As was said earlier by apmyhr, I believe many of these comments are due to a lack of knowledge in extremely large scale network deployment and government/military operations.
1) Windows is easy to use and most US citizens are familiar with basic operation reducing the amount of training required per soldier.
2) There are many qualified (and many more unqualified) technicians both active duty and civilian. Moving to a proprietary UNIX system would reduce the experienced active duty technicians by a severe amount without having an additional pool of experienced civilians to recruit. Right now it is easy to find an Exchange/SQL/Sharepoint/ASP technician or developer as the civilian market is inundated with these personnel.
3) There are thousands of programs being used on military systems designed for Windows, some written by army programmers, many more commercially. Every single one would need to be ported or designed from scratch on a proprietary UNIX system. Or some from of Windows emulation would be required defeating the purpose. Add independent support for each one we would have a collective IT department larger than the current fighting force.
As for the move to Vista, support for the OS is VITAL to our military. Imagine a vulnerability is found in XP (not too hard to imagine), but XP is no longer supported. Now add hundreds of thousands of computers in the Department of Defense susceptible to that vulnerability without any hope of a timely or even promise of a patch/fix. The move to Vista is logical insofar as national security requires an updatable OS. And based upon the sheer number of patches I see every week I feel better moving to Vista.
Although my personal preference would be to move to a virtual environment with thin-clients. Most users are e-mail/web users with a smaller percentage requiring intensive graphics (falconview, full motion video, audio, ect..) or external devices initially incompatible with thin-clients (ID card generators, proprietary military data readers, ect...) Would make patching fast and easy, mobility is increased with TCO (read your tax dollars) reduced. 1/3 of computers are life-cycled (replaced) every year for upgrades. Thin-clients would significantly reduce the edge life-cycle cost (although some would be refocused to terminal server upgrades). There are some cases where thin-clients are already used, but is very few and far between.
The point is, the Army (and the rest of the DoD for that matter) are doing exactly the right thing, even if at first reaction they seem crazy. The Army is just too big to not do the simple thing.