I got this article (Wolfgang Gruener, ConceivablyTech) from a Slashdot link.
This patent actually amuses me for several reasons. First, it shows just how (and why) the shutdown process on Microsoft Windows is so long and complicated. As it turns out, if there are graphical programs running, there are 3 different ways for the application to be terminated by force and the shutdown process restarted. That is astounding by itself, but not surprising to many people who use Microsoft Windows regularly. Even afterwards, when a top-level program is hung up, there is a way to abort the shutdown process altogether; that's another factor in the process taking so long.
The author of the article laments the absence of a patent that just shuts the system down (ideally in 5 seconds or so). For one, I would argue against a patent for that, as that would be too simple and wide-ranging to be patentable. Secondly, the diagram in the article leaves out one last reason for shutdown taking so long (which is so well-illustrated in this Linux in Exile blog post): automatic updates which are downloaded and installed after the shutdown button is clicked (but, of course, before the system actually shuts down). So let me add in a corollary to the flowchart provided, a sort of mini-flowchart, if you will.
As the Linux in Exile post said, the Windows Update process holds the computer hostage during shutdown, and given the relatively high frequency of updates that need to be installed at shutdown, this is a major contributor to long shutdown times (especially considering that there are a lot of Microsoft Windows users who do not have access to broadband).
Well, there you have it: Microsoft's long shutdown times demystified. You know what? Let them have the patent. Why would anyone else want to license such a long and complicated shutdown procedure anyway?