It's nice to learn, Gustavo, that CDF can be made to work. But it requires CDF-Enterprise and, as far as I understand, that is quite expensive so that only large companies and institutions would have it. Even though I bought the Professional version of Mathematica and Premier Service for many years I would not have it available. Nor a second computer at the moment. I did have my CDF files working at one time but WRI closed off the path that allowed them to work.
So, in my view, CDF is dead in the water for what might be considered its principal application - technical communication between scientists. Look at http://arxiv.org/ and show me examples of CDF files or even Mathematica notebooks.
I did watch the Yu-Sung Chang video, which is fairly daunting. I don't think he ever answered the question of how one would make one's own packages available to readers. If WRI provided more liberal usable CDF facilities, it would succeed - and then largely fade because everybody would buy Mathematica and communicate via notebooks.