I think the biggest confusion about REST is that it’s not a protocol, it’s a way to think about extremely big information systems. To compare prior models with REST, one has to think about the problem of information management & manipulation in network.
Referring to Pat Helland’s data on the inside/data on the outside article, Stu points out how, similarly to a service in SOA, a resource in REST can hide implementation details. I emphatically agree — at least in general, a resource is more than a DB entity with CRUD operations. This is exactly what makes me slightly suspicious of implementations that simply expose rows in a DB as resources.