Two halves, and most artists only own one
A song is two legal objects, not one.
The work is the composition — the melody and the words, the thing that exists before anybody presses record. It gets an ISWC, an International Standard Musical Work Code, and in the United States it gets registered with The Mechanical Licensing Collective, which is the body that pays songwriters for streaming mechanicals.
The recording is the master — a specific performance of that composition, captured on a specific day. It gets an ISRC, an International Standard Recording Code, and it lives at the DSPs.
Different objects. Different identifiers. Different money. One song can have one ISWC and forty ISRCs, because forty different recordings of it exist.
I am going to publish exactly where I stand on both halves, including the half where I am failing, because the useful version of this article is the one you can check rather than the one you have to believe.
Where I actually stand, and how to check it in thirty seconds
47 musical works are registered with The Mechanical Licensing Collective under the writer name DANIEL DAJAI STEWART-HANDY (IPI 00773316238), publisher Daniel Stewart-Handy, MLC publisher number P741NI. Every one of the 47 shows Total Known Shares: 100%.
You do not have to take my word for any of that, and you should not.
- Go to portal.themlc.com/search. There is no login. The public work search is anonymous.
- Set the search criteria to Writer Name.
- Query DAJAI.
That is the whole recipe. Verified 2026-07-30.
One practical note that makes the recipe actually usable: search the writer name DAJAI, not the surname. A STEWART DANIEL search on that same date returned 492 writer rows, effectively none of which are mine. The distinctive token is the middle name, which is why the registry string is worth publishing verbatim rather than paraphrased.
The reason I am leading with this instead of a play count is that it is the only substantial claim on this entire domain that a stranger can confirm without asking me for anything. Everything else I publish — catalog size, cumulative plays, mastering grades — traces back to my own files. A single-source entity is a fragile entity. This one has a third-party record behind it.
What "100% Total Known Shares" means, and what it does not
Total Known Shares is the percentage of a work's ownership that has been claimed by somebody in the MLC's database. At 100%, there is no unmatched share sitting in the black box waiting for a claimant.
That is a real and specific thing, so let me be equally specific about what it is not:
- It is not a revenue figure. It says the shares are claimed, not that they earned.
- It is not a quality signal. The registry does not listen to the record.
- It is not a chart position, a certification, or an award, and anybody presenting it as one is stretching.
What it does mean is that when a stream generates a mechanical royalty in the US, the ownership question for these 47 works has an answer on file instead of a gap. Unclaimed shares are the default failure mode for independent writers, not the exception. Closing that gap is administrative work, and it is the least glamorous and most load-bearing thing on this list.
The gap: 19 of 47 have no ISWC — and it lands on the collaborations
Here is the part that does not flatter me.
19 of the 47 registered works carry no ISWC. The shares are claimed; the international identifier is missing. Domestically that is survivable. Across territories it is not, because the ISWC is the key societies use to recognize that the thing registered in one country and the thing registered in another are the same composition.
The honest headline is not "19 works are missing an identifier." It is this: the works most likely to fail a cross-society match are my collaborations — which is precisely where more than one society is involved and the matching actually has to work.
Three of the 19 resolve to multi-writer records in the catalog:
| Registered work | Recording it resolves to | Plays | |---|---|---:| | GIVE ME CHIP REMIX | Big Sad 1900 x Daj' x Tizzy — Give Me Chip (Prod. NoLimitAustin) | 341,093 | | BADDIES | YT & Daj' — Baddies (Missed Calls) | 27,066 | | TAKE IT TO TRIAL | Daj' & Tizzy — Take It To Trial (Freestyle) | 5 |
Play counts are the summed stream counts for that recording in the catalog export, read 2026-07-30. The top row is the highest-play MLC-registered work in the catalog, and it is a three-way credit with no ISWC on it.
