Songwriter Network · the credit-and-split side of the craft · early access — nothing is running yet
The split was agreed in the room. Nothing wrote it down.
Songwriter Network is being built as one page per writer for the part of songwriting that has no home: who wrote what, in what share, agreed with the people who were actually there — plus the provenance of a lyric and a composition, and a catalog indexed the way a cue is searched for. It is the writer’s record, not the artist’s release page. It is not a publisher, an administrator, a rights society, or a licensing desk, and the section on what it refuses to do is the most important one here.
Read this before anything else: none of it is running. There is no live credit record, no co-writer confirmation flow, no catalog, and no cue index at this address today. This page describes a design and states its refusals. It claims no songwriter count, no song count, no roster, and no waitlist, and the one example it shows is a placeholder.
What actually goes wrong, in order
A split is usually settled out loud, late, between people who like each other. Nobody writes it down, because writing it down in that moment feels like an accusation. That is not carelessness — it is the correct social read of the room, and it is exactly why the record never gets made.
Then time passes, and the agreement has to be reconstructed from memory, a group chat, and the modified date on a session file. Reconstruction is where writers who still like each other start to disagree, because everyone remembers their own contribution most vividly. That is not dishonesty either. It is how memory works.
The four moments it surfaces — always at the worst time
- A sync request with a short fuse. Someone wants the song for a scene and needs to know, this week, whether one person can say yes. If the shares were never settled, the answer is not “no” — it is silence, which the buyer reads as no and moves on. The song did not lose on merit.
- A registration that has to be filed. Registration forms want shares that sum to one hundred. They will accept any numbers that do. Filing forces the argument, and it forces it after the fact, in writing, with money attached — the worst possible conditions for a conversation that would have taken ninety seconds in the room.
- A late clearance that re-proportions everyone. A sample or an interpolation surfaces and takes a share off the top. Every previously agreed number is now wrong, and the people who agreed to them were not told what they were actually agreeing to. If the original record never named the sample, nobody can even tell which agreement was made under which assumption.
- Someone dies, or stops speaking to someone. An estate, or an ex-collaborator, inherits a question no living person can settle. This is the case that turns a friendship-scale problem into a decade-scale one.
The specific arguments, named
These are not hypotheticals; they are the recurring shapes of the disagreement, and a record that does not have a field for them will not prevent them:
- Production is not a writing share, until someone decides it is. A person who built the track often expects a writer share, and a person who wrote the melody and words often does not expect to give one. Both positions are defensible. Neither is written down, so both are held silently and discovered in conflict.
- The title line. Someone said the phrase that became the hook, once, and did not write another word. Whether that is a share, a credit, or nothing is a genuine judgment call that has to be made by people, at the time — not inferred later by whoever is holding the form.
- Topline over a track that already existed. When the music arrived first and the topline came second, the split depends on terms that were probably never stated: whether the track was a work for hire, a spec beat, or a co-write.
- The person who was in the room and did not write. Presence is not authorship, and everyone knows it — right up until the credit list is being assembled and one name is missing from it.
The fix is not clever software. The fix is a record made at the time, by the people who were there, that every one of them confirmed. That artifact is what this address is for.
What a credit-and-split record will hold, and why each field earns its place
None of this is running yet. This is the shape the record is designed to take, field by field — and the reason each field exists, because a form with fields nobody can justify is how a record becomes something people fill in badly and never trust.
- The work, and every name it has been called
Title, plus working titles and the session file name. This is not bookkeeping fussiness: songs change names, and the number of disputes that begin with two people describing the same song and not realising it is not small. The alternate names are what make a record findable years later by the only name someone remembers.
- When and where it was written
Date and place. This is the field that makes the record contemporaneous, which is the entire source of its weight. A split written down the day it was agreed carries an authority that the same numbers typed in three years later simply do not have.
- Each writer, as a person and as a credit
Legal name, the name they are credited as, and their society affiliation and writer identifier if they have one. The credited name and the legal name are routinely different, and the gap between them is where a writer’s catalog quietly splits into two catalogs that never get joined back up.
- The share, as a number, summing to one hundred
Stated plainly, per writer. Not implied by the order of names, not “we will work it out”, not “equal” where the number of writers later turns out to be disputed. A record that does not force the numbers to sum has not settled anything.
