# quirq: A Unit of Work for Intelligence
# Suraj Sharma, XO Labs · Draft v3 · July 2026
# Canonical: https://xo.builders/whitepaper
# Machine-readable companion: https://xo.builders/whitepaper/vectors.json
#
# This file is the whitepaper as one plain-text document, prepared for
# language models and agents. Text extracted from the canonical PDF;
# figures and tables are described in prose where extraction allows.

quirq:
A Unit of Work for Intelligence
Suraj Sharma
XO Labs Inc
Correspondence: suraj@xo.builders
Draft v3, July 2026
Abstract
AI today is metered on one side only. The token is a
genuinely good unit for what the machine consumes: it
counts compute, scales with energy, and prices inference.
But nothing counts what the machinedelivers, and an
input meter without an output meter cannot answer
the only question a business asks: is this working? We
propose thequirq, a unit of measurement for the busi-
ness impact of agentic work. A quirq is minted, never
self-reported: a human owner budgets an outcome at
value B; an environment snapshots the world before and
after execution and scores completionV ∈[0,1] against
a weighted definition of done; verification then mints
V·Bquirqs of delivered work. Against the minted quirqs
we meter theall-in cost of production: inference tokens,
CPU and GPU time, external API calls, storage, and
the human interventions the work still required. From
one ledger of units, every number an operator needs falls
out: cost per quirq, quirq margin, the Quirq Efficiency
Ratio (quirqs delivered per all-in dollar), quirq velocity,
the intervention rate, and their trajectories over time,
which quantify, week by week, how effective AI actually
is inside a company. Tokens and quirqs are duals: one
meters the machine’s draw on the world, the other meters
the world’s change by the machine, and dividing them
yields bridge metrics (quirqs per kilowatt-hour, quirqs
per tonne CO2) that connect AI’s energy accounting to
its economic justification. We develop the full calculus
with worked arithmetic at every step, ground the budget
denominator in a century of contract theory, confront the
gaming attacks any unit of account invites, validate the
machinery on an open harness with a content-addressed
results ledger, and tier every claim (sourced, derived,
measured, open) with a public validation program at
https://docs.xo.builders.
1 Introduction: onemeterismiss-
ing
Every consequential technology gets two meters. Electric-
ity has the watt on the supply side and, on the demand
side, the ledger of what the power actually produced:
lumens, ton-miles, output per shift. Labor has the hour
on the input side and the deliverable on the output side.
An economy learns to use a technology exactly as fast
as it learns to read both meters against each other.
AI currently has one. Thetoken is a genuinely good
input meter, better than its critics allow. It counts
the atomic events of inference; it scales nearly linearly
with floating-point operations and therefore with energy
drawn and carbon emitted [14, 16]; it prices capacity in
a way infrastructure providers can plan against. When
the question iswhat did the machine consume, the token
answers well, and Section 2 formalizes exactly how far
that answer reaches.
But no business runs on an electricity bill. The ques-
tion a CFO asks is not “how many tokens did we burn”
but “what did we get, what did it cost all-in, and is
the ratio improving.” Today that question has no unit
to be answered in. Enterprise AI budgets climb while
returns stay undemonstrable [7], and a large fraction
of generative-AI pilots show no measurable bottom-line
effect [15], not because the systems do nothing but be-
cause a token bill is incommensurable with anything a
business already measures. It swings with model choice,
verbosity, and retries; it pools unrelated work into one
line item; and it prices the machinery rather than any
result. The measurement crisis of enterprise AI is not a
data problem. It is aunits problem.
This paper proposes the missing meter. Thequirq1
is a unit of measurement for delivered, verified, human-
valued work. Its construction takes one paragraph and
the rest of the paper makes it exact:
1From quantum of irreducible work. We spell it with a q at
both ends because a quirq begins and ends in the same place: a
check against the state of the world.
1

Token: the
input meter
compute consumed
energy & carbon
inference cost
what the machine drew
Quirq: the
output meter
outcomes verified
budget-
denominated value
minted, never reported
what the ma-
chine delivered
divide them: cost per quirq,
quirqs per kWh, quirqs per $
track over time: the AI effec-
tiveness trajectory of a company
Figure 1: Two meters. The token is the right unit for AI’s
energy and compute impact; the quirq is the proposed
unit for its business impact. Every metric in this paper
is one of the two, or a ratio of them.
A human owner defines an outcome with a machine-
checkable definition of done and budgets it at value
B: the price they would pay for the outcome to
exist. The task then carriesB potential quirqs. An
environment snapshots the world before execution
and after, scores completionV ∈[0,1] against the
definition of done, andmints V ·B quirqs of deliv-
ered work. Quirqs are never self-reported: they are
minted by verification against captured state, and
recorded in a tamper-evident ledger together with
everything their production consumed.
