Scribblenauts is a Nintendo game. The object is to solve a puzzle. Say there's a kitten stuck in a tree. If you type in 'ladder', the puzzle is solved. If you type in 'jetpack' the puzzle is solved.
So it's definitely different than the rest. It's innovative. I'm still not sure exactly what makes it so cool, but I see connections with libraries and potential to somehow apply the cool factor.
I see the game play as a form of searching. It sort of works the way most people assume or maybe just hope that search works. Type in your word and the puzzle is solved. There are many words that will work to solve the same puzzle.
The problem is that search doesn't work like this. Especially in scholarly research. There are multiple places to enter keywords. The keywords that come to mind are often not the ones that will bring relevant results. All relevant results do not share the same set of keywords.
The game was basically built by researching and recording each noun's metadata. Librarians and indexers have done a lot of that work already, so why isn't our version working?