- What each person actually contributed
Music, lyric, topline, arrangement — recorded as a class alongside the share. This is the field that prevents the most common argument, because it separates what you did from what you get, and lets people agree on the first before negotiating the second.
- Whether anyone controls a writer’s share
If a writer’s share is administered or published by someone else, the record says so. A co-writer discovering this at clearance time, rather than at writing time, is a preventable surprise — and it is the single fact most likely to determine whether a song can be cleared quickly.
- Samples and interpolations, declared at the time
Named when the song is written, not when a clearance arrives. If a share will later come off the top, everyone agreeing to a number deserves to know that before they agree to it. A split agreed in ignorance of a sample was never really agreed.
- Each writer’s own confirmation
Per person: confirmed, or not yet. This is the field that makes the difference between a record and an assertion. One writer typing numbers into a form is a claim about other people. A record where each named writer has confirmed their own share is an agreement.
Illustrative example · placeholder values only · not a real song, writer, or agreement
One split record, as it is designed to read
Every value below is a placeholder chosen to show the shape of a field. There is no such song, no such writer, and no such agreement. Nothing here has been cleared, registered, or filed anywhere, and the percentages are illustration — not a rate, not an earning, and not a sum of money.
| Writer (placeholder) | Contributed | Share | Confirmed |
|---|---|---|---|
| Writer A | Lyric, topline | 40 | Confirmed by Writer A |
| Writer B | Music, arrangement | 40 | Confirmed by Writer B |
| Writer C | Title line | 20 | Not yet confirmed |
| Total | 100 | Incomplete — would not publish |
Read the last row, because it is the design: the shares sum correctly and the record still would not publish, because one named writer has not confirmed their own share. A record that publishes on the say-so of whoever typed it fastest is not an agreement — it is a claim about absent people. The design fails closed instead. That is a deliberate refusal, described here in the future tense because the confirmation flow does not exist yet.
The writer’s page and the performer’s page are different pages
Search a song and you will find the person who sang it. That is correct — it is what most people are looking for. But it means a songwriter’s body of work is scattered across other people’s pages, filed under other people’s names, and there is no single address where it resolves into a catalog.
The two pages hold genuinely different facts about the same song, and neither is a substitute for the other:
- The performer’s page answers: who recorded it, on what release, in what year, and where can it be heard. Its unit is the recording.
- The writer’s page answers: who authored it, in what share, with whom, from what starting point, and who has to say yes. Its unit is the work. One work can have many recordings, and a writer’s catalog is invisible if you only ever index recordings.
Three kinds of work an artist page structurally cannot show
- Cuts by other artists. A song you wrote that someone else released sits under their name. Across a career, that can be most of the catalog — and it is exactly the part that demonstrates you write for other people.
- Songs on hold, and songs that never came out. Work that was written, pitched, and shelved is still work, and it is still evidence of what you can do. A release-driven page can only show what was released.
- Co-writes where you are not the credited artist. The room you were in does not appear anywhere unless the credits are published, and credits are published inconsistently.
Provenance is the other half of it: where a lyric or a composition came from, what it started as, whether it was written to an existing track, and whether it contains or interpolates something else. Provenance is the difference between a credit and a claim — and it is a fact about the work that no release page has a place to put.
Stated honestly: there is no such page running at this address today. No catalog is indexed, no cut is listed, and no provenance is recorded. This describes what the page is designed to hold.
A catalog indexed by cue, not by mood
Catalogs are usually tagged by genre and mood, because those are easy to enter. They fail the person searching, because a supervisor is not looking for a sad song — they are looking for something that works under two people not speaking in a car, ends cleanly at thirty-eight seconds, and can be cleared by Friday. Mood describes the song. A cue index describes whether the song can do the job.
One-stop, or not
The single most decisive field, and the one most often missing: can one party say yes to both the song and the recording, or does the request have to travel to several rights holders. A buyer with a deadline treats an unknown here as a no, because the cost of finding out is theirs to bear.
How fast a real answer comes back
Not a promise of a licence — a stated turnaround for a yes-or-no. A supervisor is not choosing between songs; they are choosing between songs that can answer in time and songs that cannot.
