Mike Champion says that XML liberates data from the applications that created it.
In simple applications - especially data oriented - this is true. In more complex applications, it rarely is.
Take web browsers for example. Pages for display to a human. The "meaning" is in the presentation - how the content looks. Now, lets say the content is in an XML format and everyone has access to the schema. Does it follow that all browsers will display the data the same way?
No. In any non-trivial presentation-oriented application, there is a processing model which must be adhered to if the semantics of the data are to be faithfully preserved.