1.1Records and method
The records come from the public catalog of Cosmic Press, which the studio serves as a manifest and a set of numbered detail files. The manifest names 24 topics and 768 entries and gives the catalog's source as "AI-generated examples"; each of the 768 entries also marks itself AI-generated. An entry has a question in the voice of a person, a compressed general form of it, a context, a need, an assumption, a line on what the compression keeps (preserved), a line on what it loses (lost), a counterquestion, a defect, and six scale notes. The scale notes run from the person outward and keep one order; this study reads the six, in the sixteen records, as the person, the people nearby, a deciding body or a rule, other living things, the place or planet, and the largest frame. A record has no options and no judgment, and it never answers its question.
Two marks on the surface of the entries divide the catalog into stretches. The first is the mood of the scale notes. In 504 entries all six notes end in a question mark, and those entries form one unbroken run, from q241 to q744. Outside the run, 199 entries have no question among their notes and 65, all of them before it, have some; the last 24 entries of the catalog return to statements. The second mark is the opening of the context. 288 contexts open with the label "Hypothetical situation:", and 244 open instead with "A hypothetical ..." followed by a setting. Within the questioning run the two openings alternate in long blocks: q241 to q264 carry the label, all 240 entries from q265 to q504 open with "A hypothetical", and the label returns from q505 to q744. The study does not know how the stretches were made and claims nothing about it. It treats them only as two manners of writing that can be told apart on the page.
The selection rule is one sentence: every entry in the studio's seventeenth detail file, q257 to q272. The studio's files hold 16 entries each, and there are 48 of them. The seventeenth lies inside the questioning run and holds the boundary between the two manners: q257 to q264 are the last eight entries of the first labelled block, and q265 to q272 are the first eight of the long block that opens with "A hypothetical". The file therefore lets the paper read the questioning ladder in both manners under one rule, with one entry from each of 16 of the catalog's 24 topics. The cost is that sixteen records are a sample of one boundary, not of the catalog; the conclusion returns to this.
Cosmic Press publishes its records in Korean. The English in this paper is a translation made once by the editorial session of this study, sentence by sentence, with nothing added; each case's English seed names the Korean original by its hash, and the appendix gives the English of the fields the paper reads. Every block quote is that English, word for word. Every count is made in Korean by a plain pattern named beside it in the study's numbers, so that the translation does not decide it; the tests are crude, and Section 3.3 says where.
All counts come from a script (scripts/count.mjs) run on the kept records and a snapshot of the whole catalog, fetched within the same minute. It also takes the series' frame measure (spine §6): for each field, the share of its five-word runs that recur in at least 5% of the records that fill it. Every field measures 0%, in the whole catalog and in the questioning run in Korean, and in the sixteen in English. The record set does not repeat sentences. Its frame is in its slots: every record has a compressed form, a lost line, a defect line, a counterquestion and six notes in a fixed order, and in this run every note is a question. Those slots are guaranteed by the form, and the paper does not report their presence as a finding. What it reports is what the records put in them.
All 16 contexts call their situation hypothetical. The records mention generators, freezers, medicines, saws, security lights, migratory birds and cities. None of that is information about the world, and nothing in them is advice; the paper treats each only as the setting of a hypothetical question.
1.2A ladder that hands the question on
In this study's reading, the ladder of questions does not narrow a question. It passes it along. The clearest case, in an AI-generated record, is the hypothetical shift worker who keeps receiving notice of a neighbourhood meeting held during working hours. The question is a complaint in the first person:
If the meeting that decides where the benches in the neighbourhood shelter go is always held during my working hours, is my opinion open to it too, merely because I am told I may attend?
and the ladder that follows it reads as a chain of addressees:
How should one see the choice of not speaking out of tiredness, even when there was a chance to speak? When passing on the opinion of a neighbour who is not present, what should be checked? Should there be a way to take part in the decision that sets the meeting time, too? Who speaks for the needs of beings that cannot take part in the meeting themselves, and on what grounds? How can people who live on different hours make the rules of one place together? May only those attending now make a choice that affects residents who have not yet come?
Each rung asks a different party. The first asks the absent person how to regard a silence that tiredness produced. The second turns to those who were there and might speak for the absent one. The third moves to the procedure itself and asks whether the time of a meeting should be open to decision. The fourth reaches beings that cannot attend any meeting, the fifth a place shared by people on different hours, the sixth residents who do not yet exist. By the end the shift worker's grievance has been put to six audiences. None of them is told what to do, and none is named as the one who must answer. The ladder says who else could be asked.
The pattern holds across the file. In 15 of the 16 records the first rung speaks in the first person singular, and in all eight of the second eight it does so even though only one of those eight questions is itself in the first person. The second rung turns to people near at hand in 15 records; the third asks who should decide, design or pay, or names a public body or a rule, in 12, and in all of the second eight; the fourth reaches a being other than human in all 16; the fifth names a place, a city, a region or a planet in 13. Across the long block that the second eight open, the first rung is in the first person in 240 of 240 entries and the third names a decider or a public body in 155 of 240. The ladder's order is the studio's form, so its presence is no finding. What the records put in it is: in this run, each rung is an address.
