Abstract

Accounting is the oldest and most successful symbolic technology human beings have built for rendering life legible, and it is for that reason the strongest possible test case for a more general question: where does mathematical and symbolic quantification actually fit the world it claims to describe, and where does it merely resemble fit while producing something else? This article takes double-entry bookkeeping, invented for Venetian and Genoese merchants and codified by Luca Pacioli in 1494, as its sustained example. Double-entry succeeded spectacularly in its original domain because that domain, completed, discrete, already-agreed transactions, was already a nonrecursive symbolic fact before any ledger touched it. The article then traces what happened as the same technology, buoyed by that early success, was extended into domains transactions never prepared it for: the valuation of goodwill and other intangibles, the allocation of overhead inside automated factories, the pricing of illiquid financial instruments under fair value rules, and the audit-driven governance of universities, hospitals and public services. In each extension, the same pattern recurs. Formal consistency, the fact that debits still equal credits, is mistaken for a guarantee of external fit, when it is nothing of the kind. The article draws on the true and fair view, a legal override built into accounting itself, as evidence that even accounting's own institutions know that judgement, not further rule-following, must have the final word. Accounting, examined this closely, turns out to be simultaneously the best argument for quantification and the clearest diagnosis of its limits.

1. Introduction: Accounting as the Strongest Case

Any argument that mathematics and quantification cannot fully capture lived reality invites an obvious rejoinder: choose a harder case. Do not reach for love, or health, or meaning, domains everyone already suspects resist a number. Reach instead for the discipline that has spent five centuries perfecting exactly the kind of symbolic rendering such arguments doubt, and see whether it holds up. Accounting is that discipline. No other quantitative technology has been more successful, more durable, or more consequential for how economic life is actually organised. Double-entry bookkeeping outlived the empires that used it, survived the transition from parchment to paper to spreadsheet to database, and remains, after five hundred years, essentially unchanged in its basic logic: every transaction recorded twice, once as a debit and once as a credit, so that the books must always balance.

This article takes accounting seriously as the strongest case for quantification precisely because it wants to show something more interesting than the usual objection that some things resist measurement. Accounting demonstrates, within a single technology and often within a single set of financial statements, both an extraordinarily durable fit between symbol and reality and a systematic, repeated, well documented tendency to lose that fit the moment the technology is extended beyond the domain it was built for. Watching where accounting succeeds and where it fails, often within the same balance sheet, teaches more about the general question of symbolic fit than any argument conducted at the level of abstract principle.

The argument proceeds historically. It begins with double-entry bookkeeping's original domain, the recording of completed merchant transactions, and asks why that domain proved so hospitable to symbolisation. It then follows accounting's own historical extension beyond that domain: into the valuation of the firm as a whole, into the internal allocation of cost inside the factory, into the pricing of financial instruments that never trade in a liquid market, and into the governance of institutions, universities, hospitals, public services, that never kept accounts in the first place. At each stage, the same question recurs. Does the symbol descend back into the mesocosm, the lived, coordinated world it claims to describe, or does it detach from that world while retaining, and even amplifying, its institutional authority?

2. Pacioli's Gift: Why Double-Entry Fit So Well

Double-entry bookkeeping did not originate with Luca Pacioli. Venetian and Genoese merchants had been using versions of it for at least a century before Pacioli, a Franciscan friar and mathematician, codified the method in 1494 within a broader mathematical treatise, the Summa de Arithmetica, Geometria, Proportioni et Proportionalita (Pacioli, 1494). What Pacioli actually gave the world was not a discovery about commerce but a symbolic technology of remarkable elegance: every transaction entered twice, once as a debit to one account and once as a credit to another, so that the sum of all debits must always equal the sum of all credits across the entire ledger. If the books do not balance, an error has certainly occurred somewhere, even if nobody yet knows where. This self-checking property, arithmetic consistency built directly into the recording method itself, is part of why double-entry proved so durable and so exportable, travelling from Venetian countinghouses to every commercial economy that has existed since.

