The fourteen cost records ask how much and give no sum. This chapter shows how they do it and what they put in the sum's place, and it adds to the debate named in the Introduction: what written help for a person without a lawyer can supply of the inputs that the economics of litigation treats as known, and where the rest has to come from.
1.1Data and method
The data are PRO SE's public library as it stood when it was read for this study. The library list gives 483 questions in four sections: 122 on scope, 121 on documents, 120 on timing and 120 on the courtroom. All 483 are marked synthetic. The list carries an editorial reference date and says that this date is not a case date and does not confirm current law. Each question has its own page, whose data were fetched by plain requests on 2026-10-06 and kept as snapshots in the study's folder; the whole library's English text was fetched the same way, for measuring the studio's repeated phrasing. The snapshots hold, in English and Korean, five fields that this study reads: the question; the evidence, which says what material is examined; the readings, always two; the judgment, which in this library is itself a question; and the revisit condition, which says when the matter would be looked at again. All fourteen records read here are tagged court and law, and none carries a publication date of its own. The English text is PRO SE's own; nothing was translated for this study.
The selection rule is one sentence: every question in the Scope section whose English question contains the word cost or costs. The rule gives 14 records. It is lexical. It leaves out one question in the Courtroom section that names the costs of an expert's appraisal, and two Scope questions that speak of aid or of expenses without the word cost: one compares legal aid with litigation aid, and one asks in what order principal, interest and expenses are paid out of a bankruptcy estate. The set is defined by the studio's wording, not by this study's view of which questions belong together. Of the whole library, 23 records mention cost somewhere in their English text; the rule takes those whose question is about it.
Some properties belong to the studio's form and would be found in any record. Every one of the 483 has exactly two readings, every judgment ends in a question mark, and every record has a revisit condition. These are the shape of the library, not findings about costs. In the cost set the form is also more uniform in its phrasing than elsewhere: 26 of the 28 readings open with "One interpretation holds" or "Another interpretation holds", and 13 of the 14 revisit conditions open with "The matter is looked at again". These openings are not read as content. To see how much of the wording is repeated phrasing, the study measured, for each field across all 483 records, the share of five-word runs that recur in at least 25 records. In the fourteen cases the share is 0% in the questions, 3% in the evidence, 2% in the readings, 1% in the judgments and 15% in the revisit conditions; across the library the revisit conditions stand at 12%. By the series' threshold of 30%, no field here is formula, so what follows each opening can be read as the record's own.
Every count in this paper comes from a script that reads only the snapshots: whether a record contains a figure, whether a question asks how much, whether a record speaks of money paid before the end of the case, what a revisit condition turns on. The counts describe the records, not courts or costs.
PRO SE marks every record synthetic and publishes its library as general information, not legal advice. All fourteen cost records also say in their evidence that they are practice materials, against 110 of the 122 Scope records and 372 of the 483 in the library. Where the records name people, they name roles: an applicant, the opposing party, the winning party, a foreign plaintiff, an administrator or trustee. This study draws no fact about any law, court or person from them. What it studies is the frame: the way a decision about money is put to a reader who has no lawyer.
1.2How much, without a sum
The plainest finding is again an absence. None of the fourteen cost records gives a sum of money, a rate, a share in figures or a number word for one. Across the library, none of the 483 English records contains a digit. In a set about costs, the absence is the first thing a reader meets.
Three of the fourteen questions ask how much or about the amount. One asks how much of the money spent on a lawsuit counts, and its two readings show what stands where a figure might:
Of the money spent on a lawsuit, how much counts as litigation costs?
One interpretation holds that the limit of what was necessary comes first, so only spending tied to the proceedings counts as litigation costs.
Another interpretation holds that for items with a fixed amount, that amount is the measure whatever was actually spent.
The first reading answers "how much" with a test: what was necessary, what was tied to the proceedings. The second answers with a reference: a fixed amount, which the record names and does not state. Neither is a quantity. Both are ways of deciding a quantity, and the record leaves open which way comes first. The question it puts where a judgment would stand says so in its own terms, asking how "the general principle and the item by item methods of calculation" fit together. A reader learns that there are two methods and that their relation is the issue. The number at the end of either method is not in the record.
The question about a lawyer's fee is the sharpest case, because it names a cost that a person who goes without a lawyer may never meet:
How much of the fees paid to a lawyer is recognized as litigation costs?
Another interpretation holds that apart from the agreed fee, only the part within the range the Supreme Court Rules set is recognized as litigation costs.
The reading names a range and the body that sets it. It does not give the range. It also draws a line, in the record's words, between the agreed fee and the part recognized as litigation costs. The record puts that line where a judgment would stand:
What standard separates the agreed fee from the amount recognized as litigation costs?