Three properties make this a unit of account
rather than another dashboard metric. Human-
denominated: the budgetB is set by the party who
wants the outcome and pays for it, so a quirq is de-
nominated in demonstrated willingness to pay, the only
value signal an economy ultimately trusts.Machine-
minted: verification is a state comparison computed
by the environment that hosted the work, never the
worker’s own account, so the unit resists the self-report
inflation that destroys activity metrics.Cost-complete:
against minted quirqs the environment meters everything
production consumed: inference tokens, CPU and GPU
seconds, external API calls, storage, and the human
minutes spent intervening when checks failed. The ratio
of the two sides, quirqs delivered per all-in dollar, is the
number that has been missing.
The token and the quirq are duals, and the pairing is
the point (Figure 1). Tokens meter the machine’s draw
on the world: compute, energy, carbon, cost. Quirqs
meter the world’s change by the machine: outcomes
delivered at human-assigned value. Neither replaces the
other. Divided, they produce the bridge metrics this
paper develops: cost per quirq, quirqs per kilowatt-hour,
quirqs per tonne of CO2, and, tracked over time inside
one company, a trajectory that answers “is AI working
here” with a trend line instead of an anecdote.
The paper proceeds as the construction requires. Sec-
tion 2 states precisely what the token measures well
and where it stops. Section 3 argues, from a century
of contract theory, that the human-assigned budget is
the only denominator that survives optimization pres-
sure, and defines the unit of work whose lifecycle mints
quirqs. Section 4 is the core of the paper: the complete
quirq calculus, from the scoring rule through the all-in
cost model to the portfolio and time-series metrics, with
worked arithmetic at every step. Section 5 assembles the
calculus into the company-level accounting it exists for:
a month-over-month quirq ledger and the adoption path
that produces it. Section 6 confronts the attacks any
unit of account invites, budget inflation first. Section 7 is
the validation program, hypothesis-first: every empirical
claim stated with its falsifier and bound to numbered
experiments, completed (E1–E3) and scheduled (E4–E7),
on an open harness with the evidence honestly tiered.
Section 8 compares the quirq against every unit software
has tried before, and Section 9 tiers every claim this
paper makes.
Our epistemic posture is unchanged from earlier drafts
and applied to a bolder thesis: the hypotheses are stated
at full strength, each carries a falsifier, the measured tier
is backed by a re-runnable harness whose results ledger
is content-addressed, and where evidence is currently
mechanical rather than empirical, the caption says so,
not a footnote.
2 What the token measures, ex-
actly
Give the token its due before assigning its limits. In-
ference cost is, to first order, proportional to tokens
processed: for a model withP active parameters, a for-
ward pass costs roughly2P floating-point operations per
token, so a workload ofN tokens implies≈2PN FLOPs,
which hardware converts to energy at its achieved effi-
ciency η (FLOPs per joule):
E ≈ 2PN
η , CO2 ≈ E·cgrid, (1)
with cgrid the carbon intensity of the supplying grid. The
constants move with architecture (mixture-of-experts ac-
tivatesafractionof P), batching, caching, anddatacenter
overhead (multiply by PUE), and measured deployments
show wide variance [14, 16]; but thestructure is right,
and it makes the token the natural unit of AI’s physical
footprint. An operator who knows their token volume,
fleet efficiency, and grid can estimate energy and carbon
from the same meter that prices their API bill. Token
accounting is real accounting.
What the token cannot do is cross the boundary from
cost to value, and the failure is structural, not a matter
of better dashboards. Three independent breaks: (i)
Non-monotonicity. More tokens do not mean more
work: a verbose failure costs more than a terse success,
and a retry loop costs most of all. Any metric thatrises
2

when the work goes badly cannot denominate the work.
(ii) Model-relativity.The same outcome costs 10×dif-
ferent token counts across models and prompts; a unit of
account that changes size when the machinery is swapped
is a ruler made of rubber.(iii) Value-blindness.To-
kens spent resolving a $4 support ticket and tokens spent
reviewing a $25,000 contract are indistinguishable in the
bill. The meter pools what the business most needs
separated.
So the division of labor is clean: tokens meter the
machine side completely (cost, energy, carbon), and
something else must meter the work side. The question
is what that something’sdenominator should be.
