In my previous post Platform Push, I lamented the fact that I'd may have to redistribute various runtimes and libraries over the next few years with my applications. From a different perspective, having the various technologies available broadly earlier (because Longhorn will ship in early 2006 and because the technology will be available in downlevel OSes) means that I can write to those technologies earlier although with some distribution headaches. The benefits of having technology earlier, IMHO, outweigh the costs of redistributing and installing additional libraries.
I don't know if you caught this but, according to Thurrott of Winsupersite.com, Longhorn will not just ship sometime in 2006, but will be "broadly available" in that year, meaning that it will ship in early 2006, not just in the waning months. This is a stronger assertion than in the past and possibly minimizes the difference in impact to ISVs of Longhorn between shipping in late 2005 and in early 2006, especially with Avalon's new availability in Win XP.
Future releases of Longhorn builds will be beta versions, not alphas. Next PDC is coming in 2005.
I am not too worried about the loss of WinFS, especially if it is going to still be available at the time in which its inclusion would have caused Longhorn to potentially slip to.