Finding effective ways to orginize files isn't so bad.
Like for example I use two programs together (I am a linux user obviously): One is called "ripperX" the other is "Zinf". You could use any two programs of this type...
RipperX is a ripping and encoding program for Linux. It pulls sound files off a audio CD using a 3rd program called cdparanioa (that is specificly designed to recover audio files and compisates for things like skips and cdrom jitters to create a 100% quality output). It then takes these .wav files and encodes them into ogg vorbis compressed audio files.
I have them all go into their respective folders based on almum name in a folder called: Music.
Ogg Vorbis is like mp3's in that it is a lossy audio compression format, but more importantly they store information about the album, artist, genre, etc etc in the file's header. RipperX uses cdda server to stick that info into each file.
All this is done in automated steps.
Zinf is a itunes-like program (formally known as freeamp) that has a automatic way of finding songs. I tell it to look in my "Music" directory and then orginizes a "MyMusic" listing by artist and album. It looks much like the explorer mode of windows desktop were you have the + and the - buttons on the tree. (also can have internet feeds and other stuff)
Then from that you can generate new playlists by double clicking on the song/album/artist/or entire database names. Then it lists it by album then artist then track number. then you can randomize, edit it, and create saved playlists. A lot like making mix tapes.
All this is very simple. All I have to do in the whole thing is press go on ripperX and have zinf scan for new songs. Then select the songs or playlists then play the music.
NOW that is a good way to use a database to orginize stuff.
I'd bet that windows longhorn is going to aim for the same way of doing stuff, but with EVERYTHING.
Your pictures may not be automated at first, but I bet if you make any new pictures using the apps in longhorn then it will then have all that info recorded.
However not all file formats like mp3's have the data built in. Like jpeg. And I doubt that MS is stupid enough to try to force everyone to adopt a new jpeg format.
So they are probably going to do something like apple does (which I don't like) and create a new NTFS-based format like HFS+ (apple's format) does.
Each file becomes 2 parts. The data part and the metadata part. Now in Macs this is totally transparent. Each file is REALY one file, just 2 parts. When transfered over to a format that doesn't support that format it simply strips the metadata away or puts it into special files.
The metadata contains things like file type, creation date, program that made it, what to open the file up with, etc etc. That lots of files don't have extensions like Window's files and the OS simply "knows" what they are. In Apple land it makes this very slick and easy to deal with, but makes networking and interaction with other OSes a bit annoying at times.
So I figure that's what the NTFS extensions stuff people are talking about. NTFS already supports metadata a bit with stuff like ACLs and the like.
That way a person can create a file on one longhorn machine then transfer it to another and not loose the info. But you will lose that info if going to a Linux machine or a Windows XP machine.
Now on top of that I figure Longhorn is going to run a functioning but minimalist database. That way it can keep track of the orgination of the files and the filing system keeps track of the file information.
Probably store the information in a binary form in the database's own form of the registry.
That way when you make a jpeg file (for instance) it will ask for any "extra info" and then encode your user name, file creation date, etc etc. Then it would place the file in a "mypictures" orginization. It will catalog it with any sort of information you will find usefull. Real world location, what album (family, work, taco bell advertising project, etc etc).
Then if your using frontpage you open up the picture orginization and it will have that already orginized and ready to go. Only jpegs and giff become aviable and you press "show this album" and it will rotate around a bit and find what you looking for. You'd have date name and revision numbers to help you locate files.
Then each filetype (or actually information type, written, movie, finacial, image etc etc) will have it's own specific orginizational method. What every MS thinks that will make it's programs seem most usefull.
The apple "finder" program maybe does this already a bit, now that I think about it. (wouldn't be the first time MS ripped off apple's ideas.

) But it doesn't orginize stuff exept by real world location.
Now the parts I don't like would be that the OS will attempt to divorce you from the filling system. Very few tools will be aviable to manage the files, and files will probably be stored in a way that would be virtually unmanageable anyways.
Plus with the DRM would tie your files in with each particular machine making a pain in the a** to do anything with your files that the orginal authors didn't think of or want you to do.
One big trouble is that the more rigid and controlling the OS or a program is the usefullness of it decreases rapidly.
Also you could imagine the pain in the butt when a worm can use the database stuff to infect and automaticly spread itself thru the easy-to-use internet.