The second eight make the address explicit, and all eight are set in a shared space or around a shared thing. The hypothetical cooperative childcare centre in another AI-generated record, whose shared pram creaks so that no one dares use it, shows the result most plainly in its first three rungs:
Do I have the honesty to fix, or to report, a small scratch on a shared tool I used? Can neighbours build a culture of mending damaged care goods together instead of blaming one another? Should the basic local government give public support to the upkeep of shared care infrastructure in apartments and row houses?
One person's honesty, the neighbours' own arrangements, a public body's support: the three rungs lay out side by side the three places where the upkeep of a shared good could rest. These are the parties of the debate Ostrom, Governing the Commons (1990), describes, with one missing: the record has no market rung, though another record of the eight names the logic of the market in its assumption, only to say that neither of its two uses of a shared resource can easily be given up or replaced by it. In this study's reading, the ladder sets the record's question at every level of the debate and declines, as the record form always declines, to say which level should hold it. Ostrom's subject is the conditions under which such problems have been solved; the records cannot speak to that, since nothing in them is solved.
The record on the shared yard, whose soil bed a meeting proposes to pave, shows the same three rungs in a harder key:
Can I propose to the neighbours that we set a weekend and do the work of clearing the droppings and rubbish from the bed together? How should a city legislate ecological building rules that make soil area, as a habitat for living things within residential districts, mandatory?
The second rung offers the neighbours a weekend of shared work; the third asks a city how it should legislate. Between a weekend and a statute the record draws no line and states no preference.
In this study's reading, the questioning ladder is a device for drawing the kind of link Mills, The Sociological Imagination (1959), asks the imagination to see, and it draws it in a fixed order, from the person to the institution and beyond. Blumer, Social Problems as Collective Behavior (1971), gives a longer path than the ladder walks. In this study's reading, the ladder reaches official policy at its third rung, in several records in the words of obligation ("should"). It never shows the steps between. No rung says that anyone took the question up. The ladder names the parties to whom a question could belong, which is the first step of definition, and stops there.
1.3A top rung that turns to time
At its top the ladder of questions looks past the present. All 16 sixth notes of these AI-generated records do so by the word test: they name the future, a time after something has ended, or ask what will be. In the first 240 entries of the catalog, whose notes are mostly statements, only 21 of 240 sixth notes do; there, 155 of 240 deny, limit or separate, by the word test named in the numbers. In the long block the second eight open, 238 of 240 sixth notes look past the present; in the later labelled block, 156 of 240. Only one of the sixteen sixth notes names the universe.
In this study's reading, most of the first eight records turn the top rung toward a time after the asker. The hypothetical family photo ends there:
After no one remembers that day any more, what will become the meaning of the photo?
and so do the records on unthanked work and on a promise kept late:
In a future where my name does not remain, what will the fact that I did this work mean?
In a far future where only records remain, can it still be said that the promise was kept?
These notes do not widen the question to a larger space. They remove the asker from it. The question of whether the cleaning of a shared entrance meant something is put to a time when the cleaner's name is gone; the question of whether a promise was kept is put to a time when only records are left. In this study's reading the move tests whether the question still stands when the person who asked it is not there. It does so by asking, and it leaves the answer open.
The meeting record's sixth note, quoted above, asks whether those attending now may decide for residents not yet come. In this study's reading, that is presentism in a neighbourhood's terms, the bias Thompson, Representing future generations (2010), names, and the record stops where his argument starts. It asks whether the present may decide for the future; it does not ask what the present would have to keep in order to decide well. In this study's reading, the third rung of the same record (whether there should be a way to take part in the decision that sets the meeting time) is the nearest the ladder comes to Thompson's concern with the process itself, and it is put about a neighbourhood meeting, not about generations.
The second eight turn the top rung toward a future society remade. Seven of their eight sixth notes name a world made by technology or a civilisation, and all eight ask "will it" rather than "may it":
In a highly technological society where total surveillance is possible, will the classical penal concept of collective responsibility disappear?
In a settlement on an artificial planet where every surface is covered with synthetic material, what kind of memory will the primal medium called soil remain as?
In a far future where civilisation has collapsed or communication has been cut off, will humans still be able to share common resources for the survival of others?
Here the top rung neither removes the asker nor limits the question. It asks for a forecast. A question about charging every tenant for thefts from a shared fridge ends as a question about whether a concept of punishment will survive total surveillance; a question about paving a flower bed ends on an artificial planet. The single sixth note among the sixteen that names the universe belongs to the rooftop record, and it too is a forecast of governance, not a view of the sky: in human settlements in distant space, to whom will the authority over the artificial sun and the artificial night be given? Gardiner, A Perfect Moral Storm (2006), is concerned with why acting may stay hard even once the ethical questions are answered. This study reads the second eight's top rung as not reaching that worry, because it does not ask anyone to act: it turns a present choice into a question about what a remade future will have kept. In this study's reading, the forecast form lets the small question ride out to the largest frame and leaves it there.