Where the vocal enters, and whether there is an instrumental
A scene needs a specific number of seconds before a voice arrives, and often needs the voice never to arrive at all. Whether a clean instrumental and a usable stem set exist is a practical yes-or-no that determines usability more often than mood does.
Tempo, key, and the shape of the arrangement
Where the song lifts, where it drops out, and whether it can be cut to length without collapsing. These are structural facts an editor works with, and they survive genre labels that different people apply differently.
What it contains that belongs to someone else
A declared sample or interpolation, stated up front. Discovering it late is what turns a placement into a withdrawn placement, and a buyer who finds out late remembers who did not tell them.
Notice what is not on that list: a price, a rate card, a fee, or any figure at all. This address does not quote money, does not clear rights, and does not run a licensing desk. A cue index here would exist so a writer’s catalog is legible to someone with a deadline, and so the conversation starts with the right person — not so a transaction happens on this page. No cue index is running today. No song is indexed and no catalog exists at this address.
What the field already offers, honestly — and where the gap is
Every category below is real and some of it is essential. None of it is named here, because naming a competitor to argue with it is a weaker move than describing what it does and where it stops. A working songwriter reading this already knows which product each row describes.
Collection societies and rights registries
What it does well: Register a work and collect against it. Genuinely important, genuinely irreplaceable, and this address does not compete with them.
Where it is thin: They begin after the split is settled. A registration form will accept any set of numbers that sums to one hundred and will never tell you the co-writers disagree — it has no way to know. They also disagree with each other across territories, and reconciling them is a job, not a lookup.
Publishing administrators
What it does well: Register, collect, and chase money across territories on your behalf. For a working writer this is often the correct thing to pay for.
Where it is thin: Their record starts when you sign, and they take a percentage — which is the honest trade, not a criticism. But it means the record of the write itself is a byproduct of a commercial relationship, and a writer without one has nowhere to put the same facts.
Split-sheet templates and forms
What it does well: Free, universally available, and better than nothing. Any writer can print one today.
Where it is thin: It becomes a file in an email thread. It has no confirmation state, so a sheet one person filled in looks identical to one everybody agreed to. It is not findable years later, it does not survive a changed phone, and nobody can tell which version was final.
Creator and collaboration networks
What it does well: Connect writers to each other and to sessions, which is real value and a different job from this one.
Where it is thin: They are built around the profile and the release, so the unit is the person or the recording — never the work and its authorship. The credit-and-split record is at best an attachment inside them, not the thing itself.
Sync and licensing catalogs
What it does well: Put music in front of buyers, which is the point, and some do it well.
Where it is thin: They index on mood and genre because that is what is easy to enter, and they bury or omit the field that actually decides a placement — whether one party can say yes and how fast. Many also require you to grant them something to be listed.
The gap, stated in one sentence
There is no writer-owned, contemporaneous, co-writer-confirmed record of what was agreed in the room — made before anyone registers anything, and without anyone taking a percentage to hold it. The reason the gap exists is not oversight: nobody in the field makes money at that moment. The registries make money at collection, the administrators make money at their percentage, the catalogs make money at placement. The write itself is commercially uninteresting to everyone except the people who were there, which is precisely why it is the thing that goes unrecorded.
The intended win is not a better version of any row above. It is the artifact none of them produce, positioned so it feeds all of them — a settled, confirmed record that makes a registration accurate, an administrator’s job easier, and a clearance answerable. This is a claim about design, not a measured result. Nothing here has been built or tested against a real catalog yet, and this page does not pretend otherwise.
Three objections a working songwriter will have — answered, not deflected
1. “I have been burned before. Something took a piece of my songs, or fed my lyrics to a model.”
That is the right instinct and it should not be argued away with reassurance. The answer is structural rather than a promise of good character, and it is checkable against this page:
- There is no rail here that could take a cut. This address collects no money, pays no money, and runs no checkout. A percentage cannot be quietly introduced by a product that has no payment path to introduce it into — and if one is ever built, its arrival is visible, because it would have to be built.
- Listing is not a grant. Recording that you wrote something does not transfer, licence, or encumber anything. The record is a description of an agreement you made with other people; this address is not a party to it.
- No lyric, melody, or composition is fed to an outside model for training. Stated in the present tense, as what this does now — not as an eternal absolute, because absolutes about the future are the kind of promise that gets quietly broken.