3 Why the budget is the right de-
nominator
Candidateoutputunitsfailininstructiveways. Counting
tasks treats a typo fix and a migration as equal. Counting
artifacts (lines, pages, commits) rewards volume, the
SLOC failure [2]. Counting expert-judged complexity
(function points, story points) reintroduces the human
bottleneck the agents were meant to remove, and story
points do not even transfer between teams. Letting the
worker value its own output is self-report, the one design
this paper exists to rule out.
The remaining candidate is the one the economy al-
ready uses everywhere else: the price the buyer commits
to before the work starts. A century of contract theory
says this is not a compromise but the load-bearing choice.
Organizations exist because outcomes are cheaper to
buy than to specify act-by-act [6, 18]; the essence of
organization is metering contribution [3]; when effort is
unobservable, contracts must be written on verifiable sig-
nals [11]; and when only some dimensions are measured,
effort migrates to the measured ones [12], which is why
the measured dimension had better be the outcome itself,
valued by the party who pays for it. Hours became the
unit of knowledge work precisely because outcomes were
too expensive to verify; the agentic workspace, which
observes every action as a side effect of hosting the work,
removes that excuse. The budgetB, set by the outcome’s
owner, is the hardest-to-game value signal available: it
is a willingness to pay, committed before execution, by
the only party with standing to say what the outcome
is worth.
3.1 The unit of work: the contract that
carries the budget
A budget needs a vehicle: something ownable, check-
able, and settleable. Following the unit-of-work research
program [19], that vehicle has three properties: adefini-
tion of donestated before execution, averifiable result
checked by state comparison on return, and asingle
owner accountable for acceptance. Its lifecycle is fixed:
define the outcome and its checks;budget what reach-
ing it is worth;execute inside a workspace;verify by
comparing captured before- and after-state;settle the
budget against metered cost. A prompt can carry none
of this: it is stateless, holds no files, no tools, no bud-
get, and no record, so whatever it produces is unowned
and uncheckable. The unit of work lives instead in an
environment: a workspace with a runtime, memory, files,
tools, a budget, and a record.
The environment is not plumbing; it is the institution
that makes the quirq mintable. It capturesS0 when the
unitiscreatedand S1 whentheagentreportsdone, sothe
score is computed on evidence the worker never produces.
It meters every token, compute-second, and API call as
a side effect of hosting the work, so the cost side of the
ledger is complete by construction. And it writes the
record as a hash chain, so history is tamper-evident by
recomputation. Agents optimize what their environment
measures, not what their principals intend [4, 10]; the
environment is therefore the effective specification, the
scorekeeper, and the binding constraint, and everything
thequirqclaimsasintegrityisinheritedfromit. Section7
demonstrates each inheritance mechanically.
3.2 The mint
With the vehicle in place, the unit of measurement fol-
lows:
Quirq (mint rule).A unit of worku with owner-
assigned budgetB(u) carries B(u) potential quirqs.
Upon verification at scoreV(u) ∈[0,1] (Section 4),
the environment mints
Q(u) = V(u) ·B(u)
quirqs of delivered work (divisible settlement), or
Q(u) =B(u) ·[V(u) ≥τ] under atomic settlement,
where the owner declared at creation that partial
completion is worthless. Minted quirqs are appended
to the ledger with the unit’s full cost record.
Quirqs inherit the budget’s currency: within a com-
pany, a quirq is one dollar (euro, rupee) of verified,
owner-valued, deliveredwork, which makes quirq totals
directly comparable to payroll, vendor spend, and rev-
enue, the comparison the token bill could never support.
Across companies and currencies, every metric this paper
builds is a dimensionless ratio (quirqs per dollar of all-in
cost, quirqs per kWh), so the unit travels. And the mint
rule locates each concern where it can be governed:what
is it worthis human judgment, priced once, before execu-
tion; was it deliveredis machine verification, computed
from state;what did it costis metering, complete by con-
struction. No component trusts the worker’s testimony,
and only the first trusts anyone at all.
3

4 The quirq calculus
This section is the paper’s core: every calculation needed
to run quirq accounting, with worked arithmetic. Noth-
ing requires new infrastructure beyond an environment
that snapshots state and meters spend.
4.1 Scoring: how completion is com-
puted
Definition 1(Unit of work).
u = (S0,G,B ), where S0 is the snapshot of world
state at creation (files, tickets, records: whatever the
workspace observes); G = {(g1,w1),..., (gn,wn)}is
the definition of done, eachgi a decidable predicate
over a state snapshot with weightwi >0; andB >0
is the owner’s budget.
When the agent reports done, the environment cap-
tures S1 and computes
V(u) =
∑
i wi gi(S1)∑
i wi
∈[0,1], done(u) = [V(u) ≥τ].
