I shudder to imagine what a big project it would be assembling a soundtrack to High Fidelity, the book.

As for MP3 metatagging, here's a repurposed, long comment I posted on a friend's LiveJournal last fall.

Metatagging has become a Major, All-Consuming Project for me in the last year, since I bought a 4G iPod with a nice screen and decided to get anal about my metatags. Although I had always been careful about fixing flat-out typos when I imported songs, I was often lazy about tags like genre, year of release and even, occasionally, album. I also started adding album art JPEGs to my songs in iTunes to take advantage of Apple's enhanced tagging technology and my new color screen.

When I first started ripping, I hadn't really come up with a scheme for organizing songs into albums vs. playlists. There are still a number of songs tagged with names of personal compilations of mine, like Jay Smith's Now We Are Thirty-Three (a birthday mix from '04) or Chris's Non-Stop Erotic Cabaret.

One of the things I've discovered, as per your metatagging nightmare, is that year of release is one of the most poorly annotated metatags on the web. In other words, the dates provided by all of the major tag providers (from Gracenote to Amazon) are riddled with errors. Recently a friend of ours was building his own application to collect metadata, and he tested it out on the J. Geils Band song "Centerfold." The metadata providers on the web offered 1990, 1994 or 1996 as suggestions – instead of 1981, the correct answer. This is the result of a web-wide tagging system entirely tied into CD release patterns, which often do not conform with original album release dates. In short, it's a music geek's (read: my) nightmare. So you end up at AllMusic a lot, double-checking all releases of music that came out before, say, 1990.

As for my collection and year of release, I basically decided last year to get super-anal and start tagging all songs, wherever possible, with original year and album of release, not the date or title of the CD I'm ripping. Two results of this policy:

(1) Ripping greatest-hits albums becomes a time-consuming process.
(2) My collection is now a mix of stuff I tagged properly and stuff I tagged lazily.

For example, a few months ago, I decided to rip a 2pac hits disc I'd owned for years. Did I let 'em rip under the 2Pac's Greatest Hits album tag? Hells, no: under the new Molanphy policy, "Keep Ya Head Up" gets tagged as Strictly 4 My N.I.G.G.A.Z. (1993), "How Do U Want It" is filed under All Eyez on Me (1996), etc. When I'm feeling extra-anal, I even re-tag the track numbers to conform with the track order and length of the original album.

I know, impressive, right? Except anything I've tagged prior to...oh, late 2004 or so, is just a mess. I am such a huge Prince fan, and yet, until recently, many of the dozens of tracks by him in my collection – which I ripped years ago – had the wrong year and were credited to The Hits/The B-Sides.

To be fair, Gracenote users have gotten a lot better lately about labeling the individual songs on a greatest-hits with their original year of release, rather than labeling all the songs with the year of the compilation's release. But that's not across the board. And frustratingly, Apple's iTunes Music Store almost always tags old songs with the year of the compilation album's release; so I find myself having to fix that shit. It's maddening to me: why should an Temptations chestnut be labeled "2004" just because I happened to get it off of the Motown #1s collection? Ridiculous...'cuz then, on the rare occasion when I tell the iPod to find me songs from the '60s, the Temptations don't come up becuase iTunes thinks that song is from the '00s. So stupid.

CMOM, I don't envy you with the tagging. I think I've decided to just ignore the year, it's too damn hard. But genres... how the hell do you keep that straight?

Oh, and Chris's Non-Stop Erotic Cabaret is, like, every single damn party we had junior year.

C-Mo:

So what do you do with boots like Prince collections that represent outtakes from studio sessions? What happens when song X was 'supposed' to be on album Y but ended up on album Z? What's the release year then?

I actually favor the semantic notion that the year of release is the year of the collection that I acquired. If "Chris Molanphy's Non-Stop Erotic Cabaret" [note the accurate use of the family name] has a song on it that was released the year before the collection was, my care level is low.