But the deeper reason double-entry succeeded runs beneath its elegant mechanics, and it is the reason this article opens with the method rather than with any later development. A merchant's transaction, by the time it reaches the ledger, has already been through a great deal of the work that quantification usually has to do on its own. A sale has already been agreed by two parties. A price has already been settled in a shared currency. Goods have already changed hands, or a contract has already fixed the terms under which they will. The ledger does not have to translate a raw, unformed piece of lived coordination into symbolic form. It only has to record a distinction that has already been drawn, symbolically, by the transaction itself. A completed sale is already a discrete, dated, quantified, mutually agreed social fact before an accountant ever opens a ledger to enter it.

This is why double-entry bookkeeping achieved what centuries of subsequent quantitative social science have struggled to match: near-total fidelity between symbol and referent, within its proper domain. The domain in question, cash received, cash paid, goods delivered, debts incurred, was already nonrecursive by the time accounting arrived, in the sense that matters here: nothing about how the transaction is subsequently described or recorded alters what the transaction was. A completed transaction does not un-happen, does not reconsider itself, does not respond differently depending on how it is later entered into a ledger. It sits in the past, fixed, exactly the kind of counterpart that a symbolic system built from discrete, durable marks is equipped to represent without loss. Pacioli did not invent a technology powerful enough to capture any economic reality whatsoever; he formalised one exquisitely well suited to a narrower, already-hospitable domain: the recording of completed exchange.

3. What the Ledger Actually Records

It is worth being precise about what falls inside this hospitable domain and what, from the very outset, sits at its edge, because the rest of this article depends on the distinction. Cash is the cleanest case of all. A sum of money received or paid is unambiguous, dated, and countable, and its recording introduces essentially no distortion between the transaction and its symbolic trace. Accounts receivable and accounts payable, amounts owed to and by the business, sit almost as comfortably within the domain, since they too rest on discrete, dated, contractually specified obligations rather than on anything requiring interpretation. Inventory, at least when counted physically rather than estimated, belongs here too: a warehouse either holds four hundred units of a given item or it does not, and counting them introduces no meaningful gap between symbol and referent.

Tangible fixed assets, land, buildings, machinery, present a slightly more complicated case, but one that remains, for the most part, within the hospitable zone. A factory building can be bought for a specific, recorded price. What happens afterward, how its value changes as it ages, as markets shift, as the surrounding area develops, is where accounting first has to make a choice about which convention to follow, historical cost or something closer to current value, and different jurisdictions and different accounting traditions have made that choice differently across the twentieth century. But even here, the underlying fact the ledger is trying to represent, a specific building, in a specific location, purchased on a specific date for a specific sum, remains a discrete, bounded, already-symbolic fact rather than an ongoing recursive process resisting translation.

What all of these hospitable cases share is that the mesocosmic work, the work of turning a lived, coordinated fact into something countable, has already been done before accounting ever touches it. A sale, a debt, a physical count, a purchase price, are already deposits rather than living processes: nonrecursive by the time the accountant records them, requiring no further translation from felt, embodied, relational reality into symbolic form. Accounting's original genius was recognising this and building a recording system of extraordinary internal discipline around exactly this class of fact. Its later history, as the following sections trace, is the story of what happened once that genius was mistaken for a general licence to render anything economic in symbolic form.

4. The Overreach Begins: From Stewardship to Valuation

For roughly four centuries, accounting remained substantially what Pacioli had described: a stewardship technology, a discipline for keeping an accurate record of what a merchant, and later a company, owned, owed, and had transacted. Its purpose was retrospective and custodial. It told an owner, or a set of investors, whether the people managing their money had handled it honestly and kept accurate books. This stewardship function fits comfortably within the hospitable domain described in the previous section, because the questions stewardship accounting exists to answer, what was received, what was paid, what remains, are exactly the questions double-entry was built to answer.

