Agreed too. Make many small games first. Problem with starting big is that you'll improve so much so fast that you'll want and need to rewrite your game over and over again to keep it on par with your skills. There is no way you can start out with something you'll be able to build your whole game upon (maybe unless you're a skilled programmer already). This is why small games is the best way to start; each game will be better and better until you're ready for the big one.
Also, know that starting too big is the single most common mistakes new game developers do. I did exactly that and almost game up game dev because I could never finish anything. It took off the moment I decided to try something small and now game development is my only income. One week per game max (and put them up on the web), that's a great way to start.