1.4Two accounts of what the frame costs
Every record, AI-generated and set in a hypothetical scene, says what its general question loses and what is defective in it, and closes its account with a counterquestion. The slots are the form's. What the two manners put in them differs, and the difference is the clearest change across the file.
The first eight name a scene that drops out and a party the general form leaves out. All eight of their lost lines end with the words "drop out" or "drops out", and they are short. The snail record's is typical:
The wet path, the small body and the moment when the way it was going cannot be known drop out.
Their defect lines, in six of the eight, say what the general question may cause someone to miss, and four of the eight name a burden, toil, loss or fatigue that falls on a particular person. The record on the unthanked glass says so directly:
If quiet labour is seen only as beautiful, the problem of work piling up on one person can be missed.
and its counterquestion turns the same concern into a question about the community:
Might the advice to find meaning for oneself come to stand in for the community's thanks or a fair sharing of the work?
The record on the room left open for the future does the same with money and space. Its general question asks whether an unplanned place can be left for the future; its defect line says that the burden of leaving possibility open differs with what one can spare, and its counterquestion asks:
Can the same way of preparing be recommended to someone who cannot afford to leave space empty?
The automatic-light record names the person who cannot work out the settings, and the photo record names the next person, who may not read the photo as the family does. In several of these eight, in this study's reading, the record's own account brings in someone absent from the question: the one who does the work, the one without room, the one who comes after.
The second eight change both accounts. Their lost lines no longer end in "drops out", and all eight name a feeling or a felt sense, in several records on two sides at once. The locked table saw is the clearest:
The heavy sense of responsibility of the keeper who wants to guard the tools amid fear of an accident that severs a finger, and the helplessness of the beginner hovering outside the door.
and the rooftop record sets a child's excitement against a resident's unease in one sentence:
The excitement of a child lying on the cold rooftop platform waiting for a dark sky, and the unease of a resident afraid of missing a step on a dark stairway.
What the general question has lost, by these records' account, is the felt weight of both parties at once. The defect lines then turn to a different fault. They say less often that someone may be missed (three of the eight, against six of the first eight); in six of the eight they say that the general form simplifies, splits into two, reduces or distorts, by the word test:
There is a risk of simplifying the many-sided standards of distribution in an emergency disaster into the binary frame of 'individual heating versus shared medicine'.
The risk of reducing the drawing of boundaries to mere selfish territoriality, or of idealising hospitality only as an unconditional virtue.
In this study's reading, the counterquestion takes the other side. In five of the eight it is a negative question, which in English invites the answer yes, and in this study's reading the others argue against the question's leaning in plain form. The yard record's need is to keep a handful of soil in the city; its counterquestion speaks for the people who clean:
Can one insist on ecology while ignoring the real hardship of the ground-floor residents who suffer, clearing the smell and the rubbish every day?
The corridor record's need is the right to have one's identity respected beyond one's own door; its counterquestion speaks for a neutral wall:
Is it not basic courtesy to others to keep a shared passage that everyone goes in and out of neutral, tidy and without colour or smell?
and the fridge record's need is an innocent tenant's wish not to pay for another's theft, while its counterquestion argues for paying:
Rather than the worst outcome, in which no one takes responsibility and the shared fridge itself is closed, is it not better to keep it going even by paying a share?
The second eight's counterquestion is, in this study's reading, a devil's advocate written into the record. The two studies of dissent named in the introduction, by Nemeth, Brown and Rogers (2001) and by Greitemeyer, Schulz-Hardt and Frey (2008), report different things about such a role. Both studies concern people in groups, and the records are not groups: the question and its counterquestion come from the same generated text. The lens shows what the records cannot settle. A counterquestion that argues the other side is contrived dissent by construction, and nothing in a record can show whether it widens a reader's thinking or, as the earlier study that Nemeth and colleagues report would suggest, mostly serves to bolster the first view. In this study's reading, the first eight's counterquestion is in several records a different device: it brings in a party the question forgot rather than arguing the other side.
The shift is visible in the whole account. The first eight's lost lines run to 37 Korean characters on average, the second eight's to 60; the six notes together run to 212 and 301. None of the sixteen defect lines uses the Korean words for compressing or bundling the question, which 54 of the first 240 entries' defect lines did. In this study's reading, the first manner's account of its costs is brief and points at someone absent, and the second manner's is longer, weighs two parties against each other and warns against its own flattening.
1.5What this chapter adds
The chapter has argued three things about sixteen AI-generated records. Asked in the shift worker's case and the pram's, the ladder names six possible holders of a question and gives it to no one. Its top rung looks past the present, toward a time after the asker in the first eight records and toward a remade future society in the second eight. And the record's account of its own costs, in this study's reading, changes with the manner of writing: the first eight name a scene that drops out and a party left out; the second eight weigh the feelings of two sides, warn that the general form splits the question in two, and argue the other side in the counterquestion. To the debate on private troubles and public issues the chapter adds, as this study's reading, a device that draws the links in a fixed order and stops before anyone takes the question up. To the debate on shared goods, a record that lays the levels of governance side by side without choosing. To the debate on the future, a top rung that asks the future rather than binding the present. To the debate on dissent, a counterquestion that is contrived by construction.