The twentieth century shifted the discipline's ambitions considerably, and the shift is visible in the language accounting standard-setters themselves came to use. Where earlier accounting frameworks spoke of stewardship, twentieth-century conceptual frameworks, developed by bodies such as the Financial Accounting Standards Board in the United States and later the International Accounting Standards Board, increasingly described the objective of financial reporting as providing information useful for economic decision-making, a considerably more ambitious mandate. Financial statements were no longer meant merely to record what had happened. They were meant to help investors, lenders, and other users predict what a business was worth and what it would become. This is a significant ontological escalation, even though it rarely gets named as one. Recording a completed transaction and estimating the future value of an ongoing enterprise are not the same kind of activity, however similar the resulting numbers look once printed side by side in an annual report.

The escalation had an obvious institutional driver. As firms grew larger, as ownership separated from management, and as capital markets expanded, investors needed more than a record of past transactions; they needed some symbolic handle on a firm's likely future performance, and accounting was the only discipline positioned, institutionally and historically, to supply it. But supplying it required accounting to move from recording discrete, already-completed facts to estimating the value of things that had not yet happened, or that had no discrete transactional history at all: a brand's reputation, a customer relationship, a research pipeline, an acquired company's competitive position. None of these are nonrecursive deposits waiting to be counted. They are ongoing, contested, interrecursively sustained processes, the kind of thing this article's opening sections identified as sitting outside accounting's original, hospitable domain. The rest of this article follows what happened when a technology built for the first kind of fact was asked to produce numbers for the second.

Brand valuation and human capital reporting extend the same ambition further still. More recent reporting standards have begun asking companies to disclose metrics such as employee turnover, training investment, and workforce composition alongside traditional financial figures, treating an organisation's people as a form of capital comparable, at least in the framing the reporting requires, to its plant and equipment. The framing is revealing. A workforce is not a stock of a homogeneous, interchangeable resource the way a warehouse of raw material is. It is thousands of ongoing, interrecursive relationships between individuals, teams, and an organisation's shifting culture, each one continuously renegotiated rather than sitting in inventory awaiting use. Reporting it as capital does not describe this reality. It substitutes for it a countable proxy that can be compared across companies and years, exactly the manoeuvre goodwill performs at the level of the whole firm, examined in detail in the section that follows.

5. Goodwill: A Residual Wearing a Number's Clothes

No single line item on a modern balance sheet illustrates the overreach more clearly than goodwill, and it is worth spending real time with it, because its very definition contains an admission the rest of accounting rarely states so plainly. When one company acquires another, accounting standards require the acquirer to identify and value every asset it can, the target's cash, inventory, equipment, patents, and customer contracts, each valued individually according to established methods. Goodwill is then defined as whatever remains: the difference between the total price paid for the company and the sum of the fair values of everything that could be individually identified and valued. Goodwill is not so much measured as left over, the residue that remains once measurement has done everything it can.

This residual character is not a technical footnote. It is the entire point, and it deserves to be read as exactly what it is: an accounting standard formally acknowledging that a company's value regularly exceeds the sum of what can be itemised, without being willing to say what the excess actually consists of. In practice, goodwill captures things like an acquired workforce's accumulated skill and working relationships, a brand's accumulated trust among customers, an organisation's capacity for coordinated action, and the countless small, unrepeatable, historically sedimented advantages that make one going concern worth more than an equivalent pile of its individually listed parts. These are exactly the kind of ongoing, interrecursively sustained coordinations this article has already identified as resistant to symbolic capture. Goodwill is what accounting calls the part of a company it cannot itemise, forced, by the double-entry requirement that the books must balance, into appearing on the balance sheet as a single number anyway.

Having produced this number under duress, accounting standards have then had to decide what to do with it, and the history of that decision is instructive. For much of the twentieth century, goodwill was amortised, written down by a fixed amount each year over an assumed useful life, in exactly the manner applied to a machine wearing out through use. This treatment at least had the modest virtue of admitting, through the act of steady write-down, that the number was a rough approximation rather than a precise fact. In 2001, United States accounting standards abandoned amortisation in favour of an impairment-only approach, later adopted internationally through IFRS 3, under which goodwill is left unchanged on the balance sheet indefinitely unless an annual test concludes that it has become impaired. That test typically relies on discounted cash flow projections built from assumptions about future growth rates, discount rates, and terminal values, each one a judgement call dressed in the borrowed authority of a spreadsheet. The result is a figure that looks, on the page, exactly as precise as the cash balance sitting two lines above it, though it was arrived at through a completely different kind of reasoning, one considerably closer to informed narrative than to counting.

