Self-as-an-End
SAE Theory of Monopoly · Paper 5

Technology and Monopoly
技术与垄断

Han Qin (秦汉) · Independent Researcher · 2026
DOI: 10.5281/zenodo.24377558 · Full PDF on Zenodo · CC BY 4.0
Abstract

This paper reads technology: operating systems, cloud, chips, standards and interfaces. Technology is built layer upon layer; a position that controls one layer can use it to shut the door on the next. A position that controls one layer does not, by its control of that layer, acquire the power to shut the door on the next; what is extended into the next layer is examined separately at that layer, and is not approved along with the legitimacy of its position in the layer already controlled. Where standards and interfaces serve as gates, the side that writes a standard and the side that adopts it each answer for what they wrote and what they adopted; where the writer and the adopter are the same firm, both accounts are on it (LD10). Vertical integration can be the capacity to produce and to serve, and checks are not aimed at breaking up capacity; where integration makes substitution, separation or taking over difficult, the structural influence is already there, and it is examined by actual dependence. Preparation for separability and for taking over is one item of checks before the fact: separate interfaces, separate accounts and arrangements for taking over made in advance can serve as examples at the rule layer, and the particular form is examined by effect, cost and alternatives (LD11). The constructed examples are three: a cloud company's scheduling of computing power under "strategic partners first," later the two firms sharing a team and merging accounts, and, asked how long separating them would take, the answer "at least two years, and the risk is very high"; a model company's agent interface, serving as a gate; and, in the far future, farm machines that accept only one firm's designated seed and fertilizer specifications, machines, seed and food extended layer into layer. In reality, the court of appeals in United States v. Microsoft upheld the finding of liability that Microsoft maintained its operating-system monopoly by suppressing browsers, did not find that it monopolized the browser layer, and vacated and remanded the remedies order; AT&T was broken up by settlement in 1984; in Nvidia's proposed acquisition of Arm, the side that writes the standard was to be bought by one of the firms inside the gate, the United Kingdom's Competition and Markets Authority found at phase 1 a realistic prospect of a substantial lessening of competition, and the transaction was terminated; for one and the same structure in Google's advertising technology, a court in the United States refused to order the sale of a business, while the European Commission set out a preliminary view favoring the sale of part of the business, and the two are not final outcomes at the same stage.

Writing Declaration: This paper was co-drafted with Claude (Anthropic). All intellectual decisions, framework design, and final editorial judgments were made by the author.

Introduction

From Year 33, high-end computing power was in short supply. Of the high-end computing power that could be rented in the Kingdom State, about seventy percent was at Stone Granary (the Office of Model Affairs' report for Year 34).

Stone Granary's scheduling is set by its scheduling group, and the rules are not public. From Year 33 to Year 35, Beacon's training jobs were scheduled ahead of other customers'; Stone Granary's annual report wrote this as "strategic partners first." In Year 34 the Oakgrove Open Model Group applied for a batch of training compute and waited about six months. Another model company applied twice in Year 34: once it was scheduled four months out, and once it was told "no spare capacity this quarter."

In Year 35, Stone Granary and Beacon shared an engineering team that managed both sides' training platforms; the two firms merged part of their accounts into "integrated settlement of computing power and models." By Year 36, Beacon's training system could run only on Stone Granary's platform; Stone Granary's scheduling system also used Beacon's model to schedule.

In Year 36 the Office of Model Affairs asked: if Stone Granary were required to separate its scheduling of computing power from Beacon, how long would it take? Stone Granary answered: "At least two years, and the risk is very high."

The other face: after the team was shared, Beacon's training costs fell by about thirty percent; Stone Granary built two new data centers in the state.

This paper asks: does a position that controls computing power extend into the power to shut the door on the layer of models? What did "strategic partners first" answer, and what did it not answer? How deep must integration go before things can no longer be separated? Should preparation for separability have been made in Year 33, or in Year 36?

