A review by Matt Stephens of JavaFX points out exactly what I think are some of its shortcomings. If I were to highlight the craziest part of all that Matt agrees with me on is that the mid-game switch in syntax, after books were being written on the subject, damaged the usability of tools, damaged the value of books (made them near worthless), and damaged the usefulness samples out on the web. Folks pull up samples and they don't compile. Folks get a book and its syntax descriptions no longer match. Folks pull up tools and either have an old version that won't compile new stuff or vice versa.
Additionally, JavaFX has seen two JavaOne conferences come and go with no formal release. Take this in contrast to the ever climbing version numbers of Silverlight and Flex and you can see that Sun is late to the game with a weak solution that keeps getting modified and doesn't have concrete tools or books out yet. Draw your own conclusions based on those facts.
Lastly, if you love to give up your privacy in even more ways, then Project Hydrazine (you know, that deadly gas from the spy satellite that came down unexpectedly last year) will let you spill your usage guts to Sun.
Final quote from Matt:
"A bigger problem for Sun in the short-term is getting JavaFX - the building block for Hydrazine and Insight - out the door. A year since announcing JavaFX, Sun had nothing to offer JavaOne but shipment dates and shaky demos based on Java SE 6.0 update 10, which kept crashing during the JavaOne keynote."
Comments like that must really irk Sun..