6. Depreciation, Provisions, and the True and Fair View

Goodwill is the most dramatic instance of judgement smuggled into apparent precision, but it is far from the only one, and two more ordinary examples show how pervasive the pattern actually is, even within accounting's supposedly hospitable core. Depreciation, the systematic allocation of a tangible asset's cost across its useful life, requires an accountant to estimate, in advance, how many years a machine will remain productive, what it will be worth when finally retired, and which mathematical pattern, straight-line, reducing balance, units of production, best approximates how its value actually declines. None of these estimates is dictated by the asset itself. A machine does not announce its own useful life. The convention chosen shapes reported profit substantially, year by year, and different companies, facing identical machines, routinely choose different conventions and arrive at different reported figures, all of them equally compliant with accounting standards.

Provisions extend the same pattern further still. A company expecting some customers to default on their debts, some products to be returned under warranty, or some pending lawsuits to result in a payout, is required to estimate, today, a single number representing an uncertain future cost, and to enter that estimate into the current period's accounts as though it were as solid as a cash balance. The provision is not wrong to exist; ignoring foreseeable future costs would badly distort a company's reported position in the other direction. But the provision is, irreducibly, an estimate wearing a number's clothing, precise to the decimal point on the page while resting on a judgement about a future that has not yet arrived and might arrive quite differently.

Accounting law itself, remarkably, contains an explicit acknowledgment of exactly this problem, and it deserves to be treated as one of the discipline's most theoretically interesting features. Company law across most jurisdictions that follow International Financial Reporting Standards requires financial statements to give a true and fair view of a company's financial position, and, crucially, provides that where strict compliance with a specific accounting standard would not achieve a true and fair view, departure from that standard is not merely permitted but required. This override clause is a formal, legally binding acknowledgment that rule-following and truthful representation can come apart, and that when they do, judgement about whether the resulting picture actually matches the underlying reality must take precedence over mechanical compliance with the rule that produced it. No algorithm determines when the override applies. No further rule specifies the conditions under which the true and fair view requirement should trump a specific standard. The determination is left, deliberately, to the professional judgement of directors and auditors, who are asked, in effect, whether the accounts, read as a whole, feel right against the reality they are meant to represent. Accounting, at its own foundations, has built in an escape hatch that looks remarkably like a mesocosmic disclosure clause: a place where felt misalignment between symbol and reality is permitted, by law, to override the symbol.

7. Relevance Lost: Management Accounting Leaves the Counting House

Accounting's overreach was not confined to the valuation of the firm as a whole. It also travelled inward, into the internal management of the factory floor, and the history of management accounting supplies one of the best documented case studies of what happens when a costing system, developed for one production environment, is applied unmodified to a very different one.

Cost accounting techniques developed across the nineteenth century, particularly within American railroads and textile mills, in an environment where direct labour constituted the overwhelming majority of production cost and where allocating shared overhead by labour hours therefore produced a reasonably accurate picture of what each product actually cost to make. These techniques calcified, over subsequent decades, into standard costing systems that continued to allocate overhead by direct labour hours long after the underlying production environment had changed beyond recognition. In their influential historical study Relevance Lost, the accounting scholars H. Thomas Johnson and Robert Kaplan traced how management accounting practice, having achieved genuine early success, stopped developing to match the manufacturing processes it was meant to inform, remaining substantially frozen at a level of sophistication reached by around the 1920s even as production technology moved decisively beyond it (Johnson & Kaplan, 1987).

