Why I don’t use iTunes metadata

I am taking quite a beating in the comments section of my previous post. Apparently I am a soon-to-be-crestfallen old man with OCD (if I combine Kelstar’s comment with the one from “Mr. D.”). Thankfully there are also messages from fellow Luddites who support my alternative lifechoice.

The object of the scorn I’m getting? Nothing to do with the Automator scripts I shared (since I am new to Automator I was hoping to get some feedback/correction/suggestions on that). It’s all about the use case that lead me to Automator in the first place, my refusal to rely on iTunes metadata to organize my music collection.

Look, I’m a software engineer. My employer (Oracle) knows a thing or two about structured data. I work in systems management, which is heavily model-driven. As an architect I really care about consistent modeling and not overloading data fields. I’m also a fan of Semantic Technologies. I know proper metadata is the right way to go. As a system designer, that is, I know it; any software I design is unlikely to rely on naming conventions in file names.

But as a user, I have other priorities.

As a user, my goal is not to ensure that the application can be maintained, supported and evolved. My goal is to protect the data. And I am very dubious of format-specific metadata (and even more of application-specific metadata) in that context, at least for data (like music and photos) that I plan to keep for the long term.

I realize that ID3 metadata is not iTunes specific, but calling it a “standard” that’s “not gonna change” as another commenter, Vega, does is pretty generous (I’m talking as someone who actually worked on standards in the last decade).

Standard or not, here are a few of the reasons why I don’t think format-specific metadata is a good way to organize my heirloom data and why I prefer to rely on directory names (I use the artist name for my music directories, and yyyymmdd-description for photos, as in 20050128-tahoe-ski-trip).

[Side note: as you can see, even though I trust the filesystem more than format-specific metadata I don’t even fully trust it and stubbornly avoid spaces in file and especially directory names.]

Some of the pitfalls of format-specific metadata:

  • Metadata standards may guarantee that the same fields will be present, but not that they will be interpreted in the same way. As proven by the fact that my MP3 files carry widely inconsistent metadata values depending on their provenance (e.g. “Beattles” vs. “The Beattles” vs “Beattles, The”).
  • I can read and edit file names from any programming language. Other forms of metadata may or may not be accessible.
  • I don’t have to download/open the file to read the file name. I know exactly which files I want to FTP just by browsing the remote directories.
  • I often have other types of files in the same directory. Especially in my photo directories, which usually contain JPEGs but may include some images in raw format or short videos (AVI, MPEG, MOV…). If I drop them in the 20050128-tahoe-ski-trip directory it describes all of them without having to use the right metadata format for each file type.
  • File formats die. To keep your data alive, you have to occasionally move from one to the other. Image formats change. Sound formats change. The filename doesn’t have to change (other than, conventionally, the extension). Yes, you still have to convert the actual content but keeping the key metadata in file or directory name makes it one less thing that can get lost in translation.
  • Applications have a tendency to muck with metadata without asking you. For example, some image manipulation applications may strip metadata before releasing an image to protect you from accidental disclosure. On the other hand, applications (usually) know better than to muck with directory names without asking.
  • You don’t know when you’re veering into application-specific metadata. I see many fields in iTunes which don’t exist in ID3v2 (and even less in ID3v1) and no indication, for the user, of which are part of ID3 (and therefore somewhat safer) and which are iTunes-specific. It’s very easy to get locked in application-specific metadata without realizing it.

[In addition to the issues with format-specific metadata listed above, application-specific metadata has many more, including the fact that the application may (will) disappear and that, usually, the metadata is not attached to the data files. That kind of metadata is a no-starter for long-term data.]

So what do I give up by not using proper metadata?

  • I give up richness. But for photo and videos I just care about the date and a short description which fit nicely in the directory name. For music I only care about the artist, which again fits in the directory name (albums are meaningless to me).
  • I give up the ability to have the organization of my data reflected in metadata-driven tools (if, like iTunes, they refuse to consider the filesystem structure as meaningful). Or, rather than giving it up, I would say it makes it harder. But either there is a way to automatically transfer the organization reflected in my directories to the right metadata (as I do in the previous post for iTunes) or there isn’t and then I definitely don’t want to have anything to do with software that operates on locked-away metadata.
  • I also give up advanced features that use the more exotic metadata fields. But I am not stripping any metadata away. If it happens to all be there in my files, set correctly and consistently, then I can use the feature. If it isn’t (and it usually isn’t) then that piece of metadata hasn’t reached the level of ecosystem maturity that makes it useful to me. I have no interest in manually fixing it and I just ignore it.

