There was a parallel with the problems you mention in that the www looked like eye candy but didn't deliver much obvious new functionality; however, the eye candy served the important function of making things much more attractive and accessible to an audience, whereas effectively using the services mentioned above required a kind of command-line machismo that most people had no wish to acquire - and so those services which delivered information of equivalent or better quality to that available on the web withered away because their cost:benefit ratio was too high.
I think TB-L is on the right track here because although much content of value is silo-ed in multiple ways, once there is enough good open-source (or 'linked data' or whatever you like to call it) material available, people will either abandon or never get into the websites which provide it.
As far as I understand things, linked data is just a euphemism for RDF (or RDF + microformats), and there's no shortage of open source RDF implementations.