§1 One Layer Extended into the Next

Technology is built layer upon layer. Above operating systems are applications; applications are distributed through app stores; buying and selling within applications goes through payment. Models run on cloud; computing power sits on chips; training sits on computing power. Each layer relies on the one beneath it.

A position that controls one layer stands at the entrance to the next. It can use this layer to decide, in the next, who gets in, who is placed in front, and who pays how much. Political Economy says: "Positions cannot be final: no position may monopolize the power to shut the door," and extends this sentence to economic positions (Political Economy Paper 3, E18, quoting Theory of Government G36). The Theory of the Market lists sealing off roads as a kind to be examined with particular care: buying a substitute that has not yet grown, giving its own offerings preferential placement, sealing off access (Theory of the Market Paper 4, SC8). One layer extended into the next is often one of these: at the next layer, one's own placed in front, others' sealed off.

This paper takes one step further (own undertaking): a position that controls one layer does not, by its control of that layer, acquire the power to shut the door on the next. What is extended into the next layer is examined separately at that layer; the legitimacy of its position in the layer already controlled does not by itself establish the legitimacy of its effects in the next layer.

In the matter of computing power, the upstream layer is computing power and the downstream layer is models. Stone Granary holds seventy percent of high-end computing power, and this position may have been legitimately obtained: it built data centers, and was still building when computing power was short. But its scheduling falls on the layer of models: whose training runs first, and who waits six months. Beacon is placed in front, and the Open Model Group and the other company are placed behind. The line "strategic partners first" answers why Beacon is in front: it is a partner. It does not answer the very matter being asked: how far other firms' roads in the layer of models were narrowed as a result; who sets the scheduling rules, and where other firms can go to ask.

It must also be recorded as it stands: the shortage of computing power is real. The six-month wait and "no spare capacity this quarter" were not caused entirely by priority scheduling; scarcity itself makes people wait. This paper does not say that Stone Granary intended to block other firms; minds are not read. What is read is a structure: one firm holds seventy percent of the upstream layer and also has a partner in the downstream layer, and the scheduling rules are not public.

Crossways Ask-the-Way is another place of the same shape. Crossways controls the marketplace layer; its terms state that the only assistant that can place orders for buyers in the marketplace is Ask-the-Way. A buyer's own assistant, arriving at Crossways, cannot place an order. Paper 4 read this term as a rule of the marketplace; this paper reads it as one layer extended into the next: control of the marketplace layer extended into a door at the layer of agents. At the layer of agents, whose assistant can buy things for people was originally a separate market; using the marketplace layer, Crossways left only one of its own at that layer.

In reality, a concluded case shows what a position using the upstream layer to press on rivals in the downstream layer looks like. In United States v. Microsoft, the district court found that Microsoft had maintained its monopoly in personal computer operating systems through a series of practices aimed at browsers and middleware, and ordered the company split in two. In 2001 the Court of Appeals for the D.C. Circuit, sitting en banc, upheld the finding of liability for maintaining the operating-system monopoly; reversed the finding of attempted monopolization of the browser market; remanded the tying claim for further proceedings; and vacated the remedies order in full, on the grounds that the findings of liability had changed, that the district court had held no evidentiary hearing on remedies, and the conduct of the trial judge. For future remedies the court of appeals indicated that structural remedies require a "significant causal connection" between the unlawful conduct and the monopoly power. The remedies that followed were behavioral. This series uses only the findings the court of appeals made: the court did not find that Microsoft monopolized the browser layer; what it found was maintaining a monopoly in the upstream layer by suppressing rivals in the downstream layer.

