ISWC, ISRC, IPI Name, IPI Base, ISNI, UPC… this article is about why we need unique identifiers and what is wrong with them.

What’s wrong with …

ISWC?

Or let’s start with the “wrong” part. ISWC is a unique ID for musical works. Well, kind of. To be a unique ID, one should not be assigned to two musical works, but also one work should not have two IDs. Which is the lesser of two wrongs. Multiple ISWCs get often assigned to one musical work.

The bigger problem is, as described in the second article on music metadata, that ISWC issuing is slow. Read about it there.

ISRC?

Unlike ISWC, ISRC, a unique ID for recordings, is self-assigned. Labels get their own ID prefix, and then they can assign ISRCs themselves. So, it is fast, but there is no central ledger. Well, the smaller issue is that labels mess things up and assign one ISRC to multiple recordings. (Multiple ISRCs to one recording is not really a problem without the central ledger.)

But without a central ledger, it’s kind of useless as an identifier.

IPI Name Number?

Usually just called IPI #, this is actually a IPI Name Number, which uniquely identifies a name, either writer’s or publisher’s. Not the writer or the publisher. The name. So, a “John James Doe” may have several, as “DOE JOHN”, “JOHN DOE”, “JOHN J. DOE”, “JOHN JAMES DOE”… you get the picture. And then this misinformation that it has 9 digits…

I guess it is really OK as a name identifier, but it makes a mess, since one person can have several ways the name is written and therefore several IPI Name numbers. Plus the pseudonyms.

And, as this was linked to affiliations with societies, different societies issued different IPI Name numbers to publishers affiliated with more than one.

IPI Base Number?

Well, IPI Base number is the other side of the medal. One person (real or legal) should have only one. It is actually a good thing and there is nothing really wrong with it, not counting the issue with it and GDPR. Except that almost no-one has ever heard about it. But it is actually the only one so far that should be called “unique identifier”.

Can we start over?

Ideally, one identifier must reference only one object (be unambiguous) and one object must have only one identifier for one purpose (be unique). (A publisher can have both IPI Base # and a VAT #, they have different purposes, but a publisher can only have one of each.)

With an identifier, one should be able to get to this object and other objects that have a relationship with it (e.g. the composer of a musical work), including their unique identifiers. This actually kind of works for musical works + ISWCs and songwriters (composers and lyricists) + IPI Name Numbers. I say “kind of”, as neither of these identifiers is unique, though both are unambiguous.

The panacean database linking ISRCs to ISWCs

I have been hearing about plans to build one such database since I started working in this industry. A database that links recordings and musical works, by linking the unambiguous, bit not unique ISWCs to neither unambiguous nor unique nor centrally ledgered ISRCs.

And to make things even more complicated, some propose that the database should be decentralized and built upon the blockchain technologies. And run by dragons and unicorns, for good measure, I guess.

I am not saying that it is not possible (well, unicorns and blockchain is not possible, but that should be obvious), it is, but it would be really really prudent to start using good identifiers. For those who really want to dig in:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5490878/

ISWCs are probably fixable, they (CISAC, or rather their new tech partner) need to figure out how to assign them really really quickly (in milliseconds) without making duplicates.

ISRCs should really get a central ledger. Solution for managing them should really be included in the price of suffixes, but any central ledger build on current technologies should do.

ISNI as a good example

ISNI is actually doing a very good job. While maybe not ideal (I will not go there this time), ISNI is actually a very good example of a modern identifier.

And UPC?

Dead. As a music-related identifier at least. EAN too.