The distortion this produces in a modern, highly automated factory is considerable and well documented. Where direct labour might account for only a small fraction of total production cost, and machine time, engineering overhead, and materials handling account for the rest, allocating overhead by labour hours systematically misprices products, making labour-intensive product lines appear artificially expensive and highly automated product lines appear artificially cheap, regardless of what either actually costs to produce. Managers making pricing and product-mix decisions on the basis of these distorted costs can be led, with complete confidence in numbers that balance perfectly on the page, toward decisions that actively destroy value: discontinuing genuinely profitable product lines because a distorted allocation makes them look unprofitable, or expanding lines that are quietly losing money because the same distortion makes them look attractive.

A stylised example makes the distortion vivid. Imagine two product lines manufactured in the same automated plant, one requiring twenty hours of direct labour per unit and minimal machine time, the other requiring two hours of direct labour and extensive computer-controlled machining. A standard costing system that allocates the plant's substantial overhead, machine depreciation, energy, engineering support, by direct labour hours will load the vast majority of that overhead onto the labour-intensive product, because that is the variable the system happens to be watching, even though the machine-intensive product may be the one actually consuming most of the overhead resources being allocated. The labour-intensive product then appears far less profitable than it is, and the machine-intensive product appears far more profitable than it is, and neither appearance has anything to do with which product actually earns the company more money.

What makes this case so useful for the present argument is precisely that nothing has gone wrong with the arithmetic. Every calculation is performed correctly according to the standard costing method in use. The books still balance. The failure occurs one level up, in the fit between the categories the costing system uses, direct labour hours as a proxy for overhead consumption, and the actual production process those categories are meant to describe. A costing convention that fit its original environment extremely well, allocation by labour hours in a labour-intensive plant, was carried forward unexamined into an environment it no longer fit, and the resulting numbers, though internally flawless, systematically misrepresented the reality they claimed to report.

8. Goodhart's Law and the Performative Ledger

A further complication enters once accounting numbers stop merely describing an organisation and start being used to manage it, and this complication has its own well established name. The economist Charles Goodhart observed, in the context of monetary policy targets, that any measure adopted as a policy target tends to lose its statistical relationship to the underlying phenomenon it was chosen to track, precisely because people begin optimising toward the measure itself rather than toward the reality it was meant to represent (Goodhart, 1975). The anthropologist Marilyn Strathern later generalised the observation into the widely quoted form now attached to Goodhart's name: when a measure becomes a target, it ceases to be a good measure (Strathern, 1997).

Accounting numbers are especially vulnerable to this dynamic, because they are simultaneously descriptive and consequential in a way few other statistics are. A reported earnings figure does not merely describe a company's performance; it determines executive bonuses, share prices, credit ratings, and continued access to capital. Once a number carries this much weight, the incentive to manage the number directly, rather than to manage the underlying business the number is supposed to describe, becomes considerable, and accounting history supplies no shortage of cases where that incentive was acted upon.

Enron's collapse in 2001 remains the paradigmatic case (McLean & Elkind, 2003). The company used mark-to-market accounting, ordinarily reserved for assets with observable market prices, to recognise the entire estimated future value of long-term energy contracts as current profit at the moment a contract was signed, long before any of that value had actually been realised, and used a web of off-balance-sheet special purpose entities to keep mounting debt and losses out of the consolidated accounts investors could see. WorldCom, collapsing the following year, achieved a similar effect through a cruder mechanism, reclassifying billions of dollars in ordinary operating expenses as capital expenditure, spreading costs that should have reduced current profit across many future years instead, a fraud eventually uncovered by the company's own internal auditor, Cynthia Cooper, working after hours with a small team to trace the discrepancy herself (Cooper, 2008).

What both cases demonstrate, and what makes them relevant to this article's argument rather than simply scandalous, is that the books balanced throughout. Debits equalled credits at Enron and at WorldCom on every day of the fraud, exactly as double-entry requires. Formal internal consistency, the fact that a ledger's own arithmetic checks out, is a property internal to the symbolic system itself rather than any guarantee of what lies beyond it, and it was perfectly maintained even as the numbers ceased to bear any honest relationship to the underlying business. A ledger can be arithmetically flawless and referentially empty at the same time, and nothing about the arithmetic itself will ever disclose the difference.

