What Music's Royalty Infrastructure Is For

Picture a track that gets 900 streams a year. It’s not a hit, but people play it. Until April 2024 its recording earned a little money on Spotify. Now it earns nothing, because Spotify stopped paying recording royalties on tracks with fewer than 1,000 streams in the previous 12 months and gives that money to the tracks above the line. (The songwriter’s smaller share still gets paid.)

Spotify’s reason is worth reading closely. Labels and distributors usually make artists build up $2 to $50 before they can withdraw, and banks charge $1 to $20 to send the money, so for tracks earning a few cents a month, “this money often doesn’t reach the uploaders.” Spotify put the stuck money at about $40 million a year. Its fix wasn’t to make paying people cheaper. It stopped paying them.

Given what Spotify has to work with, that’s a reasonable call. But look at what that implies: paying the artist behind that track costs more than the track earns. This post is about why that’s still true, when almost everything else about moving information and money has become cheap.

Everything Else Got Cheaper

In 1844 a letter sent more than 400 miles cost a quarter, around $8.64 in today’s money. A WhatsApp message costs nothing beyond your data plan. Sending $200 home to another country cost almost 10 percent in fees when the World Bank started measuring in 2008, and about 6 percent in 2025, and the World Bank credits part of that drop to new digital providers.

Better technology made those things cheaper to do. Competition, we think, is most of what made them cheaper to buy. Where competition is missing, a price can sit still for decades. For more than 40 years every Japanese bank paid every other bank the same fee, ¥117 (about $0.74) or ¥162 (about $1.03), for each transfer between them, while one economist put the receiving bank’s actual cost at about ¥44 (about $0.28). The fee came down, to ¥62 (about $0.39) in 2021, only after the Japan Fair Trade Commission published a report on it and the government got involved.

Music has both halves of that story. Moving and storing data, the part that gets a song to a listener, got far cheaper over the same years. Paying for the song is the part that didn’t.

Three Simple Jobs, Done Slowly

Strip it down and paying for music is three jobs. Capture what the fan did. Calculate who’s owed what. Settle, meaning actually pay them. None of this is hard for a modern computer.

Capture is already fast. Spotify for Artists updates stream counts every day, and every two seconds during a new release’s first week. The money runs on a different clock. Stores generally report a month’s sales two months later, so January’s streams pay in March. The songwriter’s side is slower still. BMI pays a quarter’s earnings three and a half to five and a half months after the quarter ends, and when the song was played abroad, most of the money arrives “between one and two years after the performance.”

That’s the trip our 900-stream track’s money would have taken: months of passing through companies, each with its own minimums, fees and calendar. The amounts weren’t stuck because the arithmetic is expensive. They were stuck because of the trip.

Nobody at the Payment Layer Has to Compete

This is where music differs from messages and remittances. The companies that collect and pay out royalties mostly don’t have to win the business.

In the US, the Mechanical Licensing Collective, or MLC, is the single collective set up by law for one of the two songwriting royalties on streams, the one for reproducing the song. Royalties for songs being performed go mostly through ASCAP and BMI, whose rules go back to 1941 settlements with the Justice Department. Worldwide, 228 collecting societies each license their own territory and pass money to one another. And because labels and publishers own the copyrights, the money also passes through their accounting. Each of these positions comes from a law or from ownership. None of them has to be won again every year by being cheaper.

Their costs look like it. Canada’s SOCAN spends 13.4 percent of what it collects on running itself.

Japan tried the obvious fix, which was to let in competitors. JASRAC was the only organization allowed to collect songwriting royalties from 1939 until a 2001 law opened the market. When the Fair Trade Commission found in 2009 that JASRAC’s licence for broadcasters was “substantially restraining competition,” the case went through two courts for seven and a half years before JASRAC dropped its challenge. Today its main competitor, NexTone, is still much smaller, and the two charge the same 9.5 percent on streaming. More collectors of the same kind didn’t make collecting cheaper.

Whoever Holds the Money Earns on It

Time is money. If a payment system has waiting built into it, someone is probably earning on the money while it waits.

The MLC is the clearest example, because its tax filings are public. At the end of 2024 it held $1.4 billion in securities, nearly all of it owed to other people. That year it reported $72.3 million of investment income, about 65 percent of its revenue, against $44.5 million of interest it owed rights holders. It told the Copyright Office it keeps returns above the rate it owes as reserves. That’s within the rules. But the songwriters whose money it is didn’t choose the investments, can’t see them, and don’t decide when the money comes back. In June the Copyright Office noted that the MLC hasn’t disclosed its financial advisor, the funds it invests in, or how much.

We don’t think anyone delays payments on purpose. Some of the wait even looks unavoidable from the inside, like matching streams to songs nobody registered. We’d call that a design failure: the system accepts money for a song before it knows who owns it. Either way the incentive is plain. Money that waits earns for whoever is holding it, so nobody holding it has much reason to make it wait less.

There’s a second reason, and it’s structural. Every company in the chain keeps its own books. Money moves from the streaming service’s books to a distributor’s or a collective’s, then to a publisher’s, and each handoff waits for one company to close its period and the next to reconcile it. If one company got faster tomorrow, the money would just wait at the next one.

So instead of a faster system, artists get sold the wait. A royalty advance pays you your own future earnings now, minus a cut. Amuse alone has made more than 50,000 of them, and more than half of the artists who pay one off take another, at least four on average. Several of the companies selling advances are distributors, repaid out of royalties already passing through them. Advances help artists who need cash. But there’s a whole business in selling the delay, and none of it makes the delay shorter.

Why Nobody Goes First

If every company moved to one shared record of who owns what and who’s been paid, all of them would spend less reconciling with each other. So why doesn’t it happen?

Game theory has a name for this situation: the stag hunt. A group of hunters can catch a deer only if every one of them holds their position, but any one of them can catch a hare alone. Counting on everyone else is risky, so each hunter goes for the hare, and the group ends up worse off without anyone doing anything wrong. In music, the hare is staying put. Each company keeps its position, and the interest on money in transit, whatever the others do. (Some of them actually do better with the slow system, which makes the deer even less tempting.)

The way out is for the people who gain most to start without waiting for the rest. Spotify says independent artists and labels earned more than $5 billion on Spotify in 2024, about half its royalties, and they don’t need any institution’s permission to get paid a different way.

What We Built

Miso is our attempt to bring competition to the part of the system that pays people. It’s live on Sui Mainnet, the main public network of the Sui blockchain, with Miso Record sales running on it.

Here are the three jobs on Miso. Who owns the song and the recording, and how each track’s money is split, is written down before anyone pays, and it can’t be edited later. A fan pays the release directly, from anywhere. From there the money moves in steps that anyone can kick off, but it can only go where the splits say, and each owner collects their share when they want it. One payment doesn’t pay everyone at once. But nobody waits for another company’s books to close, because there’s only one set of books.

Miso enforces the splits people agree to; it doesn’t decide who wrote a song. Taxes, and converting to and from ordinary bank money, still happen outside it. Miso Streaming is in development, and early prototypes show the design works.

Because anyone can build on it, a store or a player has to compete on being a good store or player. If an artist sells through a different store tomorrow, the money still goes to the same release under the same splits. That’s where we think the competition belongs.

Back to the track with 900 streams. Paying its artist a few cents shouldn’t cost more than the few cents. Elsewhere, the cost of moving money fell where somebody had to compete to move it. Music’s payment layer already has the technology. What it’s missing is anyone who has to compete.