Stefan Tilkov has written an article addressing doubts about REST, it's posted on InfoQ.
He delivers a top-10 list of weaknesses of REST, and then proceeds to address each one. These are things like, lack of tools support, lack of a formal contract language (WADL isn't real (not yet anyway)), it only works with HTTP, and there's no support for transactions, reliability or pub/sub.
The things Stefan says in the article are all very reasonable, and I can't really disagree. Some of the "doubts" he raises and then dismisses seem to be negligible, ot not even valid. For example, "Who would actually want to expose so much of their application’s implementation internals?". But those don't take away from the article. Still, something isn't quite right about the article itself. It seems to be written for people who don't get the Zen of REST. It's like comparing a cheetah and an eagle and explaining why a cheetah does not have wings, or why an eagle does not have four legs. If you need this explanation, you understand neither an eagle nor a cheetah.
It seems to me the whole point of REST is that, it lacks tools, it lacks pub/sub, it lacks transactions, it only works with HTTP. It's quick and dirty and it just works, and that's all there is to that.
Nice to see Tilkov mention the Microsoft ADO.NET Data Services as well as the WebGet stuff in WCF for .NET 3.5.