9. Fair Value and the Crisis That Followed

The debate between historical cost and fair value accounting is often presented as a technical dispute between competing conventions, but it is better understood as two different strategies for managing the same underlying problem: an asset's value is not a fixed, countable fact the way a completed cash transaction is, but a claim about future benefit that can only ever be estimated. Historical cost accounting sidesteps the estimation problem by declining to update the number at all, reporting an asset at whatever it cost when purchased regardless of how circumstances have since changed, which produces a stable but increasingly stale figure the longer an asset is held. Fair value accounting solves the same problem differently, by importing a number generated somewhere else, a market price, on the theory that a market made up of many independent, informed participants will produce a more accurate estimate than any single accountant working alone. Both strategies are attempts to compensate for the same underlying fact: that an asset's value, unlike a completed transaction, is not already a discrete, symbolic fact waiting to be recorded.

Accounting's twentieth-century response to its own valuation problem was to import an external anchor: rather than relying on internally generated estimates of an asset's worth, fair value accounting instructs firms, wherever possible, to value assets and liabilities at the price they would fetch in an active market. For assets that trade continuously in deep, liquid markets, publicly listed shares, government bonds, this solution works remarkably well. The market supplies a price nobody inside the reporting company had to estimate, and that price updates continuously as circumstances change, giving fair value accounting a claim to accuracy that historical cost accounting, frozen at whatever price an asset happened to be purchased for years or decades earlier, cannot match.

The solution depends entirely, however, on the market in question actually being liquid, and the 2007 to 2009 financial crisis exposed what happens when it is not. Complex mortgage-backed securities and collateralised debt obligations, many of them thinly traded even in ordinary times, saw their markets seize up almost entirely once the underlying mortgages began defaulting at scale. Under fair value rules, banks holding these instruments were required to mark them to whatever price could still be observed in the market, and as trading dried up, the only observable prices were increasingly desperate fire-sale transactions by the small number of sellers still willing to trade at any price. Banks were forced to write down assets to reflect these fire-sale prices even where the underlying mortgages continued paying and the securities' eventual cash flows might well have justified a far higher valuation. The resulting write-downs reduced regulatory capital, forcing further asset sales to meet capital requirements, which depressed prices further still, a self-reinforcing spiral that the economists Christian Laux and Christoph Leuz, examining the episode afterward, argued fair value accounting had significantly amplified rather than merely reported (Laux & Leuz, 2009).

The lesson generalises directly from the argument developed across this article. Fair value accounting is itself a second-order attempt to solve accounting's original fit problem, substituting an external market price for an internally generated estimate. That substitution works precisely to the extent that the market supplying the price remains an actual mechanism of price discovery, itself a hospitable, functioning coordination between many independent buyers and sellers. When that coordination breaks down, as it did in 2008, the market price stops reflecting anything resembling the asset's underlying worth, and fair value accounting, rather than correcting for the loss of fit, imports the resulting distortion directly onto the balance sheet with all its usual borrowed authority intact.

10. The Audit Society

Accounting's expansion did not stop at the edge of the commercial firm. Across the final decades of the twentieth century, the logic and apparatus of financial audit, independent verification of an organisation's own reported figures, was exported wholesale into universities, hospitals, schools, and public services that had never kept financial accounts of this kind and were not, in the relevant sense, businesses at all. The sociologist Michael Power, writing in the late 1990s, described the resulting transformation as the emergence of an audit society: a mode of governance in which trust is established not through direct engagement with an institution's actual work but through the proliferation of audit trails, performance indicators, and verification rituals that certify the work has been done properly, without the certifying process itself ever making direct contact with the work being certified (Power, 1997).

