For me, the disease model is valid insofar as I believe there to be a genetic component that makes one person an addict, whereas a similar person could do the same thing and not become one.
I also believe that once physical addiction occurs, the disease model is legitimate, as it follows a similar predictability in terms of decline, health issues, and other effects. In this way, I believe it is a disease very similarly to diabetes. It is partly dependent on the person who has the illness to contribute to it, but once it is full blown, you can fairly accurately spell out what will occur.
But in regards to the inability to stop a relapse because its a disease, well, as the alanons say, quack quack quack.