The word "standard" carries the record's restraint. It asks for a rule of separation, not a result. Two of the fourteen records refer an amount to the Supreme Court Rules in this way, and none quotes them. This study does not supply what they say; the point is that the record shows where an amount would be found and stops there.
The third question that asks how much turns, on reading, into a question about steps. It asks how the amount is fixed when only liability has been decided, and its readings are about the other side's chance to answer:
One interpretation holds that if the opposing party does not file its papers within the set period, the matter can be closed quickly by a decision on the applicant's costs alone.
Another interpretation holds that the amount can be fixed only after the opposing party has been fully given the chance to state its position and file papers.
Here the question how much becomes a question of procedure and of speed. The amount is the output of an exchange of papers, and the readings differ on how complete that exchange must be. The record names a set period without saying how long it is.
The nearest the set comes to a proportion is a phrase about division among several parties:
One interpretation holds that equal shares are the strict rule and that a different division requires clear circumstances.
Equal shares is a fraction without a sum: it says how a burden would be cut, not how large it is. The other reading holds that costs should instead follow "each party's actual involvement and acts". Between the two, the record offers a way of dividing, and leaves the thing divided unstated.
Against the economics of litigation, this is a precise absence. Bebchuk's model lists the size of the amount at stake and the magnitude of the parties' litigation costs among what shapes whether and on what terms a case settles. The cost records supply neither. Hadfield's account of legal fees as a sunk-cost auction, in which spending can rationally exceed the amount at stake, turns on a comparison between two quantities; the records hold neither of them, so the comparison her argument makes cannot be made from them. This study's reading, which goes past both works, is that the records give a reader the variables of the calculation without any of its values.
That is not obviously a fault. The library's own notice says that its editorial reference date is not a confirmation of current law. A sum, a rate or a range quoted from a rule would be a statement about what the law is at a given moment, and the records make no such statement. The absence of amounts is, in this study's reading, a restraint that fits a notice of general information. The cost is that a reader learns from the record where an amount would come from, and not what it is.
The section's finding is that the cost records answer "how much" with a test, a reference or a procedure. They name where an amount would come from, a range set elsewhere, a fixed amount, a period for filing papers, a division in shares, and leave every amount outside the record.
1.3Paying now, bearing at the end
If the records do not say how much, they say a good deal about when and by whom. Their most consistent move is to separate money that is paid or put up while a case runs from the burden that is settled when it ends. Five of the fourteen speak of money paid in advance, deferred, put up as security or advanced from the national treasury; ten speak of how the case ends or of a final liability or ruling; four do both.
The record on advance payment states the separation in the question it puts where a judgment would stand:
How does the duty to pay in advance differ from the final liability for costs?
Its readings describe two outcomes for a step in the case that is not paid for. In one, the court may decline to take it. In the other, the money comes first from public funds:
Another interpretation holds that even without advance payment, where the proceedings would be seriously hampered the costs may first be paid from the national treasury and must be recovered later.
The words "first" and "later" carry the frame of the whole set. Someone pays now; someone else may bear it at the end; recovery is a further step. The record does not say who that someone else will be. Its condition for looking again turns on exactly that: who paid in advance, "or from whom the court decided to recover the costs it paid".
The two records on litigation aid make the same distinction for a person without means. The first asks whether such a person can apply, and the question it puts where a judgment would stand is about order: what must be examined first to decide whether "the proviso" applies. Its first reading shows what the proviso turns on:
One interpretation holds that once insufficient means are shown, litigation aid can be granted unless it is clear that the applicant will lose the case.
The sentence ties help with the price of the lawsuit to a forecast of its outcome. Whether aid is available depends, on this reading, on whether losing is clear. On this reading, the help is itself weighed against a result that cannot yet be known. The record names this connection and does not say how clear is clear.
The second aid record asks whether costs deferred under aid are collected later rather than waived, and its readings are the two ends of that question:
One interpretation holds that costs listed as waived in the aid decision are treated as no longer owed, so there is no later repayment.
Another interpretation holds that aid only postpones payment, so the costs are settled once finances change or a ruling on who bears them is made.
On the first reading, aid ends the debt. On the second, it moves the debt in time. The record does not choose. Its condition for looking again names the two events that the second reading waits on: a change in the party's finances, or a ruling on who bears the costs that is "newly confirmed". The record sets the two readings side by side and leaves open which holds.
Security for costs follows the same path from the other side. One record asks until when security can be sought from a foreign plaintiff and puts the issue as a matter of order: which must be fixed first, the time the ground for security became known or the time of argument on the merits. Another asks how security, once given, comes back:
One interpretation holds that security given by a cash deposit may be recovered by proving that the ground for security has ceased and obtaining a cancellation order.
Money is put up at one point and recovered at another, by proof and by order. The amount is in neither sentence; the route is in both.