Universities came to be measured by research output metrics, student satisfaction scores, and graduate earnings data. Hospitals came to be measured by waiting-time targets and readmission rates. Each of these measurement regimes borrowed accounting's institutional authority, its apparatus of standardisation, comparability, and audit, and applied it to domains that had never before been organised around comparable, aggregable figures: teaching quality, patient care, scholarly insight, the slow accumulation of trust between a clinician and a patient across a long course of treatment. The same pattern traced throughout this article recurs at institutional scale. Waiting-time targets can be met by reclassifying which patients count as waiting, or by prioritising easily treated minor cases over complex ones that would consume disproportionate time against the target. Research output metrics can be met by fragmenting a single substantial finding into several minor publications, or by favouring safe, quickly completed projects over ambitious, longer, more uncertain ones. In each case, the target is met and the underlying reality the target was meant to protect, genuine patient care, genuine scholarly advance, quietly deteriorates, hidden from view precisely because the metric designed to reveal it has itself become the object of management.

The audit society compounds the difficulty accounting already faces within the commercial firm, because it applies accounting's characteristic apparatus, standardisation, comparability, verification, to domains that lack accounting's original advantage. A hospital admission is not a completed, already-symbolic transaction the way a sale is. It is an ongoing, interrecursive relationship between clinician and patient, embedded in bodies, in trust, in judgement exercised moment by moment, precisely the kind of process this article's earlier sections identified as resistant to symbolic capture from the outset. Auditing it does not merely risk the estimation errors that trouble goodwill or depreciation. It risks reorganising the underlying activity around the demands of the audit itself, until the audit trail becomes more real, institutionally, than the care it was built to verify.

11. Where the Ledger Belongs

The argument developed across this article should not be read as a case against accounting, and it is worth closing by stating plainly why not. Cash still needs to be counted. Inventory still needs to be tracked. Debts still need to be recorded, and shareholders still deserve an honest account of what has been received, spent, and owed. Within its original, hospitable domain, completed, discrete, already-symbolic transactions, double-entry bookkeeping remains one of the most successful and least improvable technologies human beings have ever built, five centuries old and still, in its basic logic, unsurpassed.

What the argument does suggest is a discipline of attention that accounting's own best institutions already gesture toward but rarely apply consistently. The true and fair view override, examined in an earlier section, supplies the clearest available model: a rule-based system that has built into itself, as a matter of binding law rather than aspiration, the requirement that judgement be allowed to override the rule whenever mechanical compliance would produce a picture that does not match the underlying reality. Materiality thresholds, the accounting convention that trivial errors need not be corrected because they would not mislead anyone relying on the accounts, work on a similar logic, treating the ultimate test of an account's adequacy as its fit with what a reasonable user would actually need to know, rather than its arithmetic perfection considered in isolation. Professional scepticism, the standard auditors are formally trained to apply, asks precisely the question this article has been asking throughout: does this number, however precisely stated, actually correspond to something real, or has it detached from the reality it claims to report while retaining the reality's institutional authority.

It is worth acknowledging, in fairness to the profession, that the institutions meant to perform this checking function do not always succeed. Arthur Andersen, Enron's external auditor, signed off on the very accounts this article has already described as fraudulent, and the firm's subsequent collapse, following a criminal conviction later overturned on a technicality, stands as a reminder that professional scepticism is a discipline auditors are trained toward rather than a property automatically guaranteed by the existence of an audit. The true and fair view override, materiality thresholds, and professional scepticism are institutional resources, not automatic safeguards. They work only when the people applying them are willing to notice, and to say aloud, that a number no longer matches the reality it claims to describe, against every institutional pressure pushing them to sign off and move on.

The task, item by item and line by line, is to ask of each figure on a balance sheet the same question accounting law already asks of the statements as a whole. Does this number descend back into the reality it claims to describe, the way a cash balance does, confirmed by counting notes in a till? Or does it merely resemble that kind of number while resting on an estimate, a convention, a market price borrowed from a market that may or may not still be functioning, or a target that has, through years of being managed to, ceased to track anything beyond itself? Accounting cannot answer this question about itself from within its own rules, any more than a mathematical model can certify its own fit with the physical world it is applied to. The answer has to come from outside the ledger, from the felt, lived, mesocosmic reality the ledger was always meant to serve and, at its best, still does.

