I definitely recommend Kyatric's advice.
My space shooter is based on his 'Asteroids in 100 events or less' tutorial, and reading through that .capx and then modifying it to include enemy ships with AI and whatnot helped me learn a lot about developing with C2. I definitely recommend a similar approach, since starting with a completely clean slate when trying to make something as simple as Pong can be daunting for the first-time C2 user.
My open source turn-based strategy game is pretty much developed from scratch (although Yann, cvp, and vee41 have made some large contributions). It is basically a remake of an old Shockwave game I enjoyed playing, and I definitely recommend cloning a small game because it will allow you to not have to have great design and instead focus just on the mechanics (designing games is hard to do effectively without prior game-making experience - anyone who thinks it's easy has likely never shipped a game).
I'm actually in disagreement with many of the people here who recommend developing small prototypes. For me, making working demos of various concepts is not hard - it's the polish like full-fledged inventory systems, cutscenes (and even scripted, dynamic menu backgrounds) that will demand the most of your work when you actually want to finish a game, and by working with the goal of completing something, you will gain experience in the area of polish whereas with prototypes you likely will not.