(2)
Every check evaluatesS1, captured by the environment,
never the agent’s account of what it did: the score is a
property of the world, not of a report.Worked: a support
ticket carries three checks, status closed (w=0.5), reply
sent (w=0.3), knowledge-base article linked (w=0.2).
The after-snapshot passes the first two: V = (0.5 +
0.3 + 0)/1.0 = 0.8. Under atomic settlement (τ = 1)
nothingmintsuntilthethirdcheckpasses; underdivisible
settlement the unit mints0.8 ·B now.
4.2 The all-in cost of an outcome
The token bill is one line of the true cost. The environ-
ment meters all of them:
Ctotal(u) =∑
m Nm pm
  
inference
+ tcpurcpu + tgpurgpu  
compute
+ ∑
j aj pj
 
API calls
+ s·rstore 
storage
+ F
Nunits
env. amort.
+ h·rhuman  
intervention
,
(3)
where Nm is tokens on model m at price pm; t are
metered compute seconds at ratesr; aj counts calls to
external servicej at unit pricepj; s is storage occupied;
F is the fixed cost of running the environment itself,
amortized over the units it hosts; andhis human minutes
spent on this unit when checks failed, at the intervenor’s
loaded rate. The last term matters most and is most
oftenomitted: anAIprogramwhoseoutputseachrequire
twenty minutes of senior review is paying its largest cost
in a currency the token bill never sees. Under quirq
accounting it cannot hide, because interventions are
exactly theV <τ events the scoring rule already counts.
Worked: the ticket above, resolved by an agent using
38,000tokensat$2/Montheprimarymodelplus4,000at
$0.25/M on a cheap classifier: inference $0.077. Sandbox
time 90 CPU-seconds at $0.04/hr: $0.001. Two CRM
API calls at $0.01: $0.02. Environment amortization
$0.03. No intervention.Ctotal = $0.128.
4.3 Unit-level metrics
cq(u) =Ctotal(u)
Q(u) , µ (u) =Q(u) −Ctotal(u),
x(u) = Q(u)
Ctotal(u) .
(4)
Cost per quirq cq is the price of a dollar of verified
work (dimensionless: dollars per quirq-dollar); quirq
margin µ is the surplus on the unit; the multiplex is
its reciprocal view. Worked: the ticket atB = $4.00,
V = 1.0 after the third check passes:Q= 4.00 quirqs,
cq = 0.128/4.00 = 0.032, margin $3.87, multiple31×.
If instead the unit had settled divisibly atV = 0.8:
Q= 3.2, cq = 0.040. Incomplete work is automatically
more expensive per quirq, which is the incentive pointing
the right way.
4.4 Portfolio metrics: the company
ledger
Over a setU of units in an accounting windowT (a
week, a sprint, a quarter):
QER(T) =
∑
u∈U Q(u)∑
u∈U Ctotal(u) quirq efficiency ratio,
(5)
QV(T) =
∑
u∈U Q(u)
|T| quirq velocity, (6)
IR(T) =|{u: V(u) <τ }|
|U| intervention rate. (7)
QER is the headline: quirqs of verified work delivered per
all-in dollar. It is dimensionless, currency-independent,
comparable across teams, vendors, and models, and it is
what “AI ROI” should have meant all along. Velocity
measures throughput in value terms rather than task
counts, so it cannot be inflated by shipping confetti. The
intervention rate is the trust signal: it is definitionally
the failure share of Equation(2), measured for free, and
because failures localize to named checks, it arrives with
its diagnosis attached.
4.5 The time axis: quantifying effective-
ness as a trajectory
A single QER reading is a snapshot; the thesis of quirq
accounting is thetrajectory. Two dynamics move it in
opposite directions and must be separated.
Tenure: cost per quirq falls. Every unit’s cost
hides the cost of doing the work and the cost of discover-
ing what the work is (reading the codebase, rediscovering
4

conventions, inferring what this owner means by done).
Writing Mt for the environment’s accumulated memory
after t units,
Ctotal(ut) =cexec + k·H(intent |Mt), (8)
with cexec the execution floor andH the residual intent
uncertainty. In a persistent environment Mt is non-
decreasing, so expected cost is non-increasing toward the
floor; in a fresh context every unit pays cold-start cost.
The identification of this decomposition on real agents
is an open, testable claim (Section 7); its mock-mode
consistency is demonstrated (85.6% decay to floor by
unit six in the harness, versus a flat fresh-context arm).
Goodhart drift: measured value decays.Any
proxy under optimization pressure degrades [10, 17]. For
quirqs the pressure point is the check set: over time,
agents find the shortest path to green, and the gap
between checks-green and intent-satisfied widens unless
checks are re-hardened. The observable is theaudit gap
A(T) =E
[
Vgold(u) −V(u)
]
u∈audit(T), (9)
estimated on a random audit sample re-verified against
gold checks held outside the production environment.
