Yihong-Ding makes a good effort in explaining the web evolution. It starts off with the web 1.0 (read or write), web 2.0 (read/write) (all about the user control through the smart use of databases) and the tries to put semantic web in the next step for the evolution.
“(…) an ideal semantic web is a Read/Write/Request Web. The fundamental change is still at web space. A web space will be no longer a simple web page as on Web 1.0. Neither will a web space still be a Web-2.0-style blog/wiki that facilitates only human communications. Every ideal semantic web space will become a little thinking space. It contains owner-approved machine-processable semantics. Based on these semantics, an ideal semantic web space can actively and proactively execute owner-specified requests by themselves and communicate with other semantic web spaces. By this augmentation, a semantic web space simultaneously is also a living machine agent.”
He also pictures a good model/image of the evolution in the post.
You might also want to take a look at a demonstration of Photosynth to better grasp what semantic web is capable of. It is not exactly on the issue of semantic web, but it has some similarities: