WS-* is an easy target in this respect. I hardly know a (Java) developer who mastered the WS arena. Not because providing web services wouldn’t be useful but because it’s just so complex, confusing and certainly _not_ fun.
In my experience there’s one magical turning point for enterprise architecture: the deadline. I’ve seen more than one project throwing corporate strategy overboard when faced with the decision: delay the deadline or use working code now but without the full might of the corporate architecture. And I’m not talking about flushing the entire design of an application. It was more along the line: company policy: “you’ve got to use J2EE application server xyz”; state at the deadline: “we’ve got a working version on this servlet container but deployment on J2EE container xyz shows a couple of obscure anomalies.”; solution: “take the working version in production now, we’ll abide the company policy later.”; of course it was never deployed on the company J2EE container.
Right now the backlash is developer-driven but before too long it’ll be customer-driven.