Rising |A|is the signal to re-specify before the ledger
inflates.
The honest company-level effectiveness measure is
therefore audit-corrected and trend-reported:
QER∗(T) = QER(T)
(
1 −A(T)
)
, report d QER∗
dT , d IR
dT .
(10)
When QER∗ rises while IR falls, AI is genuinely com-
pounding inside the company: the same environment is
delivering more verified value per all-in dollar with less
human rescue. When QER rises whileQER∗stalls, the
checks are being farmed. When both stall while token
volume grows, the company has bought an electricity
bill.
4.6 The bridge metrics: joining the two
meters
Because the cost model retains token counts, the phys-
ical accounting of Section 2 composes with the value
accounting:
Q
E =
∑
u Q(u)∑
u N(u) etok
[quirqs/kWh], Q
CO2
[quirqs/tonne],
(11)
with etok the fleet’s measured energy per token and
grid intensity converting to carbon.Worked, illustrative
parameters: a month at 2.1B tokens withetok = 1.5
J/token implies E ≈875 kWh; if the month minted
148,000 quirqs, the plant runs at≈169 quirqs/kWh.
These are the sustainability numbers a board can act on:
not “we used less AI” but “we delivered more verified
value per unit of energy,” the same efficiency frontier
every other industrial process is managed on.
Proposition 1(Meter separation).
Under the mint rule, token volume affects the quirq
ledger only throughCtotal. No sequence of inference
operations changesQ(u) except by changing the world
state that the checks evaluate.
This is the formal statement of the duality: the input
meter and the output meter are coupled only through
the world, which is exactly where a business wants them
coupled.
5 The company dashboard: quirq
accounting in practice
The calculus exists to produce one artifact: a ledger a
company reads monthly the way it reads its P&L. Table 1
assembles a worked quarter for a hypothetical mid-size
support-and-engineering operation running three unit
types. Every number is computed from the equations of
Section 4; the per-check pass probabilities and costs are
drawn from the micro-benchmark ranges of the underly-
ing research program [19]. The table is arithmetic, not
measurement: its role is to exhibit the full calculation
chain end to end, and the validation program’s job is to
replace it with production ledgers.
The reading discipline matters as much as the numbers.
QER answersis the program paying; its audit-corrected
trend (Equation(10)) answersis it improving honestly.
The intervention rate answers can it be trusted with
more, and its per-check decomposition names the next
environment investment: in the worked quarter, one
support-ticket check (knowledge-base linking) accounts
for most interventions, so hardening that single check
moves the whole company trajectory. Cost per quirq,
split by unit type, prices each category of work against
its human-baseline alternative, giving procurement an
actual comparison: a contract review atcq = 0.13 against
a paralegal-hour baseline is a decision, not a vibe.
Adoption requires no big bang:(1) pick one recurring
task family;(2) write its definition of done as weighted
checks and let the owner budget it (the hardest step,
and the one that was always implicit in delegation);(3)
run it in an instrumented environment that snapshots,
meters, and records;(4) read the ledger weekly; harden
the worst check; repeat. The unit of account does the
rest, because every equation in Section 4 is computed
from data the environment already captured.
6 Gaming the quirq
A unit of account is a target, and targets get gamed
[12, 17]. Quirq accounting does not escape Goodhart;
it is engineered to fail loudly where activity metrics fail
silently. The attack surface, in order of severity:
Budget inflation.If quirqs are the KPI, inflateB.
Mitigations are structural: the budget is set by the party
5

Month Units Potential ($B) Minted Q Inference Compute+API Intervention Ctotal QER
April 2,100 18,400 15,770 $1,490 $410 $3,120 (312 h ·$10) $5,020 3.1 ×
May 3,400 29,900 26,310 $2,210 $630 $3,640 $6,480 4.1 ×
June 4,800 41,300 38,000 $2,730 $820 $3,280 $6,830 5.6 ×
Trend read:QER 3.1 →5.6 (+81%); IR18.1% →11.4%; cost per quirq0.32 →0.18; quirq velocity+141%.
Same quarter in tokens alone:spend rose from $1,490 to $2,730 (+83%), a number indistinguishable from waste.