12. A Live Case: Carbon Accounting and the Limits of the New

Not every extension of accounting's apparatus into new territory repeats the pattern traced so far, and the clearest counter-example is unfolding as this article is being written. Climate change has forced a wave of new accounting standards, carbon accounting, greenhouse gas emissions reporting under frameworks such as the Greenhouse Gas Protocol, and the sustainability disclosure requirements now being folded into mainstream financial reporting by bodies including the International Sustainability Standards Board, all of them asking companies to measure and report something no ledger had ever been built to track: the carbon dioxide equivalent released, directly and indirectly, by their operations and supply chains.

The temptation, having spent this article's preceding sections cataloguing accounting's overreach, is to expect this extension to fail the same way goodwill, standard costing, and fair value on illiquid assets have failed. The more interesting and more honest observation is that it need not, and for a specific, diagnosable reason. Goodwill and standard costing were asked to symbolise something that already existed as functioning, mesocosmic coordination before the ledger arrived, an acquired workforce's accumulated trust, a factory's actual pattern of resource consumption, and in each case the symbol displaced or distorted a coordination that had previously managed perfectly well without it. Planetary carbon cycles present the opposite case. There was no pre-existing local, felt, embodied coordination through which a company's managers tracked their contribution to the accumulation of atmospheric carbon dioxide, because the relevant timescales, causal pathways, and spatial distribution of the harm sit entirely outside anything a body, a relationship, or a single institution could register directly. Nobody feels their firm's supply chain emissions the way a factory manager feels an unrealistic production quota. Carbon accounting does not intrude on an existing coordination and force it into symbolic form. It makes available, for the first time, a domain that had no coordination to intrude upon.

This does not mean carbon accounting is immune to the failures traced throughout this article. Scope 3 emissions figures, covering a company's entire indirect supply chain, are frequently estimated through industry-average multipliers applied to spending data rather than through anything resembling direct measurement, producing numbers with an air of precision that a closer look does not support, goodwill's problem recurring in a new register. Carbon offset markets have shown patterns strikingly similar to the mark-to-market abuses traced earlier in this article, with credits sold against emissions reductions that a number of independent investigations have since found were never actually achieved. The diagnostic developed across this article applies here exactly as it applies everywhere else: ask whether the number descends back into a reality it can be checked against, or whether it has detached from that reality while keeping the reality's institutional authority. What distinguishes carbon accounting from goodwill in principle, even where it fails in practice, is that its underlying domain was never symbolisation-hostile in the way an acquired firm's culture or a family's caregiving is. It was simply unmeasured, which is a different and considerably more tractable problem.

13. Conclusion

Accounting began as a technology of extraordinary, well earned humility. Pacioli's method claimed only to record what had already happened, in transactions that had already been agreed, priced, and completed by the people party to them. Within that narrow domain it achieved a fit between symbol and reality that few quantitative technologies invented since have matched. The history traced across this article is the history of that humility eroding, gradually and for understandable institutional reasons, as the same technology was asked to do more: to value the whole of a firm rather than merely its completed transactions, to price instruments no market still meaningfully traded, to govern institutions that had never been businesses in the first place.

At every point in that history, the diagnostic this article has offered remains the same. A ledger's internal consistency, the simple, elegant fact that debits equal credits, guarantees nothing about whether the resulting figures still refer to anything in the world beyond the page. Enron's books balanced. WorldCom's books balanced. A hospital that has optimised its waiting-time statistics perfectly may have made the underlying care measurably worse. Formal correctness and mesocosmic fit are different achievements, and accounting's five-hundred-year history is, among other things, an unusually well documented record of how easily the first gets mistaken for the second.

Carbon accounting, examined in the preceding section, shows that this is not a case against extending accounting into new territory as such. Some previously unmeasured domains are, like planetary carbon cycles, waiting to be measured rather than already coordinated in a form measurement destroys. The task accounting has never adequately built into its own standards is the one this article has tried to supply: a reliable way of telling, in advance rather than only after a collapse or a crisis, which kind of domain is in front of it.

The ledger belongs where transactions already are: discrete, dated, agreed, complete. Everywhere else, it is borrowed authority, precise in appearance and, unless someone is still willing to ask whether the number matches the life it claims to report, empty in fact.