In fact I think that many socalled Web2.0 apps significantly violate RESTs architectural principles. I agree with Dan – REST is not a distinguishing feature – at least technically – between Web 1.0 and Web 2.0. I think Paul Graham nailed it to a large extent – Web 2.0 is how the web was meant to work all along. Nothing technically that interesting triggered this “movement” – it was more that people just started to “get it”.

So to this extent you may be correct – the REST style is happily gaining some traction amongst Web 2.0 proponents and to that extent I may agree with you associating REST to Web 2.0 more than to Web 1.0

BTW – where I disagree with Paul Graham is that I have no problem with grouping a collection of (sometimes) tenuously connected technologies and principles under the banner “Web 2.0″. It may be a little misleading but so what – I think it is a significant thing for the industry that things like REST and “programmable web” start to be discussed more prominently. Giving these collection of ideas a catchy buz-phrase may stimulate the hype curve – but hopefully it also speeds up a realistic adoption once we start descending the peak of the hype curve.

I think the one thing that you are overlooking is that systems pre-web2 were implicitly attempting to become client/server despite the fact that c/s was designed for running complex applications and web1 was not. Prior to REST application builders were either attempting to replicate state in the client or designing systems that really did little more than provide a nice UI to a server application. There’s a lot of history here and as you point out, a lot of overlap.

I’d disagree with that statement in respect of REST. Fielding’s original dissertation goes back to 2000 and defines a _way of using_ existing web infrastructure (http, URI’s, servers). i.e. REST is a set of design principles _not_ a technology. And those design principles are really a summary of best practice for web applications based on acquired experience of building systems to run on top of web infrastructure. i.e. REST is not a web 2.0 thing, it’s been around longer than that – it’s now more visible because it’s been explicitly identified and given a name.

And whilst REST is well proven in the case of hypermedia delivery to web-browsers driven by human beings there’s a good deal less proof that it’s a good general solution for designing systems that don’t interact with human driven systems i.e. interacting with back-end processes etc.

In the end, I think REST will stand alongside a number of other design approaches all of which should be mixed together to construct any given system.