It' an optional section, but I think it is pretty interesting. A "hard object" (so named by Douglas Crockford) uses closures to provide truly encapsulated objects. Here is an example.
The object has methods, and it doesn't leak any implementation details, other than what the methods provide.
The object state consists of the variables that the createAccount closure captures.
Should you do this in your own code? Maybe not--although I have done it for one project and found it quite satisfying. And Crockford swears by it. But I think most programmers will prefer the class syntax, even if it doesn't give true encapsulation.