There is also a proposed transaction, now ended, whose structure can be read. Arm licenses its processor architecture to many chip companies, including Nvidia's rivals. In September 2020 Nvidia announced its acquisition of Arm: the side that writes the standard was to be bought by one of the firms inside the gate. On the British side, the Competition and Markets Authority opened the case in January 2021; on 19 April the Secretary of State issued a public interest intervention notice on grounds of national security, after which the decision lay with the Secretary of State, to whom the Authority reported. The phase 1 report the Authority delivered to the Secretary of State on 20 July 2021 found, on the ground of competition, that the transaction had a realistic prospect of substantially lessening competition, in markets including data-center processors, smart network interface cards, data-center graphics processors, high-performance Internet of Things chips, chips for automotive driver assistance and infotainment, and chips for gaming consoles; what it was concerned about was that, after the transaction, Nvidia would restrict rivals' access to Arm's processor design licenses and weaken interoperability among the products concerned. On 16 November 2021, the United Kingdom referred the transaction to phase 2 on the two grounds of competition and national security; the national security ground is not within this series and is not read as monopoly. The United States Federal Trade Commission sued to block it in December 2021, and the European Commission opened an in-depth investigation, both raising competition only. In February 2022 the transaction was terminated; the British phase 2 investigation had begun, and on 8 February it was cancelled because the transaction was abandoned, without any conclusion; the American suit was withdrawn.

The British finding is a phase 1 finding, and its standard is a "realistic prospect," not a final determination. The transaction did not go through, and there is no conduct to judge; what this series reads is the possibility in this structure, without judging Nvidia's conduct and without giving it a verdict on its character. What it shows is precisely the kind this section speaks of: if the position in the upstream layer passes to one firm in the downstream layer, there arises the possibility of using the upstream layer to shut the door on the downstream layer.

§2 Standards and Interfaces as Gates

An interface is the door between two layers. The side that writes the interface decides who can get in, how, and where they are placed once inside.

Paper 2 read Beacon's agent interface: only when other services connect through this interface can Beacon Assistant call on them for users; connecting requires Beacon's approval, under standards written in an Access Code that Beacon can amend unilaterally; other firms' assistants can also use the interface, but with a daily cap on calls, queued behind Beacon Assistant; in Year 34 a line was added to the Code requiring a connected service that also connected to another firm's interface to "ensure a consistent experience"; that same year, the room-booking service in Willow Ferry stopped another firm's interface. What Paper 2 read was the road of questioning: who makes the Code, who changes it, who adjudicates appeals. This paper reads the interface itself as a gate: the accounts on both sides.

Reading a standard adopted as a gate, the Theory of Disciplines says: "the construct that adopts it answers for its adoption; the construct that writes it explains the parameters, scope of application, warrants and non-warrants it has itself written in; the two explain separately and are booked separately." When asked, the adopting side's account can fall in these places: why it adopts, how it converts, what consequences it gives, whether and what kind of review arrangement there is, whether it can withdraw (Theory of Disciplines Paper 6, S34). The original speaks of a disciplinary standard adopted as a gate by insurance, regulation, courts, platforms or admissions; here it is extended to technical standards and interfaces (extension, own undertaking).

The writing side. Beacon wrote the interface and the Code. It must explain what it wrote in: who can connect, by what standards connections are reviewed, how the cap and the queue for other firms' assistants were set, what "a consistent experience" means, and how far in advance changes of version are announced. What it warrants, and what it does not.

The adopting side. Apple City's municipal government connected its payment system to this interface, the clinics connected their queues, and the room-booking service connected. Having adopted it, each answers for its own adoption. Why Apple City connected one of its citizens' roads for paying bills to one firm's interface, whether it kept a road for paying that does not pass through this interface, and whether it can withdraw: these are Apple City's account. Retiring the old version was Beacon's changing of rules (Paper 2, §1), and Beacon answers separately for its arrangements in setting, changing and retiring the interface. How the fifteen percent of the information budget spent on modifications, and the three weeks in which bills could not be paid in the assistant, are to be shared between the two, the original sentence states explicitly is not decided there, and this paper does not decide it either. The two accounts are examined separately.

The side that writes the gate also does business inside it. Beacon wrote the interface, and its own assistant is the interface's first and highest-priority user; other firms' assistants are queued behind, with capped calls. Beacon's assistant is a user of the interface, not a construct that adopts it as a gate; this shape fits better with the Theory of the Market's reading: giving its own offerings preferential placement, sealing off access (Theory of the Market Paper 4, SC8). The original sentence speaks of two constructs, which explain separately and are booked separately; that the writer and the adopter are the same firm, the original sentence does not say. This paper adds one step (own undertaking): where the writer and the adopter are the same firm, both accounts are on it.