The final burden, when the records reach it, depends on how the case ends. The record on when the winner pays sets the general reading and its exception side by side:
Another interpretation holds that because costs follow the outcome on the merits, the costs paid by the winning party are as a rule borne by the other side.
One interpretation holds that even for an act of the winning party, if it was not needed to extend or protect a right, the winning party can be made to bear its cost.
The first sentence quoted here is the only reading in the set that says costs follow the outcome. Set beside Shavell's four methods of allocating legal costs, it reads like the indemnity system he describes, in which the losing side bears the costs, and the other reading like a limit on it. That likeness is this study's observation about the shape of the readings, not a statement about any law, and the record itself names no system. What the record adds to that likeness is the question it puts where a judgment would stand: how it is decided "whether each cost arose from an act needed to extend or protect a right". On its two readings, the allocation is not settled by the outcome alone but cost by cost.
A case that ends without a judgment shows the same dependence from another angle:
One interpretation holds that if the settlement says nothing special about costs, each party bears its own, so no separate decision on costs is needed.
Another interpretation holds that when a lawsuit ends without a judgment, the court must decide on application who bears the costs and how much, so they are not settled automatically.
Read with Mnookin and Kornhauser, the first reading places the cost rule inside the settlement bargain: what a settlement says, or fails to say, about costs decides who bears them. This study's reading, which goes past their account of divorce bargaining, is that the record shows a cost rule casting the kind of shadow they describe, in which the law supplies the framework and the parties' own terms fill it. The second reading takes the matter back to the court. The record's condition for looking again turns on the settlement's wording being "newly confirmed" or on whether an application to fix the costs was filed. On the second reading, even a case that has ended has a cost question still open.
The record on costs on appeal carries the dependence furthest. If the higher court changes the decision on the merits, one reading holds that each instance's costs stand by that instance's judgment, and the other that a later judgment absorbs the earlier ruling:
Another interpretation holds that in the judgment that concludes the case after the merits have been changed or the case has been returned, all costs are determined together, so that the prior costs ruling is absorbed into it.
On this reading, even a ruling on costs is provisional until the case as a whole is over, and the answer for one stage is folded into another.
The conditions for looking again gather these dependences in one place. In 9 of the fourteen records the condition turns on a later ruling, order, application or the end of the case; in 2 it turns on a change in finances; in 4 it names an amount, as something still to be fixed or as a set amount that might change; none names a date. The record on a lawyer's fee puts it most plainly:
The matter is looked at again when the judgment on cost liability becomes final or enforceable and the amount of costs is being fixed.
The amount appears here for the first time as an event: it is "being fixed", at a stage when the judgment on cost liability has become final or enforceable. The record's condition for looking again is set at the moment the amount is being fixed.
Who produces it? The word court appears in 8 of the 28 readings: twice in the Supreme Court Rules, and otherwise as the body that may decline a step, decide on application, approve a form of security, hold the case record or make requests and orders. It appears in none of the fourteen questions that stand where a judgment would. Those questions are impersonal: "How is it decided", "What decides", "What must be examined first". The reader is shown what the court might do in each reading and is not told, at the point of decision, who will decide.
One record steps outside the pattern, and it is the one that most directly addresses the person PRO SE writes for. It asks what help the law provides when costs run short or explanations are blocked, and its readings do not open with an attribution to an interpretation:
If the blockage lies in spoken and written expression, there is room to examine the sequence of the court's requests for questions and clarification, prohibitions on statements, and orders to appoint counsel.
In these practice materials, has it been identified whether the cause of the blockage is a problem of cost, a problem of procedural explanation, or a problem of the ability to make statements?
The record treats cost as one of three possible causes of being stuck, beside not being able to follow the procedure and not being able to speak for oneself. Its first reading says that if cost is the centre of the blockage, the procedure can be understood as leading towards "deferring payment or avoiding collateral". Its second reaches, at the end of a list, "orders to appoint counsel": help that is a person, not a text. Read with Rostain's conclusion that tools for people without a lawyer need to be integrated with human assistance, the record can be taken to point to where written information may need to hand over to someone else. Read with Sandefur's argument that just resolution does not always require lawyers' assistance, it also shows a sequence in which counsel comes last, after requests for clarification and other steps by the court. These readings are this study's; the record does not refer to either argument. Its condition for looking again is the only one in the set written in the first person: "we will examine it again" when more material is added "to these practice materials". The record is careful, even here, to say that it is an exercise.
The section's finding is that the cost records replace the amount with an order of payment. They separate what is paid now, deferred or put up from the burden settled at the end; they tie that burden to how the case ends, act by act, and to rulings still to come; and they make their own reconsideration wait on those rulings. For the debate set out in the Introduction, this is the chapter's contribution: the records give a person without a lawyer the structure of a cost-allocation rule, and they show, in their own words, that the figures the economic models take as given are, in the records' readings, fixed late in the case, by someone the records do not name at the point of decision.