Инженеры YouTube призывают создать новую профессию для решения проблем выплат артистам в эпоху ИИ

By Jay Rungta, Engineering Manager at YouTube The music business is a technology business, but the systems doing the actual work of moving money from platforms to artists are held together by engineers who aren't sure what to call themselves. There's no degree in monetization engineering. How did a song become so complicated to pay for? Before streaming and the internet, a sale was a sale, and a performance was a performance. The math may not have been easy, but at least the shape of it was simple. A unit or a ticket sold, combined with classical accounting, meant that the checks went out. Thanks to the digital revolution, that shape was broken entirely. Now, short-form video and GenAI are set to break it all again. A cascade of calculations need to be triggered to calculate royalties on a three-second clip of a licensed track in a short-form video. First, it's figuring out the actual revenue generated against the clip. Then, multiplying that against a slate of ever-evolving deal terms: which means taking into account the underlying composition rights, master rights, user-generated content licensing terms, niche deals, global scale, and many other dimensions. This has a lot of room for a lot of things to go wrong, especially when you consider its massive global scale, across region specific rules and currency conversion rates. There’s billions of these events every day, and we can’t have humans checking the final numbers before they go out. With that much data, it's just not possible to. "A system that's 99.9% reliable and accurate sounds great, until you run it against billions of dollars of calculations." What breaks when you treat this like regular software. The engineers who built streaming infrastructure successfully built them to serve content to hundreds of millions of people reliably, which in itself was an enormously hard problem to solve. The issue is that royalty systems were also made by these teams, and the same expertise doesn’t transfer. The failure modes on the product engineering and the monetization engineering side are completely different. A bug in a content recommendation algorithm may surface the wrong video, but a bug in a royalty calculation can compound quickly across millions in payments. Until a quarterly report lands on a rights holder's desk, and they start asking why the numbers don't add up. By then, the money has already moved, kicking off long disputes or clawbacks. When reporting is opaque and reconciliation keeps failing, the business relationships eventually break down. +Read more: "5 Signs a Spotify Playlist Is Fake (and How Playlist Bot Checkers Actually Work)" What this discipline actually requires. So what does it take to do this right? A few things that general software engineering doesn't ask for. The business logic has to live inside the system. You can't build rights clearance architecture without actually understanding what a rights holder needs to see — which means understanding their contracts, beyond just their data formats. To be good at this, you need to sit at the intersection of technical depth and business fluency. Financial precision at scale is a different problem than performance at scale. A system that's 99.9% reliable and accurate sounds great, until you run it against billions of dollars of calculations. We’re talking about error margins of millions of dollars even then. The bar is higher than it sounds, and the margin for "good enough" is basically zero when real money is moving. Couple this with the ever-evolving landscape of the space. The underlying deals change constantly. Rights structures can evolve over time. New content formats arrive faster than any platform's legal team can fully anticipate. A monetization system that can't adapt is inviting legal and audit troubles. Flexibility has to be a first principle, not an afterthought. The part that's about to get harder... The rise of short-form video already stressed these systems bad enough that some platforms are still catching up. Meanwhile, GenAI has arrived even as those stress fractures are still visible. When a platform uses AI to generate music, who holds the rights? When a creator uses a voice model trained on a real artist's catalog, how does the royalty get split? These aren't hypothetical edge cases anymore, they're showing up in content pipelines as we speak. The platforms that have treated monetization engineering as a real discipline — with dedicated expertise, real investment, proper architecture — will have options as the pay out structure of these assets become clearer. The ones that haven't, will be dealing with takedowns and high-profile disputes, all the while dealing with consumers being annoyed that Taylor Swift songs are missing from their music picker. "The backend, the actual plumbing that gets money to the right people in the right amounts under the right contract terms, gets treated like operational infrastructure. It’s always someone else's problem… until it isn't." Why does this really matter? As mentioned before, there is no established career path into monetization engineering. Engineers stumble in from platform infrastructure, or from fintech, or occasionally from the music business itself, and they learn by building things and getting them wrong. The music industry has spent enormous energy thinking about the frontend of the creator economy — discovery, fan engagement, direct-to-artist platforms. The backend, the actual plumbing that gets money to the right people in the right amounts under the right contract terms, gets treated like operational infrastructure. It’s always someone else's problem… until it isn't. Since it's clearly the foundation everything else runs on, the industry probably should have a name for the engineers building it. +Read more: "Music After the Generative AI Creative Big Bang" Jay Rungta is a web monetization expert specializing in the financial infrastructure behind digital content platforms, from real-time royalty calculation to high-speed trading systems. At YouTube, he has built core architecture for music royalty payouts at massive scale. Earlier, he built the equities trading application layer at Millennium Management. His work centers on making sure money moves correctly at scale. He writes about the engineering challenges underlying financial systems.