T-000.000.001-0

Fun Fact. In 1995 ABBA’s Dancing Queen was assigned the very first ever ISWC- “International Standard Musical Work Code- heralding the Arrival (yes- an ABBA pun) of identifiers intended to uniquely identify compositions, globally.

Photo by Patrick Federi

Dancing Queen is literally, T-000.000.001-0.

It is a fitting choice. Written by Benny Andersson, Björn Ulvaeus and Stig Anderson and released in 1976, it is one of the most recorded and most globally recognised compositions in existence. If any song in the world should be unambiguously identified across the world's rights databases, this was a fitting choice to start with.

So let's do something anyone reading this can replicate, and take a look at the databases- The MLC's public search, PRS's works search, ASCAP's and BMI's repertoire tools- and look up Dancing Queen to see how the ISWC has fared in its 30+ years of history.

What an ISWC is Supposed to Be

Quick primer- an ISWC is meant to uniquely identify a musical work- not a recording, not a release, but the underlying composition itself. Under CISAC's own rules an ISWC represents a unique combination of title, writers (who must be identified by their IPIs), and their roles. You cannot register a song with the same set of writers, in the same roles and with the same title. Similarly nor should one ISWC represent two different writer combinations.

That rule is what makes the identifier useful. As I've written before, identifiers are a form of currency- their value depends entirely on confidence. An identifier people have learned to second-guess is not worth the data field it occupies. It is effectively useless.

The Count

Searching CMO databases today for Dancing Queen reveals:

• 14 variants at PRS

• 17 variants at The MLC

• 1 registration at BMI, 0 at ASCAP

  
  

The same composition exists in fourteen and seventeen different forms respectively. Some of these are legitimate- translations, adaptations and arrangements are all real works with real (and distinct combinations of writers/roles) attached. Critically, the existence of legitimate derivatives is precisely what makes the illegitimate ones so hard to spot: every duplicate can plausibly claim to be a version until someone checks.

1 at BMI and none at ASCAP is to be expected- but the variants at PRS and MLC are mostly translations... yet where are they at BMI?

The Flagship Registration

The good news first: T-000.000.001-0 currently carries the correct writers at both PRS and The MLC.

The less good news is that at The MLC this has only recently become true. For at least the last two years, the world's first ISWC was registered with an additional translator credited on it- the writer of an Italian-language version, attached to a registration that had well over a hundred recordings hanging off it, most of which were plainly not the Italian version.

Mechanically (another joke) this means revenue intended for the original Dancing Queen was being watered down by the presence of another party siphoning off a %. Likely an honest mistake in this case- but often an intentional and fraudulent “theft” in others.

Encouragingly it has since been fixed, which is worth acknowledging. I only know this as it had become a favourite (and demonstrable) war story of publishing- that even the very first issued ISWC was still incorrect in some databases.

But indeed, it was wrong for 2+ years, on the most identifiable song in the dataset, in the one rights database that is public by legislation and therefore under more scrutiny than any other. The question isn't why this one was wrong. It's what the equivalent error rate looks like on works nobody is famous enough to check.

The Orphan Recordings

Here is the finding that best illustrates why this work is hard.

At The MLC, the majority of recordings are not matched to T-000.000.001-0 at all. They're matched to a different ISWC- but one which also carries the correct three writers. Take a look for yourself:

• T-001.610.075-6. 12,000+ matched recordings at time of writing

• T-000.000.001-0. 5,000+ matched recordings at time of writing

  
  

So the public record contains two registrations for the same composition, both with the correct writer combination, one holding the correct, Preferred, historic identifier and the other holding more recordings. Under the CISAC rules, these should be a single work.

And then the obvious question, which I don't think has a satisfying answer: if both registrations carry the same three writers and the same title, why have they not been merged? Nothing about that determination requires privileged information or difficult judgement. It requires someone (or technology) to be looking, and to have the authority and the mechanism to act.

The history is the tell- one registration has been updated, until the point where it effectively matches the other registration- yet they still haven’t been merged. 

This speaks to the structural heart of the problem. Registration is a distributed activity- publishers, sub-publishers, and administrators all contribute data to societies- yet surely societies should hold responsibility for deduplication?

Things Don't Stick Around

I looked at this same song in the public data back in 2023. The registration holding the bulk of the recordings then (T-001.632.322-0) is not the registration holding them now. Different ISWC, same song, same three writers, three years apart.

So the picture isn't simply "wrong and static". Recordings drift between work registrations as matching is reprocessed, admiistrators are updated, duplicates are partially merged and preferred ISWCs assigned. The centre of gravity for a song's recordings- and therefore its money- can move without anyone downstream being told.

That has an uncomfortable implication for how this industry works. Catalog reconciliation is typically treated as a project: commission it, fix the findings, file the report, move on. But if the underlying data floats, a clean bill of health is a stationary snapshot rather than an ongoing guarantee. What was correctly matched at the point of diligence may not be two years into ownership. Data quality isn't a discrete project with an end date- it needs to be an ongoing, proactive, monitoring task.

The Shares, and the Money

Two more findings, both from the same public source.

First: four of the MLC registrations are missing 25% of shares.

A quarter of the work, unclaimed. That's the "unclaimed" category from last week's piece in its purest form- not unmatched, not unknown, not disputed. Simply a share of a very famous song that nobody has connected to an owner.

Second, and inevitably the question everyone actually wants answered- what is it worth?

Running the unmatched recordings for these registrations against the MLC's published value data gives a floor of at least $5,000 in unmatched value across all 17 variants. The MLC reports unmatched value in bands rather than exact amounts, so the only intellectually honest figure you can derive is a minimum- the real number is likely higher.

Five thousand dollars is not a headline, and it isn't the point. The point is that this is one song- the single most identifiable composition in the database, with no ambiguity about who wrote it- and it still has thousands of dollars of recordings floating unmatched, plus four registrations quietly missing a quarter of their shares. It’s chastening to think about multiplying that pattern across a catalog where nobody would recognise the titles- and it starts to add-up towards the higher end figures that accompany news stories around CMO black boxes.

Also don’t forget- from 2027, unclaimed money like this begins to be redistributed pro-rata, month by month, to everyone else.

What This Actually Tells Us

It would be easy to read this as a story about one song, or one society, and it isn't. Every organisation involved is competent and well-resourced, and several of these issues have been actively corrected. The MLC in particular deserves credit for the fact that this analysis is possible at all- no other rights database in the world could be audited to this depth by effectively a member of the public.

The lesson is structural, and it's uncomfortable: If the most famous, most documented, first-ever-numbered composition in the world looks like this in the public record, the base rate of error across everything else is not small.

Consider what protects Dancing Queen: enormous revenue, sophisticated publishers, well-known writers, decades of scrutiny, and a literally unforgettable identifier. Now consider a mid-tier work from 1994 with four co-writers, two translations, a sample, an administrator change and no particular reason for anyone to look closely. That's most catalogs, and that's the concern.

Identifiers only work when the data behind them is disciplined. Muddled registrations isn't a failure of the ISWC as a concept- it's a demonstration of what happens when a good standard meets an industry with no efficient and shared mechanism for resolving conflicting truth. And because that truth drifts, checking once is not the same as knowing.

We gave the world's first work identifier to a song we all know by heart, and half a century later the databases still can't agree on what it is.

Go and look up a song you own. Not your biggest earner- one from the middle of your catalog (and particularly one that has generated derivatives). How many versions of it exist out there, and do you know which one your money is arriving through, if at all?