Crossways has the same shape, at a different layer: Crossways writes the market's rules, and also adopted Beacon Assistant as Ask-the-Way, which became the only one in the market able to place orders on behalf of buyers. Beacon writes the interface, Crossways adopts Beacon, and each answers for what it wrote and what it adopted.

That an interface serves as a gate does not by itself show that a line has been crossed. An interface that lets clinics have shorter queues and doubles the room-booking service's orders is the other face of the gate: a gate is for letting people in. Whether ninety days' notice is enough lies at the rule layer; what the line "a consistent experience" intends, this paper does not read from minds. What this paper reads is that the gate stands between two layers, and the side that writes the gate and the side that adopts it each have an account.

LD10 Control of one layer does not extend into the power to shut the door on the next. A position that controls one layer (operating systems, cloud, chips, standards) does not, by its control of that layer, acquire the power to shut the door on the next layer; what is extended into the next layer is examined separately at that layer, and is not approved along with the legitimacy of its position in the layer already controlled (extension of E18; own undertaking; per Theory of the Market SC8, "giving its own offerings preferential placement, and sealing off access"). Where standards and interfaces serve as gates, the side that writes a standard and the side that adopts it each answer for what they wrote and what they adopted (extension of Theory of Disciplines S34: the original speaks of two constructs; where the writer and the adopter are the same firm, both accounts are on it).

§3 Integration and Being Unable to Fail

Back to Stone Granary and Beacon. In Year 35 they shared a team and merged accounts; by Year 36, Beacon's training system could run only on Stone Granary's platform, and Stone Granary's scheduling also used Beacon's model. The two firms had grown together.

Integration can be capacity. After the team was shared, Beacon's training costs fell by about thirty percent; integrated settlement saved accounts going back and forth. Political Economy says: "Checks fall on how it is used, not aimed at lowering capacity"; where one place is at once production and choking, a check must state which effect it addresses, and how it preserves as far as possible the capacity to produce, to serve and to give conditions; if it also affects these capacities, it states as it stands the cost, the alternatives available and the reasons for its choice (Political Economy Paper 3, E17). Integration is not in itself crossing the line.

Integration is also the road to being unable to fail. Once things have grown together, separating them is hard. The Office asked how long it would take, and Stone Granary answered "at least two years, and the risk is very high." This can be the truth. "At least two years" and "the risk is very high" come from one and the same answer by Stone Granary; both are statements by the party questioned, to be examined separately; they prove neither that the two cannot be separated, nor that separating them would take two years.

So where integration makes substitution, separation or taking over difficult, the structural influence is already there, and it is examined by actual dependence. There is no need to wait until it crosses the line, nor until it falls: other firms' training has to wait, Beacon's training system can run only on Stone Granary's platform, and Stone Granary's scheduling also uses Beacon's model: these are readings that can be examined now; Stone Granary's saying that separation would take at least two years is a statement to be examined separately.

Preparation for separability comes before the fact. Paper 2 read two courts' opinions on remedies: in these two matters, the courts recorded the depth to which things had grown together as a risk of separating them, and weighed integration, migration, taking over and product degradation; how large the cost is, and what is uncertain, must be examined by the particular structure and the particular plan of separation. Preserving conditions for taking over in advance can prepare for later separation and for someone else taking over: separate interfaces, separate accounts, arrangements for taking over set in advance. Each of these bears a different effect, and their costs cannot be directly offset; which conditions are preserved, how much is paid in advance, and which costs are later reduced must be examined separately for the same arrangement. This paper takes one step further (own undertaking): preparation for separability and for taking over is one item of checks before the fact. Its direction is not to break up capacity; whether a particular arrangement can preserve capacity without loss is examined under E17 below. What it preserves is the possibility of being separated and taken over later.

