v0.1 (Character Blocks, Skillchain, Workchain, Federation) + v0.2 (royalty-cascade, deferred settlement) · §3 Character Blocks, §6.2 V3 Split, §7.5 Federation Treaty Layer, §9.5 Banks as Trust Sources

Crystallised Labour, Paid Forever

The sound engineer’s snare hit ends up in a Kanye song. A billion people hear it. The engineer earns $200 once and nothing thereafter. HOP turns that into roughly a trillion micro-payments — each smaller than a tenth of a cent — flowing back to the engineer for the lifetime of the snare’s propagation. This entry shows the mechanism.

The situation

Some work evaporates the moment it’s done. The nurse who comforts a dying patient for ten hours produces value that is real, needed, and gone — consumed in the moment it was given. There is no asset left over. The hour is the hour.

Some work crystallises. A snare hit recorded once becomes a sample. A function written once becomes a library. A photo taken once becomes a wallpaper. A poem written once becomes a quotation. The hour produces something that persists, and the persistence is what makes the work compound long after the work is finished.

The market rewards crystallisation, not creation. (Read the underlying framing in Paragon Asia Dataflow.)

The harder version of the claim — service-vs-crystal is not a property of the work, it’s a property of who holds the frozen time afterward:

Service = my time melts into your life and I get paid once. Crystal = my time freezes into an object and someone (maybe me, maybe not) gets paid forever. AI doesn’t break the binary — it sharpens it, because now even service work gets crystallised, but the crystal forms in a vault the worker never had access to.

The reason the worker never has access is mechanical, not malicious. The infrastructure to pay her every time her crystal is used has not existed.

The example

A sound engineer in a basement studio in 2026 spends an afternoon getting one snare hit right — exact tuning of the drum head, exact compression curve, exact room mic placement, three takes. The session pays her $200. The producer who hired her loops the snare into a sample pack and sells it for $40 a copy.

A few hundred producers buy the pack. One of them lays the snare into a beat. A vocalist tops it. The song goes to a major label. A billion people stream it. Each stream plays the snare maybe forty times. That is forty billion plays of one snare — none of which flow back to the engineer.

Compound: over the next decade, a thousand other producers sample the same snare into derivative work. Each of those tracks gets streamed. The engineer’s snare ends up in millions of derivative works, played at the scale of trillions across her career.

Total flow back to the engineer in the current system: $200.

She made a crystal. The crystal pays everyone except her.

What HOP does

Five steps. Each is a real protocol operation, not a metaphor.

1. Creation gets signed

When the engineer finishes the snare hit, her tooling emits a Character Block — the atomic unit of HOP (§3):

{
  "block_type": "creation",
  "content_hash": "sha256:8e4c…",
  "inventory": {"medium": "audio", "kind": "snare_hit",
                "license_class": "royalty_cascade"},
  "context": {"session": "studio_2026_03_14",
              "created_with": "ProTools_12"},
  "prev_block_hash": null
}

Dual-signed by the engineer and the studio. The block lands on her Skillchain. The content hash is now a globally addressable reference to this specific snare — anywhere in the world, any system that ingests audio with this hash knows who made it.

2. Derivation gets signed

When the producer drops the snare into a beat, the beat’s Character Block carries a derived_from field pointing at the snare’s content hash:

{
  "block_type": "creation",
  "content_hash": "sha256:f12b…",
  "inventory": {"medium": "audio", "kind": "beat"},
  "derived_from": [
    {"hash": "sha256:8e4c…", "kind": "snare_hit", "weight": 0.06}
  ],
  "context": {"daw": "AbletonLive_12"}
}

weight: 0.06 declares that 6% of the beat’s value is attributable to the snare. The producer’s signature attests to the declared mix.

The vocalist tops the beat; the verse block points at the beat. The song block points at the verse and the beat. Each step is a Character Block signed by the next author. By the time the song is released, every derivative work carries the full cryptographic chain back to its sources.

