Next Generation Stores of Value: Privacy, Proofs, Compute

v3.5 is the canonical full edition (572 pages). It follows v3.4, the rule-of-its-own pass of 27 September. It ships with brief v3.5 and essay v1.5. It adds no layer and no mechanism to the monetary claim, and no rating is raised.

An external reviewer read v3.4 whole. Their verdict: the infrastructure thesis is strong, the monetary hypothesis is plausible but not established, and the measurement framework is more elaborate than its causal foundations justify. Each note was checked against the source. Most held, several were narrowed, and one was refuted. The decisions taken on them are recorded in Appendix J.

Why hold rather than convert on the day

The record had answered why a holder hedging loss of access would choose the base asset over Work Credits, and why over Bitcoin as an exit asset. It had never answered the strongest alternative: hold Bitcoin, and buy the base asset on the day it is needed. §10b, Why Hold the Base Asset? now opens with that question.

• What holding buys. A balance held in advance removes one dependency: acquisition. That is the whole advantage. It exists only where conversion is unavailable, too slow, or incompatible with the transaction’s privacy and control terms, while native settlement and the service path still work.
• What it does not buy. Holding buys no capacity, price, priority, or queue place. Those are sold through service contracts.
• The counterfactual. The comparison is the best feasible Bitcoin-funded strategy, including a provider that takes Bitcoin and pays the fee from its own inventory. That provider-side conversion is also the bypass the value-capture conditions name.
• The test. An access experiment runs equal budgets, fixed before a disruption, against the same service request. If Bitcoin-funded procurement reproduces the service without material disadvantage, the mechanism is absent for that workload and state.
• Its place in the programme. The experiment is a Phase I deliverable, run first as an analog on an existing network whose fees must be paid in its own asset. It is not a counted red line.
• A pass is not yet a reserve. Agent wallets and gas inventories are operating buffers.

The base asset is not a reservation on the machines. It is a balance that may remain spendable when acquiring that balance would no longer be feasible. Whether that difference survives the best Bitcoin-funded alternative, and supports anything beyond operating inventory, is the monetary experiment.

Three red lines re-specified, before any gate opened

This release does not carry the line that no red line was relaxed. Each of the three re-specifications removes at least one way its line could trip and adds others, and each box lists both. All three changes are prospective. No network exists, no gate has opened, and no reading is erased. Each box in §27 now publishes what passes, what trips, and what cannot be identified under its new rule. The trip wires page carries the new conditions.

• Red Line 16 reads the increment in the pressure corridor the thesis predicts, rather than a linear coefficient on regime pressure. The thesis’s own model is an inverted U, and a linear fit to an inverted U can return a slope near zero while the mechanism is present.
  • It trips when the upper bound on that increment lies below a pre-registered minimum, with the power rule kept.
  • A flat positive reading means the regime increment was not established. It does not mean the service was priced at zero.
• Red Line 17 keeps its same-lender, opaque-comparable design and adds underwriting-model provenance.
  • Agreement under a shared model is not evidence that verification removed an informational advantage.
  • A panel without five independent approaches reads unidentifiable, and that reading runs on a clock.
• Red Line 18B reads freely negotiable short-tenor insider obligations at the same 25% floor, on a level-and-persistence rule.
  • It had read long-dated obligations, which the denomination theory says move last.
  • It had also tripped on a share that was merely flat, so a stable share above the floor could trip it. That failure path is gone.

Definitions that now agree

• Value capture. The Value Capture Lemma’s five conditions are the reference design’s channels: each is necessary to that route, and none to value capture in general.
• Work Credits are prepaid claims on service reserved with a named provider. Capacity is committed, the credit is sold and the capacity reserved atomically, and verified delivery extinguishes the claim and issues the receipt. §10 carries the lifecycle.
• Supply. The capacity modulator on base-asset issuance is gone. Audited capacity constrains service claims, not the monetary schedule.
• The nine properties are the specification, not a universal law of money.
• Collateral is a lockup that holds while operators stay, not a valuation floor.
• Duration-neutral means no contractual nominal duration, not insensitivity to rates.
• Capacity reporting. Delivered capacity reports binding constraints, because max-flow/min-cut fails with gain factors or shared edges.
• The value-versus-price table no longer calls any cell monetary validation.
• Denomination. Contracts quoted in the asset count as denomination, whatever settles them.
• Holders. The buyer map counts spending the asset on the service during stress as exercise, not exit.

Law, dated and per transaction

§24 rests on two sources:

• the SEC’s interpretive release of 17 March 2026 (Rel. 33-11412);
• the Division of Corporation Finance’s staff FAQ as updated on 28 September 2026, whose Q2.5 covers buybacks and protocol-funded burns on a functional system with no central party.

It examines native issuance, secondary transfers, service prepayments, staking and pools, and project finance one at a time. The chain from burn, to distribution, to managerial act, to a stronger securities case is no longer stated as a conclusion. Governance concentration and discretionary return management remain questions to examine. A fixed schedule is not clearance.

Structure, figures, print

• The chapter split. The Work Credits chapter is split. The monetary argument is now its own chapter, served as page 10b so that no other page renumbers.
• The architecture figure is redrawn as a vector with one caption, and its Layer 1 instrument now carries its real name, VerifyReach.
• The deliverables table in the Kardashev Labs chapter now breaks across pages.
• Print fixes. Two misprints present in the v3.4 PDF are fixed: the section sign printed as “ğ”, and the lemma’s negation sign as “ň”.
• History out of the body. Sentences that still narrated the revision process are restated in the present. Authorial “we” is gone from the appendices too.

Brief v3.5 and essay v1.5

The brief opens on the question the infrastructure cannot answer for itself. It carries the access experiment, Work Credits as reserved service claims, and the three red lines as re-specified. The essay moves to v1.5. It asks the convert-on-the-day question before the monetary red lines and drops “the most likely way” in favour of “the failure to watch for”. Brief v3.4 and essay v1.4 remain available, unchanged.

Declined and parked

• Declined: price as evidence through a model. Letting price become evidence “through a model” would loosen a standing guardrail, and the record already says price is never evidence “by itself”.
• Declined: the replacement abstract. The assessment’s replacement abstract dropped the delegation buyer.
• Deferred to v4.0: a full re-spine of the book around the monetary inference.
• Parked: treating repression as a specified environment rather than a diagnosis.

Reproducibility

The checksum files record the LaTeX source commit the PDFs were built from (tag v3.5; essay tag essay-v1.5).

Boundaries held

No dated outcome predictions. No fund, token, or gate. No layer added, no mechanism added to the monetary claim, no rating raised. Red Lines 16, 17, and 18B are re-specified before any gate opened, with the changes published in each box. The Bitcoin concession stands, price is not evidence, and nothing is sized.