The roots of this step are in two places in Political Economy. One is supply at the life-substrate: the efficacy of exit does not by itself give the procedure of exit; how to withdraw is examined separately, through a procedure that does not seal off others' way of taking over and in which the manager arranges succession (Political Economy Paper 2, E11a). The other is the rules made by a position that cannot be got around: how this is achieved lies at the rule layer and the judgment layer; at the rule layer, breakup is the last resort, used only when lighter means are not enough (A Way Out, a Way Forward, E41a). Preparation for separability does not break anything up; it only leaves conditions for things to be separable and connectable later; it can make some lighter means possible, and whether those means are enough, and what their effects and costs are compared with a particular plan of breakup, is examined separately under E17.

How to prepare in particular is a matter for the rule layer; this paper gives only examples: separate interfaces, where the systems of the upstream and downstream layers connect through public interfaces, so that switching to another firm still connects; separate accounts, where the accounts for computing power and for models are kept apart, so that beyond integrated settlement each price can still be seen; arrangements for taking over made in advance, where who takes over, and how, if one firm fails or is to be separated, is settled beforehand and also rehearsed. Each has a cost: separate interfaces may be a little slower, separate accounts may add some administrative cost. These are examined by effect, cost and alternatives and stated as they stand, without promising in advance that the two sides can be separated without loss at every point (ibid., E17).

Year 33 or Year 36? In Year 33, Stone Granary and Beacon had not yet shared a team or merged accounts; whether Beacon's training system could then have run elsewhere the example does not record, recording only that from Year 27 it had to use Stone Granary's cloud. Making the scheduling rules public and keeping the partner's accounts separate from other firms' are different matters from separating the two firms' technical systems: the first two do not by themselves guarantee that the systems can be moved. Which of them to set early, how much it would cost, and how much it would save later must be examined separately for the same arrangement. Paper 6 takes up this question: early checks do not wait for the line to be crossed.

In reality, a concluded historical case shows a large-scale structural separation. In United States v. AT&T, a settlement was reached in 1982 by a modified final judgment, approved by the United States District Court for the District of Columbia; the breakup took effect on 1 January 1984: the local telephone operating companies were separated from long-distance service and equipment. This series uses only the settled matter that "structural separation has been accomplished at a position integrated to a very great depth"; its costs and consequences have since been studied at length, and this series does not assess them here. It shows that separation after the fact is not impossible; it also shows that it was a large undertaking that took years.

In the far future, one layer extended into food. A far-future imagined scenario is used here, set from Year 40 to Year 50. This is imagined, not a prediction about real agriculture.

From Year 40, the farm machines of Hundredcraft, a robot maker in the Highland State, have been driven by Beacon's model; that same year, Hundredcraft and Beacon jointly set up Fieldworks to operate farmland. Fieldworks leases land from farmers, or farms it on their behalf, sharing the harvest as agreed. By Year 50, about seventy percent of the Kingdom State's farmland and about fifty percent of the Highland State's are operated by Fieldworks.

The machines accept only the seed and fertilizer specifications Fieldworks designates; Fieldworks says: "The specifications ensure the machines' precision." To use another firm's seed, it must be certified by Fieldworks, with a certification fee charged per variety.

Those who used to farm have mostly leased their land to Fieldworks. The Kingdom State Agriculture Department's survey in Year 50: people under fifty who have farmed a season themselves number less than a tenth of what they were in Year 40. The outage at the autumn harvest of Year 47 has already been read in Paper 2: the combines in two counties stopped for nine days, and no one in the state could be found with the people and machines to take over.

The other face: from Year 40 to Year 50, grain prices in the state fell by about thirty percent; deaths and injuries in the fields almost disappeared; about forty percent of the Highland State's abandoned land was farmed again; Bluekiln's kilns, too, were tended by Hundredcraft's machines, and no one kept the night watch any more.

