this regard is milestone-based funding. I mean, this is already pretty commonplace in the IT industry. You/your company is hired to implement some kind of software. But if you get all the money in advance, your employer can't "pull out" if he realizes you're doing a shit job, and if you only get the money afterward, your programmers will have nothing to eat until the product is finished. So you make a contract about implementing part of the program, and you get paid for that. That way, you can pay your programmers straight away, and if you do a poor job, your employer won't hire you to finish the rest of the product, so you'll be motivated to do a good job.