That said, of course I'm going to put original vinyl release date for cd releases or later remasters. But even there you run into trouble. When a jazz legend's seminal work is rereleased and includes tracks that were originally on another album, possibly a comp, does the release date go with the comp, the seminal work, or something else, since it's now part of a new package.

Then again, I'm more album-oriented than singles-oriented.

Tk, let me sort-of-answer your question(s) in an indirect way. Let's take the example of movie soundtracks.

If I rip the Pulp Fiction CD, I confront the question: For metadata purposes, is "Son of a Preacher Man" a song from the Pulp Fiction soundtrack in 1994? Or is it from the 1969 album Dusty in Memphis? For an anal record geek like me, the answer is obviously the latter, and so I correct the metadata – I look up Dusty in Memphis on AllMusic and add the cover art – and because I'm super-anal I even change the "Track # of #" tag to accurately reflect its placement on the original album as Track 3 of 11 (the 14 additional bonus tracks on the reissue don't count toward the total; yes, I'm that anal).

Great, so now I've satisfied my inner Rob Gordon (cf. High Fidelity), but "Son of a Preacher Man" is disintegrated from Pulp Fiction. And sometimes, I really just feel like listening to Pulp Fiction. This is where playlists save my ass: if the flow of an original soundtrack is important to me, I create a "Pulp Fiction" playlist and put the songs in there, in the correct order from the album. (Sometimes, I take advantage of my playlist freedom and stick in extra songs; my sister recently inspired me to create a playlist for "the complete Rushmore," adding in the Donovan, Rolling Stones and Vince Guaraldi tracks that didn't make the CD soundtrack, in the order they appeared in the movie.)

Taking it one step further, I also happen to own the Dusty in Memphis CD, and I decide I want to rip that to iTunes, too. Here's where my system pays off: I don't waste hard drive/iPod space ripping "Son of a Preacher Man" again; I rip everything else on the Dusty CD, and if I've done my "Track # of #" data right, "Preacher Man" will appear as Track 3 when I play the album on my iPod, right where it belongs. I don't create a playlist for Dusty in Memphis, because I can access that by artist or album title, and its songs will be in the correct order. But if I want to hear Pulp Fiction, I need the playlist.

In short, my governing principles are: 1. metadata should defer to original album release wherever possible; 2. avoid double-ripping of the same song at all costs; and 3. if a collection of various-artist or various-album tracks is important to you, make a playlist.

So in your examples, if the Prince outtakes are added to a reissue of an album, they get "credited" to that album, but I will change their track number to reflect their special status (the original album cuts are numbered "1 of 9," "2 of 9," etc. and the bonus tracks are "10 of 11" and "11 of 11"), and I might even append "[bonus]" or "[outtake 3]" to the song title. And as for song X appearing on album Z instead of the intended Y, it gets credited to Z unless it was released as a single so much earlier than album Z that to affiliate as such is silly; in that case, I might try to find cover art and track data for the original single. No kidding – I did this a lot when I recently created iTunes megamixes for John Lennon and Paul McCartney, who released some of their biggest solo hits in the '70s as singles apart from their albums.

And Mike, genre is indeed nuts. My only rule here is to try to limit it to a core of 10 to 12. For example, I killed off separate "Rap" and "Hip-Hop" tags (including Apple's preferred "Hip Hop/Rap" tag). For me, there is only "Hip-Hop." Same with "Alternative" and "Alternative & Punk"; I use the latter and dumped the former. And finally, "Soundtrack" is not a viable genre to me unless the song was only ever created/released as soundtrack music. So Mark Mothersbaugh or Danny Elfman incidental tracks count in the "Soundtrack" genre; "Son of a Preacher Man" is "R&B."

I do have fun with genres occasionally, though. I created a genre for chessy ballads called "Schlock."

Chris's Non-Stop Erotic Cabaret is very close to every party we had junior year! Does anyone still have that tape (was it one tape? or did we just put in the same songs in the same order every time?)? I definitely remember Groove is in the Heart, Freedom 90, Everybody Everybody, Everybody Dance Now, and that scary Nitzer Ebb song. I'd love to have the list if anyone remembers it.