Is there any evidence that they are more likely to fail than non-commercial projects in general? There will be countless examples of projects that people start, but do not end up with a finished project - indeed, I'm sure even experienced developers who have released non-commercial games, may have their own "failures" in their history. Not everything I write ends up being something I finish, this isn't necessarily a problem.

(Not to mention that plenty of commercial products "fail", if you count everything that a company might start on.)

And if you want examples of outlandish unrealistic goals, just check out the Help Wanted - plenty of "we're going to make an MMORPG" from complete beginners, which aren't "fan" or clones. Actually I would argue that clones have a better chance of success, as you've got existing products that you know are fun to play, to compare against. Where as competely original games have the risk of everyone disagreeing over how to do something, or they finally complete the game only to find it's no fun to play. It's no surprise that vast numbers of even commercial games tend to stick with tried and trusted rehashed ideas.

Is there any evidence that they are more likely to fail than non-commercial projects in general? There will be countless examples of projects that people start, but do not end up with a finished project - indeed, I'm sure even experienced developers who have released non-commercial games, may have their own "failures" in their history. Not everything I write ends up being something I finish, this isn't necessarily a problem.

(Not to mention that plenty of commercial products "fail", if you count everything that a company might start on.)