exzellent idea to push this thing forward. I am thinking about new ways for multimedia in the net since two years now. My conclusion is, that the internet is to fast and to impatient for hypermedia, because it requires the visitor to comply with the sequential corset of the content. The power of text is that you can browse it. A strategy which is essential to handle the information overflow in the net.
My theory thus is, that if you do hypermedia, you have to break the sequential corset, which is one of the biggest disadvantages of radio and tv. One way is of course to offer streaming. But fast forward is a bad substitution for text browsing.
You can do it better by simply cutting video and audio in very small pieces (shorter than one minute), and embed them into the text. The text has to carry the audio and video, and the mutimedia has to supplement and follow the idea of the text. Text should be the main media. No one wants to hear a one hour interview. The text should lead the reader through the interview, and deliver at important stages audio snippets.
Following this dogma of nonsequential hypermedia, I've experimented with a more advanced technical solutions of embedding multimedia in text without taking the freedom of indivdual reading speed. I made a prototype of multimedia enriched text, where all multimedia is started automatically dependent of the reading position.
The example is german, and doesn't work on a mac, but you may understand the philosophy and take a look and the technical solution. I've done it by starting the flash files (would work with quicktime too) according to the readers position in the text.