- Right now the honest answer is stronger than any of that: nothing is running, so there is nothing here that could have taken anything. The refusals below are the commitment being made about what gets built. Judge them when it exists, and hold this page to them.
2. “This is another empty network that wants me to do free data entry until it has enough people to be useful.”
It is empty. That is stated in the first paragraph of this page and in the footer, and there is no roster, no count, and no waitlist number here to imply otherwise. But the premise of the objection is that the artifact is only worth having once other people are on it, and for this particular artifact that is not true:
- A split record is worth having if you are the only person who ever sees it. Its value comes from being made at the time and confirmed by the people who were there — both of which are fully available on day one, with a network of exactly one writer and three co-writers who do not have accounts. Its usefulness does not depend on how many strangers joined.
- The value of a directory grows with size; the value of evidence does not. A record of a co-write from a room in 2027 is exactly as useful in 2034 whether or not anyone else ever signed up. That is the difference between this and a listings network, and it is why the empty-network objection lands differently here.
- What we will not do is fake the size. No count, no roster, no invented waitlist, no social proof from people who do not exist. If the honest answer to “how many writers are on this” is uncomfortable, the fix is to build something worth joining, not to print a number.
3. “My society and my publisher already have my splits. Why would I keep them twice?”
Because those are a different artifact, and this one refuses to become them. A registration is the filing; this is the agreement the filing assumes already happened.
- A registration cannot tell you the writers disagree. It accepts numbers that sum correctly. If two co-writers hold different beliefs about the same song, the registration is silent and stays silent until money moves — which is years later and much harder to unwind.
- A registration is filed by whoever files it. A record where each named writer confirmed their own share is evidence of agreement in a way that a form submitted by one party is not.
- This does not compete with either, and says so where it matters: where a registration is the thing that determines what happens in the world, the registration is what happens. This record does not override it, is not filed anywhere, and is not a legal instrument. Its job is to make the registration accurate before it is filed, and to be the contemporaneous account if anyone later asks what was agreed.
- Keep using them. If you have an administrator, this does not replace them, and any page telling a working writer to drop their administrator for an unbuilt product would be worth closing immediately.
What this refuses to do — and why each refusal is in your interest
A refusal is not a missing feature; it is a decision about what this is allowed to become. These are written down here so they can be held against the build later, and so that a future version quietly adding one of them is visible as a change rather than passing as growth.
It takes no share of any song, no publishing, and no administration percentage.
Not a reduced percentage — none. This is the one that makes every other refusal credible, because almost every incentive to misbehave downstream comes from owning a piece of the thing you are describing. It also means this cannot be the business model, and that is a fact you should factor in rather than a virtue to admire.
It registers nothing, with anyone, on your behalf.
It files no work with any collection society or rights body, and it never will silently. A product that quietly acquires the power to file on your behalf has acquired the power to file something wrong on your behalf. Filing stays with you or with the administrator you chose.
It collects no money and pays no money.
No royalty figure, no split payout, no rate, no earnings estimate, no checkout, no card. This page will not show you a number that looks like income, because the moment a page does that, every other claim on it starts being read as a sales pitch. Money is off here, and it is off deliberately.
It does not feed a lyric, melody, or composition to an outside AI model for training.
Stated in the present tense, as what this does now. It is deliberately not phrased as a permanent guarantee about all future conduct, and it makes no perimeter claim that your files can never travel anywhere — those are absolutes nobody can actually bind, and a writer who has been paying attention should distrust anyone who offers one.
It will not publish a split until every named co-writer has confirmed their own share.
This fails closed, and it will be inconvenient. If one writer will not confirm, the record stays incomplete rather than publishing the numbers the other three agreed on. That is the point: an incomplete record tells the truth about an unsettled song, and a published one would launder one person’s version into an apparent agreement.
It will not list a person who has not agreed to be listed.
No scraping, no roster assembled from credits found elsewhere, no page created for someone so they can come and claim it. Naming a co-writer inside your own record is describing an agreement; publishing a page about a person who never asked for one is something else, and the difference matters more in a field where a name is an asset.
It will not rank, score, count, or review anybody.
