Not particularly funny when you think about it. Open Source is great at generating a metric shitload of code. Most of that code is a hodge-podge of learning exercises or scratch-an-itch sort of thing. Large projects tend to have a huge tendency towards being feature dump grounds with absolutely no overarching direction at all. Volunteer hackers are more often than not lacking in the planning skills and vision departments, even if they are awesome code-producing machines.
Getting a very clear roadmap that all the developers can (mostly) agree on is a very difficult thing to do, especially when you have a ton of developers with highly divergent personal opinions or corporate goals for the project and its supporting community.
Large amounts of code are going to be produced with or without a solid set of goals for GNOME 3.0. The hard part was getting consensus on what code should be produced, which in turn helps direct hackers energy towards more useful things than "yet another Gecko shell" or "yet another file manager" or whatever those hackers might've started working on otherwise.