The semantic Web in the Berners-Lee vision acts more like a series of connected databases, where all information resides in a structured form. Within that structure is a layer of description that adds meaning that the computer can understand.Since we're on the topic of visions and dreams, here would be my answer: Imagine the lawyer, Mr. X, flipping open his laptop (which by then would be priced similarly to a cell phone), and typing in "2 o'clock meeting with Angela at Starbucks." All of a sudden, his online calendar would pop open and a series of clients names would appear, and the correct "Angela Smith" would be sent an email with details of the meeting agenda sent to the printer. Starbucks would receive an electronic notification with the usual order of Venti Chai Latte (two cups) and a newspaper -- the Globe and Mail (his favourite) to boot. Because Mr. X's car is in the shop because of a recent accident and a replacement car isn't ready yet, a taxi has been order automatically for Mr. X and will be ready for him upon arrival for 1:30 at the entrance. The ride is estimated for 15 minutes to his destination, but his preference has always been for early arrival.
Finally, it's the library's turn now. Mr. X. sends an email to the librarian, (after all, she is the one responsible for the library's more intricate databases), simply with the message "Wang V. Granville LLP" (both pseudonyms of course), and immediately, the librarian works her magic and types in the necessary key terms. All of the acts, statutes, regulations, as well as updated case files relating to the case are electronically retrieved and stored onto a file which is automatically sent to the lawyer's dossier. (The librarian's job is behind the scene - she is the one who carefully collates the materials and gives them tags which the semantic databases will translate into its own readable language).
The lawyer walks out of the firm nonchalantly and begins his afternoon with everything he needs, but taking only one-tenth of the time and effort he would need back in the days of Web 2.0. That, in my hypothetical world based on user history and preferences and interlocking databases, is how the future of Web 3.0 might look like.