Chapter 10. The Waiting and Who Pays for It

This chapter takes up the second sub-question set out in the Introduction: what the records say about deadlines, about the cost of waiting and about who bears it while the answer is not in. It reads six of the 53 records on axis A3, and closes with the ledger's hypothesis about who bears the waiting. Each reading asks what the waiting costs, who is named in the setting as affected, and what the ledger can and cannot code about that.

10.1What the records say about deadlines and horizons

The series asks what must be decided now and what would make the decision change, and its theme is the time between a question and its answer. Axis A3 was written to carry that time: whether a record states a deadline, how long the horizon of the setting runs, and who waits and pays while the answer is not in.

Of the 53 records, none states a deadline. The ledger's count is exact: cases with a deadline in the record, 0 of 53. No field in the record form asks "by when". That absence belongs to the form, and it settles less than it seems: the form does not stop a date or a period being written into the Scenario, the Question or an Option, and in these records periods do appear there.

What can be read from the records is the horizon — the stretch of time a setting itself names. The coders agreed on that cell in 30 of 53 records, or 57 per cent, and the values they chose were these: unsettled in 23 records, unknown in 18, days to months implied in 11, and years implied in 1.

In plain terms, the coders placed the horizon only in the 11 records coded days to months and the one coded years, and in none of those did a record state a horizon in words both coders found settled; each placement was an inference from the setting's prose. In 18 records they read the setting as naming no horizon at all. In 23 records they disagreed.

That matters for a series about the time between a question and an answer, and it matters in a particular way. The records are not silent about delay. Several of the six in this chapter are written around a stretch of time: ten years of rare accidents on a shipping route, a year of steady option premium, blocked redemptions that keep investor payment commitments ahead of asset recovery, a vault in which rewards are paid whenever a profit occurs while losses arrive when they arrive. The delay is present in the setting's prose, in the sentences that describe what has been happening and what is proposed. What those passages do not do is put the delay into one comparable form, and part of the reason is that three different things are being called time here. There is the stretch over which the setting's facts accumulated, there is the stretch ahead in which a decision would have consequences, and there is the stretch someone spends waiting for an answer. The ten years of rare accidents and the year of premium are the first. The horizon cells are an inference about the second. The third appears in a record only where a party is named who absorbs what is unresolved while the question is open, and among these six that happens in the vault setting, where all losses are borne by the depositor.

Time does not live in the Scenario alone. The weekly option desk's Example judgment turns the interval into a count of days; the joint venture's Invalidation line turns on the timing of asset recovery; the vault's Question offers to defer performance rewards for a period and reclaim them when losses occur. The record form has no field that carries a deadline, and the prose fields carry time regardless.

Two consequences follow for this chapter. First, the reading of axis A3 here is a reading of prose, and where the prose runs thin the coders' cells run to unsettled, which is what they did for most of the cases below. Second, the ledger does not decide the horizon hypothesis in this volume; it leaves that hypothesis to be read across the books in Volume 0, and Chapter 14 sets out the verdicts on all eight hypotheses as the ledger gives them.

The six cases read in this chapter, and the axis-A3 cells assigned to them, are these:

Where the coders placed a horizon, the setting named a stretch lying ahead of the decision rather than a date; the marine record, whose calm decade lies behind the decision, stayed unsettled on that cell. The vault record and the two average-performance records carry an unknown horizon cell, because their settings say nothing about how far ahead the consequences run. The periods in these records are written in ordinary units — years, weeks, days — so they can be set beside one another; what the ledger gives is the cell each pair of coders chose, not the reason they chose it. This chapter therefore treats the cells as readings of the prose and the periods as facts of the settings.

10.2Selling the tail in two times

Two of the records in this chapter sit in the same thread, and the thread's own line states the shape of the problem in both.

The present record's line says the same thing about position size, substituting the worst path for the average return. What the two settings do differently is where the small gains are collected and where a rare loss would fall.

Synthetic scene — 1739

In a synthetic setting dated 1739, a marine insurance underwriter faces a shipping route on which accidents have been rare for ten years and on which insurance has been greatly increased, so that stable annual premium revenue has been coming in. The question in front of the underwriter is whether to increase underwriting further, share the risk with other underwriters, or set an underwriting limit by route. Hands work through the contracts that share the same route; eyes follow the rumours of war beginning to circulate between two countries. What is waited for is whether the run of calm continues, and if it does not, whether the capital can meet a total claim that arrives on many contracts at once.

