The quirq

The unit of work defines and verifies the job. The quirq is what that verification mints: the output meter, dual to the token's input meter.

Every consequential technology gets two meters: one for what it consumes, one for what it produces. The token is a genuinely good input meter (it counts compute and scales with energy and carbon), but it structurally cannot price work: it rises when work goes badly, changes size when models are swapped, and cannot tell a $4 ticket from a $25,000 contract. The measurement crisis of enterprise AI is a units problem, and the quirq is the missing output meter.

The definition

A quirq is one unit of budget currency of verified, owner-valued, delivered work. It is minted, never self-reported:

1. A human owner defines an outcome with a machine-checkable definition of done and budgets it at value B. The task now carries B potential quirqs.
2. The environment snapshots the world before execution (S₀) and after (S₁) and scores completion V ∈ [0, 1] against the definition of done.
3. Verification mints V · B quirqs, recorded in a tamper-evident ledger with everything production consumed.

potential quirqs   = B          set by the human owner, before execution
minted quirqs   Q  = V · B      released by verification against captured state

Three properties make it a unit of account: human-denominated (the budget is demonstrated willingness to pay, set by the party paying), machine-minted (verification is a state comparison the worker never produces evidence for), and cost-complete (metered against inference, compute, API calls, storage, and the human minutes spent intervening).

Divide the two meters and the missing numbers appear: cost per quirq, quirqs per dollar (QER), quirqs per kWh. Track them over time and "is AI working here" becomes a trend line instead of an anecdote. The full treatment (the token's exact physics, the contract-theory case for the budget denominator, the calculus, the dashboard, gaming, and validation) is the canonical quirq section.

The mint needs machinery it can trust: something has to capture the snapshots, run the checks, and meter the spend without ever asking the worker. Next: the research note, on why that machinery, the environment, is the load-bearing component.