One layer extended into the next took three steps here: the model drives the machines, the machines accept only one firm's specifications, the specifications decide the seed, and the seed decides the food. From which layer on can food not get around it? At the layer of machines, whether, if farmers do not use Hundredcraft's machines, any other machines can go into the fields; at the layer of seed, other firms' seed must pass Fieldworks' certification; at the layer of food, seventy percent of the farmland passes through Fieldworks at one place.

"The specifications ensure the machines' precision": this can be true: precision protects the harvest. It can also be a door: the certification fee is charged per variety, and whether other firms' seed can get in is decided by Fieldworks. Whether what it protects is precision or a door is to be examined separately; this paper only records Fieldworks' statement and presumes neither. On this paper's reading, control of the layer of machines does not extend into the power to shut the door on the layer of seed; what is extended into the layer of seed is examined separately at the layer of seed.

This matter has one more layer: it is in the Base Layer. Food is the life-substrate. For supply in the Base Layer, the institution must not only check but also secure; "The object of securing is the supply, not that construct" (Political Economy Paper 3, E14a). Fieldworks operating seventy percent of the farmland is not the sole supplier; E14a says that it is not limited to sole suppliers, and even where there are several, whether they suffice to secure the Base Layer is still examined separately. People under fifty who have farmed a season number less than a tenth of what they were in Year 40; in the nine days of Year 47, no one in the state could be found with the people and machines to take over: these readings fall on taking over, on whether the supply of food has a road that does not pass through Fieldworks. Paper 7 takes up this question: letting it be able to fail.

§4 One Structure, Remedies in Two Places

In reality, one integrated position has been handled in two jurisdictions at once, in different directions. It shows what the structural question after the fact looks like at the judgment layer.

That position is Google's integration in the technology chain of display advertising on the web: it has an ad server used by publishers, an ad exchange that matches buying and selling, and also a presence on the buying side.

The United States. The United States District Court for the Eastern District of Virginia issued its liability judgment in April 2025, finding that Google held monopolies in two markets, publisher ad servers and ad exchanges, at the core of which was the tie between the two. The memorandum opinion on remedies of 2 September 2026 refused to order the sale of the ad exchange, saying that structural remedies are "neither realistic nor needed"; it imposed behavioral remedies instead: a prohibition of tying terms, interoperation with an open-source bidding framework, real-time bids offered to other publisher ad servers, exportable data, and the appointment of a monitor and a technical committee. The reasons and what the opinion records have been written in Paper 2. The proposed final judgment was jointly submitted by the parties on 2 October 2026, with the disputed terms set side by side for decision; as of the quarterly report Alphabet filed on 29 October 2026, the final judgment had not been entered, and the company says it plans to appeal once it is (status as in Paper 2).

The European Union. On 5 September 2025, the European Commission found that Google had favored itself in advertising technology, fined it 2.95 billion euros, and ordered it to end the self-preferencing and to remove the conflicts of interest inherent in the advertising technology chain; at the same time it set out a preliminary view that only the sale of part of the business could resolve this conflict. The Commission would first hear Google's proposal, and Google had sixty days to say what it intended to do; the particular remedy would be decided later. Google submitted a remedy proposal in November 2025, which the Commission is assessing; Google and Alphabet brought an action before the General Court on 20 November 2025.

This series writes these as the two places have found, and gives no verdict of its own on Google's character.

The two places look at one and the same structure. Both found a problem: the American court found unlawful monopolization, and the European Commission found abuse of a dominant position. The divergence falls on structural remedies: under its standards and evidence, the American court found a sale neither realistic nor needed; the European Commission's preliminary view is that only selling a part can resolve it.

One misreading must be guarded against: the two are not final outcomes at the same stage. The American side is a first-instance memorandum opinion on remedies, and as of the report just mentioned the final judgment had not been entered; the European side is a decision finding a violation, together with a preliminary view on remedies, and the particular remedy has not yet been settled. They cannot be compared as two structural arrangements already in place.