3. Use gets observed

When the streaming platform plays the song, the play event is itself a signed block:

{
  "block_type": "use",
  "content_hash": "sha256:c5a3…",
  "use_event": {"played_track": "sha256:<song>", "duration_ms": 218000},
  "signed_by": "ed25519:<streaming_platform>"
}

Platforms already run this telemetry; the protocol change is that the event is signed and federated rather than locked inside the platform’s data warehouse.

4. Payment flows through the graph

The streaming platform’s Workchain has a federation treaty with the rights-holder chain (§7.5 — Federation Treaty Layer). The rights-holder chain federates with the producer’s chain. The producer’s chain federates with the engineer’s Skillchain.

The play event walks back through the derived_from graph. At each step, the chain’s configured royalty share — declared in its Anchor Block (§2.1) and the V3 split (§6.2) — cascades the payment upstream. The engineer’s wallet ticks up by a fraction of a cent.

The math: a play that pays the rights-holder $0.004 (typical 2026 streaming royalty), with the producer chain’s split sending 30% upstream, with the snare attribution at weight 0.06, settles $0.000072 to the engineer per play.

5. Aggregation closes the loop

Forty billion plays × $0.000072 = $2.88 million to the engineer, automatically, with no negotiation, no licensing department, no lawsuit.

Over the snare’s full propagation life across a thousand derivative tracks, the cumulative flow is plausibly in the range of $5–15 million across her career — a small share of what the rest of the value chain captured, but proportional to the actual contribution of the work she did.

But citation is the floor, not the ceiling

The five steps above pay the engineer for being used. That’s fair. It is also the floor.

Two snares get sampled into a thousand tracks each. Snare A is in tracks that average a million plays. Snare B is in tracks that average ten thousand plays. Under citation-only, both engineers earn the same per-use cascade × the playback rate. The engineer who made the better snare — the one that demonstrably caused songs to do better — earns more only indirectly, because her tracks ended up with more plays. The credit is downstream of the outcome rather than attributable to her work.

HOP also measures the causal contribution directly. The mechanism is the same shape as §6.3.5 Bean Chain mentorship measurement, applied to derivation instead of mentorship:

Bean Chain (mentorship)	Same shape, applied to crystallised labour
T1 — measure mentee’s skill vector at baseline	T1 — measure baseline performance of derivative work not using snare X
T2 — mentorship event; Bean staked	T2 — producer uses snare X; derivation block signed
T3 — measure mentee at +12 months	T3 — measure performance of derivatives using snare X, matched on other attributes
T4 — confirm sustained outperformance	T4 — confirm sustained lift; cumulative bonus settles

The protocol projects the performance delta onto the “this work uses snare X” direction. Positive projection = causal lift attributable to the snare; the engineer collects a bonus payment proportional to the measured lift. Negative projection = the snare actively hurts the work it’s in; the bonus is zero, or, under v0.3 clawback, negative.

The math, with numbers. Suppose her snare causes a 15% lift in average track success across the thousand derivative tracks. Average track lifetime revenue: $10,000. Total measured causal value attributable to the snare: $1.5M. Even at a small share — say 5% to the engineer — she earns $75,000 in lift bonuses from one snare, on top of the per-play cascade. The lift bonus alone is several hundred times the original session fee, settled automatically against measured outcomes.

For the engineer who made the world’s best snare sound — the one a producer reaches for first because it actually moves the needle on her tracks — the lift bonus is the dominant income stream. The per-play cascade pays everyone proportionally to use. The lift bonus pays the best proportionally to how much better they make the work that uses them. The protocol can distinguish them because the substrate already carries the signed derivation graph and the play telemetry.

This is the deeper claim of the materialisation: HOP makes excellence economically legible. Not by reputation (which can be performed) but by measured causal contribution at scale, paid out at the cadence the contribution actually unfolds at.