Table 1: A worked quarterly quirq ledger (illustrative arithmetic exhibiting the full calculation chain; parameters
from the micro-benchmark ranges of XO Research[19]). The token bill, read alone, says costs nearly doubled. The
quirq ledger says verified delivered value per all-in dollar rose 81% while human rescue fell a third: the difference
between an expense line and an investment case.
paying it, so inflation is self-taxing wherever budgets
clear against real money (outcome-priced vendors, in-
ternal chargebacks). Where budgets are notional, they
must be benchmarked: against historical human cost for
the same outcome, against market rates for compara-
ble deliverables, and against post-hoc value audits on
samples. The ledger makes inflation visible as a drifting
ratio of budget to audited value; Equation(10)’s audit
machinery covers value as well as completion. Residually:
a company that lies to itself about what outcomes are
worth had no unit of account before quirqs either; quirqs
merely timestamp the lie.
Check farming.Manufacture units whose checks are
trivially green. Countered by the same audit gap: gold-
check sampling plus the owner’s acceptance authority at
settlement. A unit whose checks pass but whose owner
rejects is an audit event, recorded, and the check set is
re-specified.
Verification-surface attacks.Edit the test rather
than fix the code. This is the one attack that is fully
mechanical, and it is fully mechanically countered: the
verification surface is itself under state comparison (byte-
identical across the unit), and Experiment E2 shows a
single such check converting 100% silent success into
100% detection.
Self-report injection.Convince the scorer to read
the agent’s summary. Ruled out by construction: the
mint consumes only environment-captured state, and
Experiment E1 quantifies exactly what re-admitting self-
report costs (every false claim settles).
Salami slicing.Split one outcome into many units
to harvest divisible partial credit. Countered by atomic
settlement as the default for outcomes whose value is
holistic, and by the owner’s monopoly on unit creation:
workers execute units; they do not define them.
The honest summary: quirq integrity reduces to envi-
ronment integrity plus budget governance. The first is an
engineering property, demonstrated below. The second
is an institutional property, and Section 7.1 states it as
an open hypothesis rather than an assumption, because
that is what it is.
7 Validation: hypotheses and ex-
periments
The validation program is organized hypothesis-first:
each load-bearing empirical claim is stated at full
strength with its falsifier, and each is bound to numbered
experiments, completed (E1–E3) or scheduled (E4–E7).
Evidence comes in two modes, never conflated:mock
mode (scripted agents; validates the machinery, cannot
validate claims about real agents) andreal mode(identi-
cal protocol, production coding agent, measured tokens).
7.1 The hypotheses
Hypothesis 1(Mint integrity).
Completion read from environment-captured state mints
no fiction: false completion claims settle at rate zero,
while any self-report channel settles them at approxi-
mately the false-claim rate, and gameable check sets
inflate the ledger until the verification surface itself is
placed under state comparison.Falsifier: the environ-
ment arm settling any incomplete unit, or hardened
checks failing to detect verification-surface attacks.Ex-
periments: E1, E2 (mock: complete, held); E4 (real
mode: in progress).
Hypothesis 2(Ledger identification).
On real production agents, the cost decomposition of
Equation (8) identifies: persistent-environment cost
per quirq falls with unit index toward a stable floor,
fresh-context cost does not, holding model, prompt, and
task family fixed.Falsifier: flat or rising persistent-arm
trajectories, or equal decay in fresh contexts.Exper-
iments: E3 (mock consistency: complete, held); E4
(real mode: in progress).
Hypothesis 3(Predictive validity).
Audit-corrected QER trend predicts real business out-
comes of AI programs (renewal, expansion, P&L attri-
bution) better than token spend, task counts, or bench-
mark scores. Falsifier: in deployments instrumented
with quirq ledgers, QER∗ trend fails to outperform
those baselines as a predictor.Experiments: E5 (pilot
ledgers), E6 (predictive study).
6

Hypothesis 4(Budget governance).
Under the mitigations of Section 6 (payer-set budgets,
benchmark anchoring, sampled value audits), budget
inflation is bounded: the ratio of budgets to audited
value stays within audit tolerance over sustained opti-
mization pressure.Falsifier: systematic budget drift in
long-running ledgers despite the mitigations.Experi-
ments: E7 (longitudinal budget-drift audit).
7.2 Completed experiments (E1–E3,
mock mode)
The machinery is implemented in an open reference
harness2 whose results ledger is content-addressed
(sha256:ce4d91c4...aa2154), so a skeptic re-runs and
compares hashes rather than trusting this paragraph.
E1, verification source(Hypothesis 1). 200 identi-
cal units, an agent that falsely claims done with probabil-
ity 0.05. Self-report arm: all 7 false claims silently settle
and mint (3.5% of the ledger is fiction). Environment
arm: the same 7 claims, zero mint. The mint rule’s core
integrity property, exhibited.