No leaderboard, no verified tier, no reputation number, no reviews of writers by other writers. Ranking authorship would turn a record of what happened into a competitive surface, and writers would then have to manage it — which is a tax on the exact people it claims to serve.
It will not hold your record hostage.
The record is designed to be exportable in a plain, readable form, because a record that only exists inside one product is a dependency rather than an asset. If this address goes away, your evidence should not.
One honest note about the whole list: these are commitments about what will not be built, and they are true in the present tense because none of it exists yet. They are not evidence of good behaviour under pressure, because there has not been any pressure. Treat them as the contract, and check them against whatever eventually ships.
When a co-writer is a minor
This is written for adult working songwriters, but pretending every co-writer is an adult would be false. A seventeen-year-old artist writes with a professional room. A student writer gets a cut. It happens, and a record designed as though it never does will handle it badly at the worst moment.
So the posture is conditional rather than absolute, and it is narrow enough to be checkable:
- A minor co-writer named inside a split record is not published as a public page and is not listed by default. Their share can be recorded and agreed — that is the protection, not the exposure — without a public profile being created for a child.
- A minor is listed publicly only where that minor’s own family has consented to it, and that consent is withdrawable.
- Being in a session, being in a credit list, or being on a release is never treated as advance agreement to a public listing.
None of this is running, because none of the record is running. It is stated now because the consent rule has to be designed in at the start — a system that publishes by default and adds a consent gate later has already published.
What is actually built, stated as a list rather than implied
Most sites put this in a footnote, or leave the future tense doing quiet work in a sentence nobody parses carefully. Here it is as a table, because a working songwriter deciding whether to pay attention to this deserves to know the answer in one glance. The answer is: almost nothing is built.
| Capability | Honest state today | Status |
|---|---|---|
| This address and this description of the design | Real. You are reading it. It is a statement of intent and a set of refusals, and it is the only thing on this page that exists. | Real today |
| A credit-and-split record you can create | Not built. There is no form, no stored record, and no account. Nothing can be entered at this address today. | Planned |
| Co-writer confirmation, failing closed until everyone confirms | Not built. This is the central mechanic described above and it does not exist. No confirmation has ever been sent or received here. | Planned |
| A writer’s catalog page and provenance history | Not built. No catalog is indexed, no cut is listed, and no provenance is recorded at this address. | Planned |
| A cue index a supervisor can search | Not built, and not even the second step. No song is indexed, there is no search, and no buyer-facing surface exists. | Planned |
| Export of your own record in a plain, readable form | Not built — there is nothing yet to export. Named here because it is a design commitment, not a shipped capability. | Planned |
| Any registration, clearance, licensing, or payment function | Not built, and refused by design rather than merely absent. See the refusals above. This is the one row that will still read the same when the rest are built. | Refused by design |
If that table makes this look early, it is because it is early. The alternative — describing a planned confirmation flow in the present tense and letting you find out later — is how a page becomes a lie with good typography.
How it is meant to work, when it exists
Written in the future tense throughout, because not one of these steps runs today.
- A writer starts a record for a work, ideally the same day. Title and working titles, date and place, and the writers who were in the room. The design target is that this takes less time than the conversation it prevents — if it does not, it will not get done, and a record that does not get done is worth nothing.
- Contributions and shares are stated per person. What each writer did, as a class, and the share as a number. The shares must sum to one hundred; the record will not pretend an incomplete set is a settled one.
- Anything that belongs to someone else is declared now. A sample or interpolation is named before anyone agrees, so that everyone is agreeing to the same thing.
- Each named writer confirms their own share. Not the person who created the record confirming on their behalf. Until every named writer has confirmed, the record stays incomplete and does not publish. This is the step the whole design exists for.
- The confirmed record becomes the writer’s own evidence. It sits on their page, exportable, and it is what a registration is prepared from, what a clearance question is answered against, and what an estate reads years later.
- Filing stays where it belongs. The record is not sent to any society or administrator by this address. That is your decision and your relationship, and this refuses to insert itself into it.
Joining is handled by conversation today, not by a self-serve signup, because there is no product to sign up to. That is stated plainly rather than dressed as an exclusive early cohort.
Common questions
Is anything on this page running today?