The full verified no-ISWC list, as the registry spells it:
| # | Work (registry casing) | ISWC | |---:|---|---| | 1 | GO | missing | | 2 | THATS MY | missing | | 3 | NASTY GIRL/ON CAMERA | missing | | 4 | GATTI | missing | | 5 | BEATBOX | missing | | 6 | WE SHOULD | missing | | 7 | TAKE IT TO TRIAL | missing | | 8 | SUMMER LEAGUE | missing | | 9 | RARA | missing | | 10 | BIG SHIT | missing | | 11 | THROAT BABY | missing | | 12 | BLACC MERCEDES | missing | | 13 | THUNUNUH | missing | | 14 | TRIPLE S | missing | | 15 | BADDIES | missing | | 16 | H TO THE IZZO | missing | | 17 | STREET SWEEPER | missing | | 18 | ATHLETICS | missing | | 19 | GIVE ME CHIP REMIX | missing |
A disclosure that belongs at the top of that table: 47 works are registered. 25 titles have been transcribed off the registry so far — the 19 above plus six more whose ISWC state was not recorded. The remaining 22 are registered but not yet transcribed here. I am not going to pad the list to 47 with plausible-looking titles. A fabricated work title under a real writer's name is a lie about a public registry, and it is the kind of lie a fact-checker can catch in one search. The complete set is retrievable by anyone with the recipe above.
The six additional observed titles, with ISWC state recorded as unknown rather than guessed: MACHO, VIBELIKETHIS, GIVE ME CHIP, CHOOSE UP, HEAVY, STILL BACKIE.
The other half is empty
Now the recordings.
The internal catalog export holds 15,405 rows. The isrc field is null on every single one of them. Not most. All 15,405.
I checked the other copy too, in case the export was the stale one. The catalog file this site actually serves holds 15,650 rows, and the ISRC field is null on all 15,650 as well. Two files, two row counts, the same zero.
(Three denominators are in play across this domain — ~14,030 deduplicated SoundCloud tracks, 15,405 export rows, 15,650 served rows — and they do not agree. That is a real open problem rather than a rhetorical hedge; it is taken apart properly in the companion study, 15,650 Uploads. Ten Tracks Are 82% of Everything.. What matters for this article is that the ISRC count is zero under every one of the three.)
Alongside it, the DSP linkage from that same export, read 2026-07-30:
| Field | Rows populated |
|---|---:|
| SoundCloud (@dajai-io) | 14,291 |
| YouTube | 631 |
| SoundCloud (@supercooldaj) | 202 |
| Apple Music | 101 |
| ElevenMusic | 58 |
| Spotify | 10 |
| Genius | 10 |
| Flagged dsp_released | 158 |
| ISRC | 0 |
So the position is lopsided in a way I did not fully see until I put both halves on the same page: the compositions are registered and fully claimed, and the recordings are essentially unidentified. 47 works have a home in a national registry. Zero masters carry the code that would let a platform, a distributor, or a rights database recognize them as specific recordings.
That is not a story about volume. It is a story about which half of the paperwork an independent artist tends to do first — and I did the songwriter half.
The root cause of the forks is a typography bug, and it is in my own files
There is a second-order problem that comes out of the same drawer.
My MusicBrainz artist record shows 22 release groups and zero registered aliases. Zero. And the artist name has an apostrophe in it: Daj'.
An apostrophe is not one character. There is the straight typewriter apostrophe ' (U+0027) and the curly typographic apostrophe ’ (U+2019). To a human they are the same mark. To a string matcher they are two different artists.
With no alias registered, Daj’ does not resolve to Daj'. So when a delivery carries the curly form, ingestion does not find the existing artist — it mints a new one. The observable consequences:
| Symptom | IDs |
|---|---|
| Canonical Apple artist | 141720605 |
| Apple forks holding real releases | 1895128322, 1895066240 |
| Canonical Deezer artist | 13032359 |
| Deezer fork | 58345252 |
| Canonical Spotify artist | 5cKcbJFuArX3O0nBFmxJ6D |
| Spotify profile holding DARK I as an orphan | 4v8xRRlsd0DnbZXo0KqMAf |
| Flight to Vegas on Spotify | not present at all |
DARK I — Outwitting the Devil, released 2026-04-07, ten tracks through UnitedMasters — sits on the second Spotify profile rather than the canonical one. The merge is platform-gated; I cannot fix it by editing my own site.
