Recently I've been observing a definite upswing the in the use of REST-friendly words in the Web Services community. REST has been mentioned without attracting a huge amount of criticism in the WSA mailing list, the WS-Description WG has been grappling to support RESTful interfaces in WSDL 2.0, WS-CAF have been mooting ideas about dereferencing applications in a RESTful way, and even in the WS-GAF mailing list people are now talking about REST. It seems that REST has become the next step up from the current "procedure calls via documents" approach that Web Services are taking - which I took issue with in my recent talk to the ACS.
While I think the REST camp has got some things right, such as a focus on message-exchanges instead of APIs, I think that the current REST love-fest is worrying. Sure REST's message-oriented approach is sensible, but it has the downside of mixing the semantics of the SOAP message with those of the transport protocol. For what it's worth I believe that the semantics of a message are a function of the Web Service (or application) that processes that message. How the message gets transported is really quite unimportant in terms of functional requirements for an application.
For example you might receive this post in HTML format in your Web browser, you might receive it through the XML feed in your blog reader, or you might get it forwarded to you by fax or email. Whichever means the message takes to get to you is unimportant because the message is still going to convey the same semantics - that REST is a point on the learning curve and not the end of it as this unscientific diagram shows:
My preference, of course, is to move the best practice in Web Services forward to the "processMessage" model (I coined the term "MEOWS" - Message Exchange-Oriented Web Services but a better acronym would be welcomed), which like REST has a message-oriented flavour, but unlike REST doesn't hinge on the use of an API, not even a uniform one. This was the central tenet of an article Savas and I wrote for Web Services Journal which was more important than the WSDL article which caused such outcry :-)
As I noted before, Savas has started some weird cult around this idea.