Illustrative Scenarios

Events are the ephemera of history; they pass across its stage like fireflies, hardly glimpsed before they settle back into darkness and oblivion.

An invented case cannot prove an institution. It can expose the decisions that an institutional design would otherwise conceal. The three demonstrations below use familiar elements—contracts, insurance, collateral, sensors, settlement rails, and specialized forums—but none reports an existing deployment or predicts that one will prevail.

The point is not to watch the book's concepts operate flawlessly. It is to discover where they stop.

A Covenant Meets an Ambiguous Defect

A solar developer has authorized a computational system to administer procurement and milestone payments for a remote installation. Its mandate is real but bounded. It may compare bids, execute standard purchases, release funds against specified evidence, and suspend a payment when records disagree. Engineers and a finance officer set those boundaries. The developer remains the principal.

A shipment arrives with intact serial records and the required port attestations. Payment releases. Weeks later, infrared inspection reveals microfractures in part of the installed array. The evidence establishes the defect but not its cause. The panels may have left the factory damaged. They may have been mishandled at the port, stressed during transport, or installed incorrectly. Each explanation activates a different warranty, insurer, and body of evidence.

The covenant can freeze the next payment and assemble what is known. It cannot turn ambiguity into breach merely because a decision is due. More sensors may narrow the possibilities, but the case is not governed by verification cost alone. It is governed by the meaning of several contracts whose categories overlap imperfectly.

Collateral does not solve that problem. A bond answers the question what can be collected if a condition is established? It does not establish which condition occurred, whether the condition was defined sensibly, or who had authority to apply it. Automatic slashing at this point would reward speed by assigning loss before responsibility had been found.

The parties therefore use an independent technical forum authorized in advance to order inspection, interpret the warranties, and issue a binding allocation of the immediate repair cost. Its decision is provisional with respect to later contribution claims among the supplier, carrier, and installer. Work resumes before the complete causal history is known.

Here the proposed architecture is reordered. Evidence does not lead directly to execution. It leads to a forum. The most valuable part of the system is not its ability to make an ambiguous judgment quickly, but its ability to preserve the dispute, finance an interim remedy, and keep the surviving principals within reach of a later answer.

Clinical Speed, Institutional Time

A hospital uses a computational triage service. The service may order routine tests, prepare a care pathway, and alert a clinical team under policies adopted by the hospital. Different classes of act carry different authority. A physician need not approve every laboratory order. An invasive intervention remains subject to professional and institutional controls.

A patient arrives with signs consistent with several conditions, one of them time-critical. The system can read the record and compare the pattern faster than the clinical team can assemble. It can also be wrong in a way that becomes legible only after treatment changes the evidence. Ground truth is not simply expensive. In some cases it is altered by the act taken in response to the prediction.

The hospital has therefore delegated preparatory action rather than final judgment. The service orders confirmatory tests, alerts the relevant team, reserves capacity, and presents the evidence used to select the pathway. If specified indicators cross an agreed threshold, the protocol permits a narrow set of reversible measures while a clinician is reached. It does not confer a general authority to treat.

Suppose the first tests support the alert and the later course shows that the dangerous condition was absent. The episode cannot be scored by asking whether the model's initial probability was “correct.” The hospital must ask whether the action was authorized, whether the threshold was defensible, whether the available evidence was represented faithfully, whether the intervention was proportionate, and whether the patient can obtain an explanation and remedy.

No single liability sink contains that inquiry. The vendor may answer for a defective system. The hospital may answer for adopting the policy. A clinician may answer for an act that required clinical judgment. An insurer may bear part of the loss without possessing the authority that caused it. The record matters because it lets those offices be distinguished. It does not decide the dispute.

This case resists a simple automation gradient. Cheap tests can widen delegated preparation while leaving the decisive act under professional authority. Better performance may reduce review in one interval and increase institutional demand for monitoring, explanation, and appeal in another. Verification changes the boundary. It does not abolish the boundary's politics.

Settlement Across a Fractured Supply Chain

A procurement firm uses computational systems to compare suppliers, prepare contracts, monitor shipping evidence, and initiate payments across several jurisdictions. Most routine orders travel through conventional banking because suppliers price goods in local currency, accounting systems already recognize the rails, and legal recourse matters more than resistance to account restraint.

One class of transaction has a different threat model. A supplier and buyer operate across jurisdictions whose authorities may interrupt payment for reasons unrelated to the commercial dispute. For that class, the parties compare bank escrow, a regulated stablecoin, Bitcoin, and a contractual reserve held by an independent intermediary.

No rail wins in the abstract. Bank escrow offers familiar law and reversible error while preserving bank and state control. A stablecoin offers programmable transfer while retaining issuer, reserve, freeze, and redemption dependencies. Bitcoin reduces dependence on an asset issuer and makes confirmed history costly to replace, but introduces price, custody, fee-market, confirmation, conversion, and key-control risks. A contractual reserve can put assets near an adjudicator while recreating a custodian.

The firm uses Bitcoin for a bounded contingency reserve and ordinary bank money for invoices. That decision does not make electricity a universal price floor. Mining has no direct bearing on the firm's procurement workload unless an operator can actually route the relevant power and capital into suitable hardware. What Bitcoin contributes here is narrower and stranger: a bearer claim and transaction history whose production and revision are conditioned by protocol-recognized work rather than an issuer's account entry.

Then a public authority prohibits payment to one supplier. The procurement system can detect the restriction and suspend execution. It cannot decide that evasion is legitimate because the underlying contract remains valid, nor can the protocol make an unlawful payment lawful. If an authorized person nevertheless uses the reserve, proof-of-work may help the network order the transaction. It does not supply authority, excuse, or immunity at the endpoints.

The scenario therefore frustrates the fantasy of a settlement layer outside institutions. Permissionless validation can remove one veto and expose another. The firm still needs an answerable principal, a lawful policy for exceptional acts, evidence that preserves what happened, and a forum able to decide consequences that the rail itself cannot recognize.

What the Cases Establish

The three cases do not share one mechanical progression. The solar dispute makes adjudication precede final allocation. The clinical case separates preparatory speed from authority over irreversible action. The procurement case uses several settlement systems at once and shows why protocol validity cannot decide institutional legitimacy.

They do share a pressure. Computation can lower the cost of proposing, comparing, recording, and sometimes executing action faster than institutions can lower the cost of bearing its consequences. A defensible design does not pretend that every ambiguity can be collateralized or every delay eliminated. It decides which judgments may be delegated, preserves the evidence needed to contest them, and keeps a principal and a remedial forum alive after the execution ends.

That is a demonstration of the volume's argument, not evidence that the demonstration will become the world.