The record's Scenario reads:

And its Example judgment reads:

The three options are to increase underwriting, to share risk with other underwriters, or to set an underwriting limit by route. Read against the waiting, each option moves who holds the claim when a common event arrives. Increasing underwriting collects more premium now on the assumption that the calm run continues, and concentrates in one place whatever a single event would trigger. Sharing risk gives up part of the premium to spread a claim that would otherwise land in one place. An underwriting limit by route caps the claims the underwriter may have to pay, and caps the premium in the same breath. The waiting in this record runs on a scale the setting names in years: the ten years of rare accidents stand behind the stable premium, and the interval that would test the capital is not given a length.

Synthetic scene — 2026

On a synthetic desk in 2026, out-of-the-money BTC put options have been sold every week for a year, and steady premium has been earned without a loss over that course. The investment committee proposes increasing the position size because returns are stable. Hands mark the week's positions as they close; eyes read the proposal and the run of premium behind it. What is waited for is the committee's answer, which the record gives no date for; the record's own judgment tests the position against past crash intervals rather than against any interval it says is coming.

The record's Scenario reads:

And its Example judgment reads:

The options are to scale up, to set the position size on the worst-case interval loss, or to buy further puts with part of the premium. Read against the waiting, scaling up collects more premium each week now and enlarges what a single interval can take; sizing on the worst interval reduces what the desk earns in the quiet weeks; buying puts with part of the premium trades a certain weekly outflow for protection against the interval itself. The waiting here is measured in weeks, and the record's own judgment converts the interval into a count of days — the unit in which a year of premium would disappear.

Who collects now and who pays later is the question axis A3 carries, and the two records code it differently.

In the 1739 setting the Scenario places the accumulated contracts in one underwriter's book, so a coder could read the underwriter as the party who would pay when a common event came. In the 2026 setting the Scenario names an investment committee that proposes, and does not say who would bear the loss if the interval arrived after the size was raised, so the bearer cell stayed unsettled. This study reads the difference as a difference in the records' prose rather than in their economics: both settings describe a stream of small gains and a rare loss, but only one of them names where the claim would land.

Two literatures bear on this pair, and each fits partially.

Raghuram G. Rajan's paper Has Financial Development Made the World Riskier? (2005) argues, as its reading note gives it, that managers paid on returns measured against their peers have reason to take risks that pay off most of the time and lose heavily only rarely — tail risk — and to herd, so that financial development can raise the risk of a severe crisis even as it spreads risk. The fit with the 2026 record is specific and partial. The committee's proposal is stated in the Scenario as resting on the stability of returns. This study reads that as a resemblance to Rajan's argument and not as an instance of it: the note's own terms are pay measured against peers, tail risk and herding, and it does not describe stable past returns as grounds for more exposure. What neither record supplies is the arrangement Rajan's argument turns on. No record in this chapter names a manager's pay, a benchmark or a peer group, so the records show a shape of reasoning whose cause is not among them. The 1739 record fits Rajan's argument less well still: it describes a book that grew over a calm decade, with no peers and no pay named anywhere in the setting.

Nassim Nicholas Taleb's The Black Swan (2007) is described in its reading note as a book about rare events of extreme impact that are explained, after the fact, as if they had been predictable. These records cannot show that habit, because they have no fact after the fact: they are settings written before any rare event arrives. What they can show is what has been set down before the outcome. Both example judgments separate a stretch without a loss from a strategy that carries loss risk, which is the material such a later explanation would work on. One distinguishes revenue earned during a period of rare accidents from the risk that a single event triggers claims on many contracts at once. The other distinguishes a period with no loss from a strategy with no loss risk. Both are dated readings attached to synthetic settings and both are revisable. This study reads the two judgments as an attempt to keep the state of each position visible before any rare event, since nothing in the records shows whether such a line would hold once an event had been explained.

The waiting in these two records runs on different clocks — years in the first, weeks in the second — and in neither is the bearer of that waiting named in words the coders could settle. The condition attached to the 1739 record's judgment is among the Invalidation lines that Chapter 13 returns to; whether conditions of that kind can be observed is that chapter's subject. Here it is enough to record that the interval, in both settings, is described as a possibility and never as a date.

10.3Averages and expectancy

Two records from the September 2026 atlas expansion ask, in the present, whether a strategy's average behaviour justifies adopting or enlarging it. Both are written in the expansion's form, with an assumption, required evidence and tags above the Question, and both carry options and an Example judgment.