No. There is no credit record, no confirmation flow, no catalog, no cue index, and no account at this address. The only real things are the address itself, the design described here, and the refusals. The build-status table above states each capability individually rather than leaving it to a general disclaimer.
Does this take any part of my song, my publishing, or my income?
No. It takes no share of a song, no publishing, and no administration percentage, and it collects and pays no money at all. There is no checkout on this page and no payment path behind it.
Is this a replacement for a collection society or a publishing administrator?
No, and it refuses to become one. It registers nothing on your behalf and files nothing anywhere. Where a registration determines what happens in the world, the registration is what happens. This record is the contemporaneous account of what the writers agreed, which is what a registration assumes and never captures.
Is a split record here legally binding?
It is not a legal instrument and this page does not claim it is. It is evidence: a dated record of what the writers said they agreed, confirmed individually by each of them. Whether that is decisive in any given dispute is a question for a lawyer in your territory, and anyone telling you otherwise on a marketing page is overselling.
What happens if one co-writer will not confirm?
The record stays incomplete and does not publish. That is deliberate and it will sometimes be frustrating. An incomplete record accurately says the song is unsettled; a published one would turn one person's version into an apparent agreement, which is the exact harm this exists to prevent.
Are my lyrics used to train an AI model?
No. A lyric, melody, or composition is not fed to an outside model for training. That is stated in the present tense as what this does, deliberately not as an eternal absolute, because a promise about all future conduct is not one anybody can honestly make.
Can a minor be named in a split record?
Yes, and that is the protective case rather than the risky one: a young writer's share should be recorded and agreed like anyone else's. What does not happen is a public page being created for them. A minor is not listed publicly by default and appears publicly only where that minor's own family has consented, withdrawably.
How is this different from thesongwriter.directory?
thesongwriter.directory is a directory of songs organised around the question a choir or ensemble director has to answer -- the song, its key and range, and whether a group may perform it without a licensing fee. That is a permission question about performing a song. This address is one writer's own credit-and-split record: who wrote what, in what share, and the provenance behind it. Different unit, different question.
How is this different from thecomposer.directory?
thecomposer.directory is for people who write original scores, indexed on forces and instrumentation and whether a score can be looked at. This is for songs and the authorship behind them. A composer's score and a songwriter's split are different documents answering different questions.
How is this different from performer.network?
performer.network is about people who perform. This is about people who write, and the distinction is the whole reason this address exists: the performer's page and the writer's page for the same song hold different facts, and a writer's catalog disappears when only recordings get indexed.
Will I be able to get my own records out?
That is the design commitment: a plain, readable export, because a record that only lives inside one product is a dependency rather than an asset. It is not built yet, and it is listed as planned in the build-status table rather than described as available.
How many songwriters are using this?
None, and this page will not print a number to suggest otherwise. There is no roster, no member count, no waitlist figure, no borrowed quote, and no review anywhere on this page. When there is something real to report, it will be reported with the unit and the date attached.
What this page is and is not claiming
songwriter.network is an early-access address with nothing running behind it yet. There is no credit record, no co-writer confirmation flow, no catalog, no cue index, and no account here today, and the build-status table above says so capability by capability rather than hiding it in a disclaimer. This page cites no songwriter count, song count, catalog count, member roster, waitlist figure, borrowed quote, review, or rating, and names no writer, artist, song, label, publisher, or competing product. The single split record shown is prominently labelled illustrative: the writers, the contributions, the shares, and the confirmation states are placeholders chosen to show the shape of a field, and no such song or agreement exists.
Money is honest-off, which matters more here than on most pages because the subject itself is money: this shows no royalty figure, no payout, no rate, no advance, and no earnings estimate, runs no checkout, and charges no card. The percentages in the example are illustration of a field, never a sum of money. It takes no share of any song, no publishing, and no administration percentage. It registers nothing with any rights society and files nothing anywhere. It does not feed a lyric, melody, or composition to an outside AI model for training — stated in the present tense, never as an eternal absolute and never as a “your work never leaves” perimeter claim. A minor co-writer is not publicly listed by default and appears publicly only with that minor’s own family consent, withdrawably — a conditional posture, not a blanket claim that every co-writer here is an adult. This is a for-profit company’s product, not a charity, and nothing here is framed as a charitable gift or a tax-deductible contribution.