I’m not an audiophile. I’m not a photographer. I have simple needs. You may have more advanced use cases which justify the risk of relying on format-specific metadata. To me, the bargain is not worth it.

I’m not saying I’m right. I’m not saying I’m not a grumpy old man. I’m just saying I have my reasons to be a grumpy old man who clutches his filesystem. And we’ll see who, of Mr. D. and me, is crestfallen first.


Filed under Apple, Everything, Modeling, Off-topic, Standards

7 Responses to Why I don’t use iTunes metadata

  1. sporobolus

    I use format-specific metadata carefully, mostly because after years of using directory structure I realized that hierarchy was really limiting me; it has saved me a lot of time, too, because there are many good tools and most scripting languages do have libraries (presence of such tools/libraries is a good test of whether the metadata format is stable)

    i _am_ a photographer & i can’t imagine trying to find one of my photos without multiple non-hierarchical categories; today’s image tools are tremendous aids, but not without metadata…

    also worth noting that iTunes maintains an XML version of the metadata stored in the id3 tags, which i see as an automatic non-format-specific backup; more than once i’ve hacked that XML to save a lot of time & hassle, and if iTunes along with every id3 tool disappeared tomorrow I’d still have everything

  2. “You don’t know when you’re veering into application-specific metadata”
    That is a very good point, and something I’ve not thought of before (since being anal about my iTunes library, and it’s metadata, since around 2001).
    Though I think it’s more of a theoretical problem in your case. Your method only allows for Artist and Album data (the folder names), and each of these are in ID3v1 and ID3v2, so I’d argue it’s a moot point.

    “File formats die”
    Very true. But, like sporobolus said in the previous comment, iTunes (being the subject of your ire) does give you a file that most people consider fairly open. An XML file. I don’t see text files going away any time in the next 50 years.

    “Applications have a tendency to muck with metadata without asking you”
    This is a bit of a stretch. Sure, I’m sure some applications do (you mention an example ” SOME image manipulation applications MAY strip”), but to say they have a “tendency to muck with…” is taking that concept too far, I think.

    Having said all that, your simple needs are perfectly met by your method.
    I also understand the Luddite stubbornness. I didn’t even consider the idea of buying a ebook reader until about 4 years ago. “I like the physicality of a REAL book!” Haha!

    Interesting discussion. Thank you for sharing.

  3. Thomas Beagle

    “As proven by the fact that my MP3 files carry widely inconsistent metadata values depending on their provenance (e.g. “Beattles” vs. “The Beattles” vs “Beattles, The”).” – unlike filenames…. oh, wait.

    In that case I think you’re confusing the way that the data is stored with the quality of the data.

    (I am sympathetic – I use tools to get my MP3 ID tags all correct – and then another tool renames all the files accordingly. Best of both worlds for me.)

    • Alex

      You can use MediaMonkey to do both. It updates metadata using information from Amazon and then it allows you to rename files any way you want based on that metadata. Conversely, you can also fill in the metadata based on the file names.

  4. Katie M.

    Architectual, structural, prickly squaredom. How nice that these things pay the bills for you. But good grief already! Try letting go and chilling out while spending more time listening to your music than trying to rule it with an iron fist.

  5. @vambenepe

    Relevant data point for this discussion: John Gruber suggests on Daring Fireball that the definition of “a post-PC device” could be “one with no user-visible file system” (in the context of comparing DropBox with iCloud).

    Except that in this post, I’m talking about iTunes on my Mac, which very much exposes a filesystem (thought maybe that’s why the MacOS finder is so bad, to progressively have us give up on looking at the filesystem…)

  6. missjanesvoice

    Hi Vambenepe,
    I wholeheartedly agree with your point relating to Metadata. We run a home network in which we try and share 4 Audible accounts. Audible and Apple signed an agreement way back which excludes almost all other media players from playing and/or seeing the files, most of which have far superior network capabilities, by network i exclude peer peer. Another aspect is that the metadata for these files can not be edited, at all. You can update the metadata within iTunes of course but lose or ‘update’ iTunes and your back to where you started.
    As a home user I’d love to be able to organize our, now over a thousand, book collection but due to this proprietary file type it’s just futile. So we actually have to rely (file names don’t do it for audiobooks sadly)on Audible getting it right, sadly I’m not surprised at how many spellings of ‘Anne McCaffrey’ there can be. So thanks again for voicing a raspberry at iTunes.
    Oh and a note to those who wish to offer an alternative to iTunes note all 3rd party media players who can work with Audible files use iTunes as a back end. And it still doesn’t solve the proprietary metadata issue.