Their Questions read:

Axis A2 is meant to hold the grounds a decision rests on and what is missing from them. Neither of these two records lists what is missing from the grounds it names, and the reason is worth spelling out, because both of them do name evidence.

The first record's Scenario names, as required evidence, the lower tail of the profit and loss distribution, execution and margin data through drawdown and stress intervals, and the recovery period after losses. The second names realised win rate, average profit, average loss, cost per trade, the frequency of large losses, and sample uncertainty. This study reads those required-evidence passages as naming tasks rather than as flagging gaps: they say which data an analysis must gather, not what the analysis would then be unable to support. That reading belongs to this study and not to the ledger, whose A2 row counts only material a record explicitly lists as missing, and no record in the 53 lists any.

The two Example judgments then say what each record's own reading makes of those grounds. The first declines to adopt on average indicators alone, and treats the loss tail, the recovery period, the margin requirement and execution availability in a crisis as separate constraints. The second declines to treat a win rate as equivalent to a strategic edge, and holds that scale is not increased until verification shows net expectancy has improved. Both are dated WSTiger readings on synthetic settings, and both are revisable; this chapter reports them as the records' readings, not as its own conclusions.

What the two Questions leave out is as instructive as what they include. Neither names a decider, and the ledger's A1 row counts cases with the decider in the record as 0 of 53. Neither names a horizon that both coders could settle: both of these records carry an unknown horizon cell. Neither names a person or a role who would bear the cost of a wrong adoption or a premature enlargement, though the first Question speaks of rare losses damaging the account and the second record's bearer cell reads as the decider, implied. The grounds are written as an analyst's checklist, while the doubt — how rare is rare, how many samples suffice, whose account absorbs the shortfall — stays outside the record's fields.

One detail cuts the other way and should be recorded. The second record's required evidence names sample uncertainty in its own words, and it is the only one of the six records in this chapter that does so. This study reads that naming as the record conceding the inference problem inside the Scenario field rather than as a statement of what is missing from the decision — a concession written into the setting, not a gap flagged for the reader. The A2 row is built to hold that distinction, and in these two records the distinction is visible in the prose.

The waiting in both records is the time before verification. Neither Question says how long that time runs, and neither says who waits through it: whether it is the analyst assembling the drawdown data, the committee holding a decision open, or the account absorbing whatever happens in the meantime. Chapter 13 reads the conditions attached to revisiting, for the records it returns to; Chapter 11 reads what can be undone. For these two records the chapter's finding is narrower: the grounds are written, the waiting is implied, and the record form has no field in which to place either the deadline or the bearer.

10.4Paid now, lost later

Two records set far apart in this study's time span describe the same shape from two positions: one party is paid while another's loss stands unresolved, and the answer is not in. The first is set in 2034, the second in 1974.

Their Questions read:

One Option from each is quoted here because each states, in the record's own words, a change in the terms of payment:

Synthetic scene — 2034

In a hypothetical on-chain operating vault of 2034, the persons in charge submit proposals, one after another, to raise leverage. Every time a profit is made, governance tokens are paid to the person in charge as a performance reward, while all losses are borne by the depositor. Hands draw up the next proposal; eyes follow the profit events that pay the rewards. What is waited for is the profit that pays and the loss that the setting places on the depositor, and the setting puts nothing at stake for the persons in charge beyond the rewards already on offer.

Synthetic scene — 1974

In a synthetic joint venture operating entity of 1974, the books show a profit while some asset redemptions stand blocked, and payment commitments to investors stand in front of asset recovery. Hands work through the schedules and the accounts; eyes track which redemptions are blocked and what cash is actually available. What is waited for is the recovery of assets that has not been confirmed, and the date on which investor payments fall due. The setting gives no date for either.

Michael C. Jensen and William H. Meckling's Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure (1976) is described in its reading note as an argument that when one party acts for another, agency costs arise — monitoring, bonding, and the loss that remains; among them, owners of a firm financed by debt have reason to move into riskier projects whose losses fall on creditors. The vault record resembles that structure in one respect. The persons in charge receive governance tokens as a performance reward each time a profit is made, the setting states that all losses are borne by the depositor, and proposals to raise leverage arrive one after another. The reward follows the gain; the loss follows the depositor. What Jensen and Meckling's note identifies is the incentive to move risk onto the party that absorbs the losses, and the vault setting supplies that asymmetry without the debt channel their mechanism requires. This study reads the vault record as a resemblance and no more: the split of gain to one party and loss to another is in the setting, while the financing conditions that explain it in their account are not. The joint venture record fits the lens poorly: its problem is liquidity and the timing of redemptions, not a choice between projects, and the option to negotiate payment schedules changes timing rather than incentive.