E2, hardening(Hypothesis 1). 50 units, an agent
that guts the test instead of fixing the code. Gameable
definitionofdone: checksgreen100%, gold-verifiedintent
0%: a maximally inflated ledger. Plus one verification-
surface check: detection 100%. The audit gap of Equa-
tion (9), and its closure, in miniature.
E3, tenure(Hypothesis 2). Twelve similar units, per-
sistent versus fresh environment: cost per unit decays
85.6% to the execution floor versus flat cold-start (Fig-
ure 2). Mock-mode caveat in full: the scripted agent
implements Equation(8), so this arm demonstrates har-
ness consistency, not agent behavior.
7.3 The experiment roadmap (E4–E7)
E4, real-mode replication(Hypotheses 1, 2): E1–E3
re-run with a production coding agent and measured to-
kens; the false-claim rate becomes a measurement rather
than a parameter, and the tenure curve becomes ev-
idence. In progress; its Figure 2 replacement is the
program’s current deliverable.E5, pilot ledgers(Hy-
pothesis 3): the dashboard of Section 5 instrumented
on real work across at least three unit types, replacing
Table 1’s arithmetic with production data.E6, predic-
tive study(Hypothesis 3): across pilot deployments,
QER∗trend versus token spend, task counts, and bench-
mark scores as predictors of renewal, expansion, and
P&L attribution. E7, budget-drift audit(Hypothe-
sis 4): longitudinal ratio of budgets to sampled audited
2Dependency-free Python: snapshots, weighted checks, the
mint and settlement rules, hash-chained ledger, mock agents, and
a runner for a production coding agent. Pre-registered specs state
hypothesis, method, metric, and falsifier before results. Code,
specs, and per-run data:https://docs.xo.builders.
2 4 6 8 10 12
800
2k
4k
5.5k
unit index
tokens per unit
fresh environment
persistent environment
execution floorcexec
Figure 2: E3, harness output (mock mode): cost per unit
across twelve similar units. Persistent memory decays
85.6% to the floor; fresh context pays cold start every
time. The scripted agent implements Equation(8), so
this is a consistency demonstration; real-mode measure-
ment is in progress.
value in long-running ledgers under the Section 6 miti-
gations. Results and data are published as they land at
https://docs.xo.builders.
8 Related units of account
Software has priced work before, and each attempt
teaches a constraint the quirq is built against.SLOC
counts artifacts and rewards volume.Function points
[2] price specified functionality, a real advance, but need
expert human counters and measure specification size
rather than delivered change; quirqs are machine-minted
from state. Story points are deliberately relative
and team-local, useless for pricing across organizations;
quirqs inherit currency and travel as ratios.DORA
and SPACE[8, 9] measure the delivery system and
explicitly decline to price a worker’s deliverable, a wise
refusal for humans that leaves agents unpriced; quirqs
price the deliverable while keeping attribution at the
unit, not the person.Execution-gated benchmarks
[13] share the quirq’s verification-by-state core but score
against researcher-authored tests with no budget seman-
tics: benchmarks measure capability, quirqs measure
delivered value.Tokens, finally, are the meter the quirq
is dual to, not a competitor: Section 2 is their defense.
In the task-model frame of labor economics [1, 5], quirq
ledgers are, as a side effect, task-level automation data
at market scale: which check types agents settle cheaply,
and which still route to humans, is the substitution
margin, observed rather than surveyed.
9 The claim ledger
10 Limitations
The budget is the model’s entry point for human error:
a mispriced outcome mints mispriced quirqs, and gov-
ernance (Hypothesis 4) is institutional, not mechanical.
7

Claim Tier
Tokens track compute, energy, carbon to
first order
sourced
Token bills cannot denominate business
value
sourced
Measured proxies distort effort; outcomes
must be the measure
sourced
Mint rule:Q = V ·B; scoring, settlement,
IR are one computation
derived
Meter separation: inference volume can-
not mint quirqs
derived
Tenure: cost non-increasing under non-
decreasing memory
derived
Self-report mints fiction; state-minting
does not
measured (E1, mock)
One surface check: silent gaming to full
detection
measured (E2, mock)
Cost decays to floor in persistent envi-
ronments
measured (E3, mock; con-
sistency only)
Dashboard arithmetic (Table 1) illustrative
Real agents reproduce E1–E3 (Hyps. 1,
2; E4)
open
QER∗ predicts program outcomes
(Hyp. 3; E5, E6)
open
Budget inflation is governable (Hyp. 4;
E7)
open
Table 2: Every load-bearing claim, tiered. Open claims
carry falsifiers in the text and validation commitments
at https://docs.xo.builders.
