Comment by pflenker
2 days ago
Somewhat related: some conscious artistic choices - such as writing down two tracks but delivering them as one (not sure if this is what happened here) can’t really be transferred into databases.
I own a cd where one track name is a small icon depicting a heart stabbed with a rather lengthy knife. To my knowledge, this track has no canonical name. Any digital version of this cd betrays the respective author‘s interpretation of the icon.
And then, of course, there’s „Love Symbol“: https://en.wikipedia.org/wiki/Prince_(musician)
> can’t really be transferred into databases
Of course they can, it's up to the person designing the database schema to anticipate what is a common artistic practice and model the data accordingly. It might be that specific databases like MusicBrainz and Gracenote haven't accounted for that, but if you own the schema you can easily set up a one-to-many (parent/child) relationship between physical track and song name.
One extreme example of this would be the "Lovesexy" album by (the artist formerly known as) Prince, which in its original CD form had only one track, containing 9 songs. I think the Spotify version is still faithful to this.
This and many other common "conscious artistic choices" ought to be collected into a "Falsehoods Programmers Believe About Recorded Music", if that is not already a thing...
In your example above, yes it's true that many song titles and artist names are fully and partially graphic symbols with no direct text representation (another thing Prince was fond of), but again given the prevalence of this there's no reason a smart data schema couldn't model a song or artist having a 'canonical' name that can only be represented by some graphic format along with one or more pronounceable/text-encodable alternatives (TAFKAP/Love Symbol) and so on; and of course tracking the fact that the 'preferred' identifier can change over time (Cat Stevens/Yusuf Islam, to mention a non-Prince example).
The person designing the database schema might have made a conscious decision not to support arbitrary track names and layout, since the added restrictions will make the client side software easier to implement.
But I guess it is also possible to design a schema that includes a lot of optional metadata, and clients only need to support up to certain core set of features if they choose.
It’s entirely possible to come up with artistic choices which precisely aim to be impossible to capture. So at times it’s not due to falsehoods someone believes, but rather the opposite - artists deliberately breaking the limits of what’s currently possible. Not a direct CD example, but one vinyl (was it by Pink Floyd?) has a special last track, where the needle is redirected endlessly, making the last track effectively endless. Or double grooved vinyls, Track 0s on CDs and so on.
But to stay with my example of the stabbed heart - even if the DB supported it, you’d still have to make choices when converting the icon as printed into the database, such as coloring.
How about "Naming the CDs"
There's a handful of albums that MusicBrainz doesn't quite have the right cd naming for since one was labeled "LEFT" and the other "RIGHT" and not 1/2 -- there is no canonical 1/2 order.
What's the CD?
The Inchtabokatables - Ultra