Memo. Executive Memo

Jason St George. "Memo. Executive Memo" in Next‑Gen Store of Value: Privacy, Proofs, Compute. Version v1.0. /v/1.0/read/front-matter/executive-memo/

Executive Memo (Front Matter)

Next-Gen Store of Value: Privacy, Proofs, Compute Private Money & AI Money — Monetary primitives for the AI era Jason St George

A cypherpunk monetary stack: privacy by default, proofs by construction, and compute that only gets paid when anyone can verify it.

The thesis in one page

Every monetary epoch begins with an argument about what is real. In the 20th century, money rode on soft guarantees: central banks, custodians, broadcast narratives, credentialed authority. In the 21st, those guarantees are under compounding stress: debt stocks that cannot be honored in real terms make financial repression arithmetically attractive; identity and network infrastructure are increasingly fused into surveillance and enforcement; and AI breaks the assumption that “seeing is believing.” In that environment, stores of value that rely on reputation, gatekeepers, or custodians are brittle.

This thesis makes a concrete claim: three cryptographic capacities can function as monetary primitives for a dense digital civilization:

• Privacy — censorship-resistant settlement that preserves agency
• Proofs — portable attestations of computation and provenance (“receipts,” not biographies)
• Compute — useful work (matmul, inference, ZK proving) wrapped in succinct guarantees and priced as a verifiable commodity

The goal is not a slogan. It is an engineering and economic agenda: make Privacy, Proofs, and Compute behave like verifiable necessities—things the world must keep buying through every cycle—on rails anyone can audit.

Why now

We assume a world of chronic financial repression and pervasive identity/network surveillance, not benevolence. In such a world:

• Compliance becomes the reserve. Post-Bretton Woods money trends toward enforcement-backed stability: capital controls, de-banking, asset freezes, and regulatory chokepoints as policy tools.
• Reality becomes contestable. Synthetic media and model-generated content erode common knowledge. Without receipts, “trust” collapses into platform policy.
• Compute becomes a strategic commodity. AI makes verified computation structurally demanded: inference, proving, provenance, and secure execution become the new industrial base.

So we look for new anchors—things that cannot be forged, that do not ask permission, and that anyone can verify.

The hinge: verification asymmetry

The economic core of the thesis is verification asymmetry: the gap between the cost to produce a claim and the cost to verify it.

For a canonical workload $W$, define:

$r(W) = \frac{v(W)}{p(W)}$

where $p(W)$ is production cost and $v(W)$ is verification cost. If $r(W) \ll 1$, verification is cheap relative to production and markets can form without trust. If $r(W) \to 1$, we’re back to “trust the prover,” i.e., platform IOUs.

We operationalize this with a public KPI vector:

VerifyPrice(W) = (p50/p95 verification time, p50/p95 verification cost, failure rate) under realistic network conditions, on reference verifier hardware (baseline: a laptop). VerifyPrice is the hinge that determines whether proofs and verified FLOPs behave like commodities (publicly checkable) or like permissions (someone must bless them).

The loop: Create/Compute → Prove → Settle → Verify

We propose a modular, end-to-end loop that turns intent into receipts and value flow into auditable artifacts:

1. Create/Compute — define an intent or perform useful work (matmul, inference, ZK proving, provenance)
2. Prove — compile the claim into succinct proofs and receipt artifacts
3. Settle — move value non-custodially on privacy rails
4. Verify — let anyone independently check correctness, policy compliance, and settlement safety

This loop is the “unit cell” of the stack: it is how private payroll clears under capital controls, how media provenance survives platform deletion, how verified inference becomes a market, and how proof/compute procurement becomes auditable infrastructure.

The stack: seven layers, from base reality to governance

The triad does not exist in the abstract; it must be supplied under adversarial conditions. We therefore define a seven-layer cypherpunk stack:

• Layer 0 — Verifiable Machines & Energy: open hardware, sampled supply chains, auditable power; if the machine can lie or cannot stay on, nothing above it matters
• Layer 1 — Reachability: communications that survive DPI, filtering, and shutdowns (and make verification real, not theoretical)
• Layer 2 — Distribution & Execution: reproducible builds, signed updates, multi-home delivery; honest clients under app-store/CDN/DNS weaponization
• Layer 3 — Identity & Claims: humans and machines prove capabilities and rights without doxxing; reputation is receipts
• Layer 4 — Truth & Work: proof systems + PoUW; canonical workloads; proof factories; VerifyPrice observatory
• Layer 5 — Value & Settlement: privacy rails and non-custodial flows; refund-safe corridors; settlement as a workload with receipts
• Layer 6 — Governance & Telemetry: “no dashboards, no trust”—public SLOs as constitution; drift and capture made measurable, not denied

The instrument: Work Credits

At the monetary center we define Work Credits: energy-anchored claims on standardized units of triad work (privacy settlement, proof generation, verified compute) produced under public SLOs. Credits are minted only when valid work receipts are accepted and telemetry confirms health bounds. They are not debt instruments: no coupons, no fiat promises—value floats with demand for triad capacity. The intended property is simple: scarcity tied to energy, hardware, and verification constraints—not decree.

What we are proposing (concrete deliverables)

This thesis specifies a modular stack of twelve primitives and four reference applications:

Primitives (selected): Proofs-as-a-Library (PaL) to compile claims into proofs; a Privacy Rails Kit (PRK) for non-custodial, refund-safe settlement; a minimal receipt schema (PIDL) so proofs and settlements become portable artifacts; canonical workload registries and harnesses (MatMul-PoUW, verified inference); SLA escrow/slashing; neutral routing; bridge/corridor safety templates; telemetry agents and a Verify* observatory.

Reference applications: (1) private treasury & payroll, (2) media provenance & authenticity, (3) verified inference, (4) proof/compute procurement markets.

The ambition is not “another chain.” It is a research and engineering agenda—a Bell Labs for proof-of-useful-work and lawful privacy—turning privacy primitives, zero-knowledge, and verifiable compute into everyday infrastructure.

The falsifiable test: VerifyPrice / VerifyReach / VerifySettle

This thesis is designed to be falsifiable. If the system cannot keep these metrics healthy, it becomes another platform IOU:

• VerifyPrice(W): verification remains cheap (p95 bounded), failures rare, overhead $r(W)$ stays $\ll 1$
• VerifyReach(N,R): reachability under censorship stays high (time-to-first-connection, success rates, peer diversity)
• VerifySettle(C): settlement remains robust (success, refund safety = 1.0, time-to-finality, anonymity-set health)

Neutrality and repression-resilience are not asserted; they are measured.

Who this is for

• Builders: if you’re building privacy rails, proof systems, verified compute markets, or censorship-resistant distribution, this is a map of failure modes and a blueprint for receipts-first infrastructure.
• Operators and allocators: if you fund or run networks, this provides a checklist and telemetry regime for distinguishing commodities from IOUs.

If you only remember one line: the next reserve asset is not a single object but a triad of verifiable necessities—Privacy, Proofs, and Compute—kept honest by public metrics and built from the machine layer up.

Tip: hover a heading to reveal its permalink symbol for copying.