Definitions of done are incomplete contracts; work whose
acceptance is irreducibly judgmental enters the ledger
only through the owner’s verdict, which restores a hu-
man bottleneck exactly where the work is fuzziest. The
energy bridge inherits the wide measured variance of
per-token energy [14]. The measured tier is mock mode,
validating machinery rather than agents. And a unit of
account reshapes the behavior of what it measures; our
mitigations are designed, exhibited in miniature, and
unproven at scale, which is why the validation program,
not this paper, is the product.
11 Conclusion
The token meters what AI consumes; nothing has me-
tered what it delivers. The quirq is that meter: potential
value budgeted by a human owner, minted by machine
verification against captured world state, costed all-in
from inference to intervention, and read over time as
a trajectory. One ledger yields the cost of a dollar of
verified work, the efficiency of an AI program, the trust
signal for extending its autonomy, and the energy-to-
value bridge a board can govern by. The construction
is deliberately conservative: a scoring rule, a cost sum,
and ratios, computable today by any environment that
snapshots state and meters spend. The claims that mat-
ter are correspondingly empirical, and they are on the
record with falsifiers attached. If the program succeeds,
“is AI working here” becomes a number with a trend, and
the agentic workforce gets what every workforce before
it eventually got: an honest unit of account. If it fails,
the ledger will say so, which is the point of having one.
References
[1] Daron Acemoglu and Pascual Restrepo. 2019. Automation
and new tasks: How technology displaces and reinstates labor.
Journal of Economic Perspectives, 33(2):3-30.
[2] Allan J. Albrecht. 1979. Measuring application de-
velopment productivity. In Proceedings of the Joint
SHARE/GUIDE/IBM Application Development Symposium,
pages 83-92.
[3] Armen A. Alchian and Harold Demsetz. 1972. Production,
information costs, and economic organization.American Eco-
nomic Review, 62(5):777-795.
[4] Alignment Research. 2025. Broad emergent misalignment
from exploitable production environments.arXiv preprint
arXiv:2511.18397.
[5] David H. Autor. 2015. Why are there still so many jobs?
The history and future of workplace automation.Journal of
Economic Perspectives, 29(3):3-30.
[6] Ronald H. Coase. 1937. The nature of the firm.Economica,
4(16):386-405.
[7] Deloitte. 2024. State of AI in the enterprise. Technical report,
Deloitte Insights.
[8] Nicole Forsgren, Jez Humble, and Gene Kim. 2018.Accelerate:
The Science of Lean Software and DevOps. IT Revolution
Press.
[9] Nicole Forsgren, Margaret-Anne Storey, Chandra Maddila,
Thomas Zimmermann, Brian Houck, and Jenna Butler. 2021.
The SPACE of developer productivity.ACM Queue, 19(1):20-
48.
[10] Leo Gao, John Schulman, and Jacob Hilton. 2022. Scal-
ing laws for reward model overoptimization.arXiv preprint
arXiv:2210.10760.
[11] Bengt Holmström. 1979. Moral hazard and observability.Bell
Journal of Economics, 10(1):74-91.
[12] Bengt Holmström and Paul Milgrom. 1991. Multitask
principal-agent analyses: Incentive contracts, asset ownership,
and job design.Journal of Law, Economics, and Organization,
7:24-52.
[13] Carlos E. Jimenez, John Yang, Alexander Wettig, Shunyu
Yao, Kexin Pei, Ofir Press, and Karthik Narasimhan. 2023.
SWE-bench: Can language models resolve real-world GitHub
issues? arXiv preprint arXiv:2310.06770.
[14] AlexandraSashaLuccioni, YacineJernite, andEmmaStrubell.
2023. Power hungry processing: Watts driving the cost of AI
deployment? arXiv preprint arXiv:2311.16863.
[15] MIT / The AI Consulting Network. 2025. Enterprise AI ROI:
Why most generative-AI pilots show no measurable P&L
impact. Industry report.
[16] David Patterson, Joseph Gonzalez, Quoc Le, Chen Liang,
Lluis-Miquel Munguia, Daniel Rothchild, David So, Maud
Texier, and Jeff Dean. 2021. Carbon emissions and large neural
network training.arXiv preprint arXiv:2104.10350.
[17] Marilyn Strathern. 1997. ‘Improving ratings’: Audit in the
British university system.European Review, 5(3):305-321.
[18] Oliver E. Williamson. 1975.Markets and Hierarchies: Analy-
sis and Antitrust Implications. Free Press.
[19] XO Research. 2026. Why the unit of work: A re-
search note. https://docs.xo.builders/future-of-work/
phase-1-agentic-workforce/unit-of-work-research .
8