Let's posit that games are fun due to players exploring some kind of game space. IE, doing interesting things inside a game. For the sake of this essay, assume some variant of that statement is correct, because I don't have the time to argue it today.
If that's true, then there's really two approaches - both of which are used in every game.
One is to come up with new and interesting kinds of play space. Letting the player do something new and interesting which creates a new game space topography. Something like a gravity gun, or gaining skill points, or whatever.
The other method is to use tried and proven game spaces, allowing the player to return to a familiar setting and explore new niches of it. That would be the genre, for example.
The standard design conventions aren't simply blind adherence to an ancient and undying law. They are a place that many gamers call "home", and by starting the game in one of these places, the gamer immediately feels at ease and excited. He isn't going to have to struggle to survive the first two minutes of this game - he already knows the first two minutes of this game. But he knows that it will get exciting. It will give him new and interesting play experiences.
To achieve this end, you don't even have to use new play elements. You can simply use tried and true conventions in somewhat different topological arrangements. Focus on different pieces of an old convention, or combine two conventions in a new way.
I'm not saying new is bad, but I'm saying that old certainly isn't worthless. When a car manufacturer wants to design a popular new car, they don't say, "well, let's start from scratch. Is there an alternative to these round rolling things?"