My notes from the NFJS expert panel, which was not as good as 2004 but still fun:
Ben Galbraith spoke about the uncertainty principle and paradox of choice. Advises that architects choose something that works, and don't get distracted by the latest shiny framework. The best choices for your project will not be those you would have made if it started earlier or later.
Dave Thomas: "Don't architect skyscrapers for garden huts."
Glen Vanderburg predicts the industry will move from highly-configurable frameworks to smaller more focused tools (this is the "lightweight" meme). I wholeheartedly agree, as one of the dissenters that wanted to see Struts stay simple and new projects for the bells and whistles.
Stuart Halloway went on his usual dynamic language evangelism rants; this year he's hijacked the fashionable "lightweight" meme to advance his cause. Last year Java was brittle, this year it's heavy.
Ted Neward predicted that agile development will go the way of client server. That is to say the current hype is an overreaction that will gradually fold into the mainstream. This observation was so insightful that it motivated me to attend his afternoon talk?, which was quite uninspired.