Two further properties citation alone doesn’t deliver:

Distinguishes the snare that gets played from the snare that drives the play. Two producers both sample her snare; one makes a song that goes viral, the other doesn’t. Citation pays the same per play. Lift attributes the difference to the snare that caused it — the engineer earns more from the song where her contribution actually mattered.
Resists gaming. Producers cannot farm her snare into a thousand identical low-performance tracks to game her royalty, because the lift score would crater and the bonus would not pay out. The §6.3.4 anti-collusion safeguards (pairwise discount caps, Christiano-style trust matrix) carry over directly from the mentorship case to the derivation case.

Why this is impossible today and tractable in HOP

Two impossibilities compose in current infrastructure.

The micropayment rail. Per-transaction cost of moving money on the cheapest rails is roughly one cent. You cannot pay anyone a hundredth of a cent because the rail overhead is a hundred times the payment. The engineer is locked out of micropayments structurally — not by malice, by infrastructure. In HOP, the payment is a state change on a chain. No bank fee. No card-network interchange. The atomic operation is write a number into the engineer’s balance. The cost of doing this a trillion times is the cost of a trillion writes on a chain, which is the cost of a trillion bytes of storage and a trillion signature verifications — measurable, finite, far smaller than a hundredth of a cent each. The chain settles in batches; the engineer off-ramps to fiat on whatever cadence she chooses (weekly, monthly, annually). The off-ramp pays the rail fee once. The micropayments inside the chain incur no per-event fee.

The causal-attribution measurement. Today, finding out which of two snares actually drove a song’s success is a marketing-attribution problem — A/B tests, incrementality studies, careful counterfactual matching. Expensive enough that nobody runs it for the engineer; only ad networks and the very biggest platforms can afford to. In HOP, the substrate already carries the signed derivation graph (step 2 above) and already observes use (step 3). Running the T1–T4 measurement is a query over data the protocol already holds, not a new study commissioned each time. The cost of measuring the causal contribution drops by orders of magnitude because the data was on the chain the whole time.

What’s different

A world in which every snare hit, every line of open-source code, every Stack Overflow answer, every news article, every dataset row, every model-training image pays its author forever is a world where the people whose work crystallises hardest are paid proportionally to how much it propagates.

It is also the operational answer to Paying the Bees for Honey: the AI lab that wants to train on human-generated text queries the global data tree, ingests only attested-author content, and pays each author per-token-included at sub-cent rates. The lab’s cost: a fraction of what they currently spend on copyright legal exposure. The authors’ earnings: distributed proportionally, automatically, by the math.

The frame from the corpus, stated plainly: the crystal still forms downstream of the work, but the crystal is now cryptographically attached to the worker who made it. The vault is no longer the platform’s. The vault is the chain.

What’s in v0.1 and what’s pending

In v0.1, today — works in the Python reference implementation:

Character Blocks with derived_from references (the field exists; the implementer adds it to the inventory schema)
Skillchains accumulating signed creations
Workchains observing use events
Federation treaties between chains
Per-attestation payment from a relying party (the §9.5 Banks as Trust Sources pattern, applied to royalty flows instead of identity attestations)

Pending v0.2:

The walk-the-derivation-graph royalty cascade — a multi-week build; the protocol is straightforward, the federation diplomacy is the gating item
BBS+ selective disclosure for plays that need to hide listener identity
Bean-Chain-style deferred settlement for delayed or contingent royalty events
A reference implementation of the cascade walker that traverses the derived_from DAG, applies per-edge weights, and emits a payment intent the off-ramp can settle

Pending v0.3:

Cross-jurisdiction tax-withholding at the off-ramp
Streaming-platform federation treaties standardised across the major rails

A v0.1 implementer could build the snare-hit case for a single platform pair (one DAW, one streaming service) in roughly a month. The harder problem is not the protocol; it is the federation treaties between the rights-holder chains and the streaming chains. That is a conversation, not an implementation.