Federated payments keep failing for the same reason

Every federated-payments project tries to solve federation, money, and identity simultaneously, and is then surprised when the intersection kills it. The pattern that actually ships (Stripe, Cash App) solves two and punts the third. The fediverse keeps attempting all three, which is how you end up with fifteen projects at eighty percent and none at a hundred.

Parent discussions

• Federated Payments — the research summary this pattern came out of

Connected concepts

• Proceduralism as Hidden Sovereignty — punting identity to Stripe means punting it to the KYC apparatus of the state; the “neutral” payment rail is a sovereignty decision wearing an API
• Human Decision as Development Artifact — “which leg do we punt” is the decision these projects refuse to make, and the refusal is legible in their architectures

The point

Name the three problems. Federation: value moves between servers run by parties who don’t trust each other. Money: real currency, with settlement, fraud, and regulators attached. Identity: knowing who is paying whom, durably, across the network. Any two are hard. All three is a research program disguised as a product.

The companies that shipped picked two. Stripe solved money and identity (settlement on the card networks, KYC on state-issued identity) and punted federation entirely: one company, one API, no peers. Cash App solved money and identity inside a closed social graph. The punt is not a compromise they regret; it is the load-bearing decision. Stripe’s whole value is that you never have to trust a stranger’s server.

The fediverse punts nothing. Mastodon’s monetization experiments want ActivityPub federation, real money, and portable cross-instance identity all at once, and each pairwise seam is itself unsolved. Money plus federation means settlement across trust boundaries. Identity plus money means KYC that no volunteer instance admin can perform. The open FEP landscape tells the same story at the spec layer: payment FEPs have been proposed and none adopted, because a standards body cannot ratify its way past a trilemma. Lens Protocol is the clever variant. It takes federation and identity (a portable social graph on Polygon) and outsources money to crypto rails, which relocates the trust problem rather than solving it. Outsourcing a leg is still punting it; Lens just punts to a rail with its own unfinished trilemma.

The lesson for the thing I actually care about, economic viability for self-hosted creators, is uncomfortable. The honest MVP federates content and discovery, keeps real money, and punts identity to Stripe. Boring. Centralized at the exact edge where centralization is the entire value. Shippable. The purist version that refuses the punt joins the fifteen at eighty percent.

Open questions

• Is punting identity to Stripe scaffolding or foundation? Does the third leg ever get built afterward, or does the punt calcify into the architecture?
• Is Lightning a real answer to settlement-across-trust-boundaries, or a fourth problem imported to solve the second?
• What is the smallest payment FEP that could actually be adopted? Federating only the pointer to payment (a standard “pay this creator” link object), moving no funds at all.