On a technical level it was pretty decentralized. RSS and OpenID specifically did not rely on any particular server. XML namespaces, RDF, "folksonomies" were ab

On a technical level it was pretty decentralized. RSS and OpenID specifically did not rely on any particular server. XML namespaces, RDF, "folksonomies" were about open distributed permissionless participation. XMPP was federated. People got too excited with the Twitter API, but even the social services were in principle meant to be replaceable with rel=me (people were supposed to be identified by their URLs, hosted wherever, not by an account on a platform). I do agree that Web 2.0 was naive in thinking that merely having open decentralized protocols was enough to ensure services built on top of them will be open and decentralized. The commercialized centralization happened despite the decentralized/open tech — on a layer above. Megacorps figured out that open protocols are merely a fertile ground for building walled gardens on.