And here is the part that stopped me, because it is the same bug looking back at me from my own repository.
The generated album index on this site, built 2026-05-24, files "Daj’s Dedication" with the curly U+2019. Same file, same generator: DARK I appears twice under one identical slug — once as "DARK I: Outwitting The Devil" with 5 tracks and 4,354 cumulative plays, and once as "DARK I Outwitting the Devil" with 10 tracks and 0 plays. One album, two rows, colliding on the same slug, with the play history stranded on the wrong one.
The bug that split my Apple artist page is in my own JSON too. I did not import it from a DSP. I generated it.
There is also a third Spotify profile that surfaced during this audit which I have not claimed anywhere in this site's structured data, because I have not confirmed it is mine. An unverified identity assertion is a false one until proven otherwise, and the safe default is silence.
Open items, stated as open
Not a roadmap. A list of things that are currently wrong.
- Obtain ISWCs for the 19. Collaborations first, because that is where cross-society matching breaks.
- Transcribe the remaining 22 titles off the registry so the published enumeration matches the registered count.
- Register a MusicBrainz alias for the U+2019 spelling. This is the upstream fix — every fork below it is a symptom.
- Pursue the platform-side merges at Apple, Deezer and Spotify. Gated on the platforms, not on me.
- Fix the duplicate DARK I row and the curly apostrophe in the generated album data. That one is entirely mine and has no external dependency.
- Get ISRCs onto the masters. 15,405 rows, zero codes, and no honest way to spin that.
- Reconcile the track-count denominators so the site publishes one number instead of two defensible ones.
The registry entry is the strong half and it is checkable in thirty seconds. The recordings half is weak and I would rather say so here than have someone discover it and conclude the strong half was decorated too.
The per-work detail, the composition schema, and the full open-metadata list live on the publishing page. The album itself is at DARK I — Outwitting the Devil.
Two companion pieces published the same day take the same approach — publish the measurement, including the part that fails. The catalog power-law study does it to the play distribution. The mastering engine write-up does it to my own masters, where a QC engine I built caught a defect in a batch I had already shipped.
FAQ
What is the difference between an ISWC and an ISRC?
An ISWC identifies the composition — the song as written, independent of any particular performance. An ISRC identifies a specific recording of a composition. One song can carry a single ISWC and dozens of ISRCs, one per released recording. They are issued by different bodies, registered in different places, and pay different people: the ISWC side pays writers and publishers, the ISRC side pays performers and master owners.
What does "100% Total Known Shares" mean on an MLC work?
It means the entire ownership of that work has been claimed by identified parties in the MLC's database, with no unmatched percentage left sitting unclaimed. It is a completeness measure for the ownership record, not a revenue figure, a chart metric, or any kind of quality signal. Unclaimed shares are the normal failure mode for independent songwriters, so closing them to zero is administrative work rather than an achievement.
How do I verify these 47 registrations myself?
Open portal.themlc.com/search, set the search criteria to Writer Name, and query DAJAI. No account and no login are required — the MLC public work search is anonymous. Search the distinctive middle name rather than the surname: a STEWART DANIEL query returned 492 unrelated writer rows on 2026-07-30, while DAJAI isolates the right writer.
Why would one artist end up with several artist pages on the same streaming service?
Usually a string-matching failure rather than anything deliberate. If the artist name contains a character with more than one common encoding — a curly versus straight apostrophe is the classic case — and no alias is registered upstream to tie the spellings together, a delivery using the alternate spelling will not match the existing artist and the platform creates a new one. The fix is to register the alias at the authority record so both spellings resolve to a single artist, then request the platform-side merges for the duplicates that already exist.