The Cost of Checking

The Cost of Checking

Every argument in this essay reduces, at its load-bearing joint, to a single legible quantity: VerifyPrice—the real-resource cost of verifying that a claimed computation, settlement, or attestation actually happened.

Production is expensive; verification is cheap. That asymmetry—the touchstone problem—is the entire technical basis for public truth at scale. When checking costs one ten-thousandth of producing, the discipline stops depending on anyone actually checking. It is enough that they could: a claim any counterparty could falsify for nothing is not worth making, so it is not made. That is what a cheap assay buys—not an audit, but the fact that lying stopped being free—and it is why delegation becomes safe rather than unnecessary. When checking costs the same as producing, you have a priesthood with extra steps. VerifyPrice is the dial on that machine, and it is measurable: milliseconds and dollars on reference hardware, for a standardized workload, published as a distribution anyone can reproduce.

Three services ride on the dial, and every one of them is a monetary surface:

• Private settlement. Moving value without exposing the transaction graph to competitors, platforms, or hostile states. The instrument: VerifySettle—refund-safe, atomic, non-custodial corridors, measured.

• Portable proof. Attesting that a computation ran, a fact holds, or a rule was satisfied, without trusting the platform. The instrument: VerifyPrice itself.

• Verified compute. Machine work with receipts anyone can check, turning capacity into a commodity that can be priced, hedged, and held. The instruments: VerifyPrice plus delivered-verified-capacity telemetry that discounts nameplate promises by stress.

And one number deepens it. The capturable wedge is what remains of the price a protocol commands after paying for the verification that justifies it: the premium over the best unverified alternative, minus the cost of proving. If the wedge is thin, the fee is a tax on captive volume, and whatever the technology achieves, the asset will not carry it. The wedge is the difference between useful and monetary, and it is published quarterly or it is nothing.

Stated without the machinery, the wedge asks whether anybody actually pays for the truth. Verification can be cheap, correct, public, and comprehensively ignored. A system that is auditable and whose auditability moves no price has produced receipts and nothing else, and that failure has no technical signature—every other instrument on the panel stays green while it happens. It is the mode in which this thesis is most likely to be quietly wrong, which is why it gets a number rather than a paragraph.

Gold was assay-able; that was never enough to make it money. The question is never “can you check it?” but “what does checking cost, and who pays?”—and that is a number, not a vibe.

The second half of that question deserves an answer, because it is the objection an economist raises first. If verification is free, nobody earns a return on verifying, and work nobody is paid for does not get done. The resolution is that the cost and the benefit fall on different parties. The prover pays, because being believed is what they are buying; everyone else consumes the result at no charge. Verification is not an underprovided public good waiting for a subsidy—it is a private purchase whose public checkability the buyer cannot withhold without destroying the thing they bought. That is also, uncomfortably, why the proving business earns an ordinary return: credibility sold into a competitive market is priced at the cost of supplying it. The property that makes verification sustainably provided is the same property that stops it from being a source of excess return, which is precisely why this thesis withdrew the claim that fee revenue was evidence of moneyness.