When I was taking a game development class in college, the professors encouraged us to avoid using genres as a way to decide what kind of game we wanted to make. By staying clear of pre-established concepts that we have in our heads about familiar genres, they said, we would be more likely to innovate in our game design.
While I recognize the usefulness of this concept for rookie designers, I'm not sure that this is a good practice for a number of reasons:
1) Genres clarify marketing of a game. Hey, maybe genres generalize your game for consumers, but can you really expect a non-hardcore gamer to understand specific descriptors like "physics-based gameplay"? Give them a genre, and they'll know what the foundation of the experience is.
2) If a genre is a confining box, the boxes could fill a warehouse. Just try to sort reviews by genre at IGN to see what I mean. There are a lot of genres out there to cover all bases -- a hell of a lot more than the movie industry has. At the top level, you don't need to distinguish any more than "Strategy RPG" or "Driving Simulator"; those are pretty damn explicit.
3) At the design level, picking a genre can help focus a designer's vision. The genre can summarize one of the biggest questions to consider when shaping a game experience -- what are gamers playing this for? Do they want to shoot each other? Fly a realistic plane? While I think it is healthy to start building a concept genre-free, in order to make any sort of creative structure, a designer needs to know what sort of experience he/she is building out of that initial concept.
4) It will save the design team and marketers a lot of breath to be able to answer, "I'm making a cool twist on a fighter," than "it's a careful mix of advanced team AI techniques and character/weapon balance.
All in all, I don't think genres are as defining as one might think. There is nothing wrong with starting from "I've never made a platformer. Let's try something new with that genre!"