Another interesting ClayShirky piece here :
Basically argues that standards are a kind of all or nothing thing. You either support the standard or you don't. And before the standard is decided upon there is nothing. Interop, is a gradual, piecemeal, process. You can be a bit interoperable with another piece of software. Application A can read a file from B some of the time. And this kind of interop can be both practical and the kind of proto-success that supports further adaption towards better communication.
See also :