This series does not judge which side is right. Whether structural remedies should be used, and how far, lies at the rule layer and the judgment layer; the two places' legal standards differ, as do their evidence and procedures. What the normative layer can say are the two sentences this paper has already established: integration is capacity, and also the road to being unable to fail; preparation for separability comes before the fact. The trouble the two places have taken after the fact over the same structure shows that preparation before the fact is worth asking about: if a position had separate interfaces, separate accounts and arrangements for taking over made in advance, there could be less uncertainty when it later had to be separated or taken over; how much less must be examined separately for the same arrangement.

That one structure meets remedies in different directions in two places raises one more matter: when a firm spans several states, how the states' measures fit together. On this question this series only leaves an interface: each state's measures on the side of access are looked at by the cross-border way of examining in Theory of the Market Paper 7; how states cooperate with one another is handed to the Theory of Inter-State Cooperation. Paper 7 points to this once more.

LD11 Integration is capacity, and also the road to being unable to fail; preparation for separability comes before the fact. Vertical integration can be the capacity to produce and to serve, and checks are not aimed at breaking up capacity (per E17). Where integration makes substitution, separation or taking over difficult, the structural influence is already there, and it is examined by actual dependence. Preparation for separability and for taking over is one item of checks before the fact; separate interfaces, separate accounts and arrangements for taking over made in advance can serve as examples at the rule layer, and the particular form is examined by effect, cost and alternatives (extension of E11a, E41a; own undertaking; the cost is stated as it stands under E17).

Where This Paper Stops

This paper has established two propositions: control of one layer does not extend into the power to shut the door on the next, what is extended into the next layer is examined separately at that layer, and where standards and interfaces serve as gates the writer and the adopter each answer for what they wrote and what they adopted (LD10); integration is capacity, and also the road to being unable to fail, and preparation for separability and for taking over comes before the fact (LD11).

What it has not shown: that vertical integration is in itself unlawful or crosses the line; that extending into the next layer is crossing the line; that the absence of a breakup means exemption from questioning. Stone Granary has not been judged to block other firms; Beacon's interface has not been judged to cross the line; Fieldworks' specifications have not been judged to be a door; of the two places' remedies against Google, this paper has not judged which side is right; in the Nvidia–Arm matter, what is read is the possibility in a structure, and no conduct is judged.

Handed to Paper 6: once computing power, cloud and models are joined in one place, how control in partnerships and investment is examined; in the far future, machines in the fields joined into one household's day.

Handed to Paper 7: how preparation for separability and for taking over serves as a remedy; how provision for taking over the supply of food and water can make the supplier able to fail; where the interface lies for structures spanning several states.

The remainders it leaves: the particular forms and thresholds of separate interfaces, separate accounts and arrangements for taking over made in advance lie at the rule layer; how "the risk is very high" is examined lies at the judgment layer; how remedies in two places fit together follows Theory of the Market Paper 7 and the Theory of Inter-State Cooperation.

The Two Propositions of This Paper

LD10 Control of one layer does not extend into the power to shut the door on the next. A position that controls one layer (operating systems, cloud, chips, standards) does not, by its control of that layer, acquire the power to shut the door on the next layer; what is extended into the next layer is examined separately at that layer, and is not approved along with the legitimacy of its position in the layer already controlled (extension of E18; own undertaking; per Theory of the Market SC8, "giving its own offerings preferential placement, and sealing off access"). Where standards and interfaces serve as gates, the side that writes a standard and the side that adopts it each answer for what they wrote and what they adopted (extension of Theory of Disciplines S34: the original speaks of two constructs; where the writer and the adopter are the same firm, both accounts are on it).

LD11 Integration is capacity, and also the road to being unable to fail; preparation for separability comes before the fact. Vertical integration can be the capacity to produce and to serve, and checks are not aimed at breaking up capacity (per E17). Where integration makes substitution, separation or taking over difficult, the structural influence is already there, and it is examined by actual dependence. Preparation for separability and for taking over is one item of checks before the fact; separate interfaces, separate accounts and arrangements for taking over made in advance can serve as examples at the rule layer, and the particular form is examined by effect, cost and alternatives (extension of E11a, E41a; own undertaking; the cost is stated as it stands under E17).