The two records draw the line in their own words, without the terms of Niklas Luhmann's Risk: A Sociological Theory (1993). In the vault setting, the persons in charge submit proposals to raise leverage and the setting describes the losses as falling on the depositor; whether the depositor has any part in the decision is not stated. In the joint venture setting, distributing the reported profit is a timing question, and the setting places the investor payment commitments before asset recovery rather than after any decision. This study reads both settings as written from the side that poses the question, which is why the bearer cells are so difficult to fill: the deciding party's problem is stated in full, and the affected party is named in passing.

Axis A3 codes the difference sharply:

The vault record is the one case in the whole corpus of 53 where a coder's reading named the bearer as those who did not decide — third parties or the vulnerable. It is also one of the five lenient readings coded as supporting H1, the hypothesis that the cost of waiting falls on someone other than whoever decides. The joint venture record's bearer cell stayed unsettled, and its horizon cell too: the Scenario names profit, blocked redemptions and payment priorities, and leaves the identity of the party who would bear the delay to the reader.

Read against the waiting, each Option moves cost between positions. Keeping the current vault structure collects rewards now and leaves the loss shape unchanged. Deferring and reclaiming ties the reward to a later check, and the setting states that whether reclamation can occur after losses must be checked together in the protocol and the execution code — a condition written as an investigation, not as a date. A leverage cap in the protocol would constrain the leverage the vault can hold and would likely cost it part of what it earns in good periods; the record states neither effect, so both are readings rather than consequences the setting draws. For the joint venture, distributing reported profit pays investors now on the strength of a book entry; the available-cash basis stops the schedule resting on unconfirmed recovery; negotiating payment schedules moves the timing of commitments to a later date, at the cost of changing what investors were promised. Each option redistributes the waiting between the entity and those who were promised payment, and no sentence in either record says when the answer arrives.

The timing in the vault setting sharpens the asymmetry the reward creates. The reward is paid at every profit event, so it is collected often and soon, while the loss the setting places on the depositor has no schedule at all. The option to defer the reward and reclaim it when losses occur would move the payment from the profit event to the loss event and bring the two into the same time; whether the reclamation is enforceable is exactly what the record's own judgment says must be checked in the protocol and the execution code, and the record's condition for revisiting asks for evidence that such a protocol has been enforced. In the joint venture the asymmetry is of another kind: the profit sits on the books, the cash does not, and the payments promised to investors come before the recovery that would fund them.

These two records can be read beside the marine book and the option desk, but not as one shape. The marine case codes the decider as the bearer of the loss, the option case leaves the bearer unsettled, and the joint venture reports payment commitments rather than a stream of distributions. What the settings share is a payment owed against a loss or a recovery that is not yet resolved, and an answer that is not in.

10.5The bearer and H1

The last question of the chapter is who bears the waiting, and the ledger's coders settled that cell only part of the time.

The coders agreed on the bearer cell in 24 of 53 records, or 45 per cent. That is the lowest agreement of the reading cells except the two reversal cells, and the values the coders chose were these: unsettled in 29 records, the decider implied in 12, unknown in 7, those nearest the stoppage implied in 4, and third parties or the vulnerable implied in 1.

The verdicts on H1, in both readings, are these.

The strict verdict needs almost no argument. H1 compares two terms, the bearer of the waiting and the decider, and the strict reading scores a case only where both coders found both terms stated in the record. On this corpus the strict counts are 53 unclear and none either way, and unclear then outranks every other count. The A1 extraction is part of the reason but not the whole of it: the ledger's row reports no case with the decider in the record, while the coder table does carry stated decider values, one individual and one team or organisation. What blocks a strict H1 score is that no case settles the two terms together, not that no setting names a decider.

The lenient reading, which also counts what the setting implies, comes to the same verdict by a different route. It codes five records as supporting the hypothesis and twelve as going against it, and leaves 36 unclear. Since unclear cases outnumber the decided ones, the verdict rule returns unclear. The share the ledger gives for this reading is 29 per cent, which is the share of the decided cases that support the hypothesis.