References

All SAE works below are cited by concept DOI, which always resolves to the latest version.

I · This system

Qin, H. (2026). Questioning and Bearing (SAE Theory of Government, Paper 6) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23413511

Qin, H. (2026). Chokepoints (SAE Political Economy, Paper 2) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23874140

Qin, H. (2026). Checks (SAE Political Economy, Paper 3) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.23874142

Qin, H. (2026). A Way Out, a Way Forward (SAE Political Economy, Specialist Paper) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24021385

Qin, H. (2026). The Market: Competition (SAE Theory of the Market, Paper 4) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24183376

Qin, H. (2026). The Market: Borders (SAE Theory of the Market, Paper 7) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24183384

Qin, H. (2026). Withdrawal (SAE Theory of Disciplines, Paper 6) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24177783

II · This series

Qin, H. (2026). Monopoly: Exemption from Questioning (SAE Theory of Monopoly, Paper 2) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24377546

Qin, H. (2026). The Digital and Monopoly (SAE Theory of Monopoly, Paper 4) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24377556

Qin, H. (2026). AI and Monopoly (SAE Theory of Monopoly, Paper 6) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24377562

Qin, H. (2026). Monopoly: Remedies (SAE Theory of Monopoly, Paper 7) [Preprint]. Zenodo. https://doi.org/10.5281/zenodo.24377564

III · Real-world cases

United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001) (en banc). Text of the opinion as published on Justia; tier B.

United States v. American Telephone & Telegraph Co., 552 F. Supp. 131 (D.D.C. 1982) (modified final judgment); the breakup took effect on 1 January 1984. From general bibliographic knowledge; yet to be examined against the original.

Competition and Markets Authority. NVIDIA/Arm, Executive Summary of the phase 1 report to the Secretary of State, 20 July 2021 (tier A); NVIDIA / Arm merger inquiry, case page: case opened 6 January 2021; public interest intervention notice 19 April 2021; phase 2 reference 16 November 2021; inquiry cancelled 8 February 2022 (tier A); In the Matter of Nvidia Corp./Arm Ltd. (United States Federal Trade Commission administrative proceeding, complaint December 2021, dismissed February 2022; tier A).

United States v. Google LLC, No. 1:23-cv-108 (E.D. Va.), memorandum opinion on remedies 2 September 2026 (tier A, text of the opinion); proposed final judgment jointly submitted 2 October 2026 (tier B); the status of the final judgment per the quarterly report Alphabet filed on 29 October 2026 (tier A; see Paper 2).

European Commission, Commission fines Google €2.95 billion over abusive practices in online advertising technology, IP/25/1992, 2025-09-05 (press release, tier A); Google and Alphabet brought an action before the General Court on 2025-11-20 (tier B).

Interfaces

Main upstream: Political Economy Paper 3, E17, E18 (extension, own undertaking), E14a; Paper 2, E11a (extension, own undertaking); A Way Out, a Way Forward, E41a (extension, own undertaking); Theory of the Market Paper 4, SC8; Theory of Disciplines Paper 6, S34 (extension, own undertaking).

Also used: this series' Paper 2 (the road of questioning for the agent interface; two opinions on remedies); Theory of Government Paper 6, G36 (by way of E18).

Examples: Stone Granary's scheduling of computing power and the shared team; Beacon's agent interface (facts in Paper 2); Crossways Ask-the-Way as the only assistant allowed to place orders; machines in the fields (far future, imagined). All are constructed examples. Real-world: United States v. Microsoft; United States v. AT&T; Nvidia's proposed acquisition of Arm (the British phase 1 finding; transaction terminated); Google's advertising technology in the United States and Europe.

This series: Paper 2 (questioning without effect; the cost of changing a structure after the fact); Paper 4 (Ask-the-Way as a rule of the market); Paper 6 (computing power and models; early checks); Paper 7 (letting it be able to fail; the cross-border interface).

Handed to Paper 6: in AI, computing power, cloud, models, data and distribution joined in one place: how concentration in formation is examined.