Why the coders could not agree is visible in three properties of the records themselves.

First, the pair the hypothesis tests never appears in words. No record names a decider, and no record names a deadline; the ledger's counts for both are 0 of 53. The hypothesis asks whether the bearer of the waiting is someone other than whoever decides, and the form gives no field in which either term could be written. The bearer cell must therefore be inferred from the Scenario prose, and inference on that cell is where agreement collapsed: 45 per cent.

Second, the inference runs into a structural feature of how these settings are written. This study reads the records as written from the position that holds the question — the underwriter's desk, the investment committee, the persons in charge of the vault, the operating entity's payment schedule. In that position the bearer is either the decider itself or a party the setting names in passing: depositors, investors, counterparties, those nearest a stoppage. Where the setting is written from inside one book or one desk, the question "who waits?" collapses into "the whole record", and a coder must decide whether the party who decides and the party who pays are the same party. That decision is a judgment, and the two coders made it differently in 29 cases.

Third, the lenient cells, read against the lenient H1 cells, show how much of the verdict rests on that judgment rather than on the settings. This study reads the lenient H1 cells as resting on that judgment: where the bearer cell read as the decider the reading went against H1, and where it read as someone other than the decider — the four nearest the stoppage and the single one coded to third parties or the vulnerable — the reading supported H1. The remaining thirty-six records leave the relation unsettled in the lenient reading as well as in the strict one. That is a limit on what H1 can score; it does not make the axis unreadable, since A3 also carries horizons, deadlines and the cost of waiting. The unsettled cells are not a judgement about markets or about the studio that writes the records.

Two rules from Chapter 2 govern what may be concluded from all of this. The absence of a decider field and of a deadline field is guaranteed by the record form, so the low visibility of the decider-bearer separation is not evidence that organisations hide who pays; it is evidence that the form does not ask. And the cell values are readings by two models of different families, whose agreement is reported rather than assumed.

The question of who waits is not the same as the question of who loses at the end. A party can absorb a loss when the answer arrives and still wait on someone else's decision in the meantime, and a party can wait for an answer without being the one who pays for it. In the six records of this chapter the two are kept apart only in the vault setting, where the depositor is named as taking the losses while the proposals come from the persons in charge; in the marine setting, on the coder's reading, the party who decides is also the party who would pay the claim, and in the joint venture the bearer cell stayed unsettled. That gap between deciding and paying is what makes the cell hard to code, and it is also the gap the hypothesis is built to test.

What the six records of this chapter show, despite the coding difficulty, is the shape of the waiting in a few concrete settings: a stream of small gains over ten years or one year, a rare loss that can land on many contracts or on a single interval, a reward paid to one party while losses stand at another, and a profit on the books whose recovery date is unknown. In each setting the record states an exposure that the waiting leaves open — claims against capital, a run of premium that an interval could take back, blocked redemptions, losses placed on the depositor — though none of these is stated as a cost caused by the delay itself, and none names who waits for the answer or by when it falls due.

10.6What would change the judgments

Every record carries a line stating when its judgment would be revisited, and those lines are the only part of the form where a future trigger appears in the record's own words. The six cases of this chapter write their triggers in very different shapes.

The weekly option desk's judgment is re-judged on a replay of the crash interval, and the trigger is a pair of measurements: losses inside the capital limit, or hedge cost above the premium.

The vault record asks for two observations that would each change the balance between the person in charge and the depositor: capital of the person in charge actually in the vault, and evidence that a reclamation protocol has been enforced.

The joint venture's trigger concerns the two facts its Question turns on, the recovery of assets and the order in which payments stand.

The marine record's trigger turns on the rumours of war and on how the contracts sit.

The two average-performance records write their conditions as tests to run rather than events to watch for, and each adds a further check on how the analysis itself was built.

None of these lines is a date. Each names something to watch for or something to test, and the two average-performance records name work to do rather than an event to see. The records give the trigger and not the clock: nothing here says how long the stretch before the trigger runs, and nothing here names who waits through it.

This chapter does not conclude that the cost of waiting falls on the decider or on anyone else. The ledger's verdict stands: H1 is unclear in both readings. In the strict reading nothing is scored either way; in the lenient reading five records support the hypothesis and twelve go against it, and the unclear cases outnumber both, which is what the verdict rule returns. Whether a decision can be undone, and what that does to the cost of waiting, is the subject of Chapter 11.