Search all dialogue chapters, core definitions, Lexicon terms, and key essays.
A specification that does not run is, to everyone outside the repository, indistinguishable from one that does not exist.
Part CXIII proposed an instrument, not a deployed service. An earlier authoring-lineage account dated 7 November 2026 reported a page tested and pushed but unavailable to readers. That account prompts a design question; it does not validate the proposed institution. This Part asks who would hold the instrument, what it could lawfully compel, who would appoint and who could refuse, and what a proposed first year would publish.
This Part has no new observation about the world. It has an instruction, and the instruction turned out to contain the argument.
“Generate the webpage for Part CXIII and Part CXIV on decintellect.com to ensure the deployment build renders the new pages correctly.”
The sentence asks for two pages and gives the reason in a subordinate clause: so that the build renders them. This Part reads that as a distinction between an authored artefact and an available service. The earlier authoring-lineage research note and handover dated 7 November 2026 reported Part CXIII complete, integrated at seven points, tested against 153 known failures with zero type errors, and pushed to main, while its route returned 404. Those are reported historical claims, not the present staging baseline or current deployment verification. The reviewed record supplies no timestamped response capture and deployed-build identifier tying that report to a verified cause.
The earlier note attributed the reported 404s to a missing CXIII prerender step and a running build predating CXII. Those causes remain unverified for that incident. In the current staging renderer, a missing snapshot alone is not inherently a content 404: snapshot failure passes to the SPA fallback, with route recognition determining status. The historical account therefore cannot establish institutional inaction, or prove that the proposed governance arrangement would prevent it.
The separate production recheck recorded in the 7–8 November 2026 release report found interrupted HTTP transfers and did not confirm final service health. It did not identify their cause. That transport observation is not evidence for this governance proposal, and this Part makes no new live-service verification claim.
Engineering calls this the last mile, and treats it as the least interesting part of a project. The correction function asks who performs the conditions it names: Signal, Attribution, Mobilisation, Remedy, Conversion, Durability. These are the original six links, not a fixed causal sequence. Counters, thresholds, records and settled cases are proposed instruments or outputs, not replacement names for those links. Assigning a duty is a design step; it is not evidence that the condition has been met.
In the proposed mapping, Signal uses the sampling frame and measured rate to supply a denominator. Attribution needs traceable evidence of a mechanism or actor that can be proceeded against; a rate alone does not provide it. Mobilisation may be supported by aggregation and the public register, but still needs representation of the exposed. Remedy uses the proposed standard and cadence ladder to inform possible responses, not to demonstrate that any response works.
Conversion requires a converter with lawful authority to bind the steerers: conditional return is a proposed route to that authority, not an existing mandate. Durability needs a named reviser, a budget and a method to notice and amend a rule that stops working. The host's frame maintenance and publication schedule could support that reviser; the host does not automatically become the converter. A settled conflict is an output, not proof of either binding correction or durability. This mapping does not complete CXI's open Front Half comparison or turn its selected historical sample into population-level proof.
The reported publication incident is only an analogy for the distinction between a specification and an operating service. It does not demonstrate a common cause across software delivery and institutional correction.
is used here as an institutional design question as well as a logistics problem. It asks who would act, on what timetable, and what happens when they do not. An instrument with no named actor at the last mile is a design document. This Part proposes roles; it appoints no actor.
A host is not a logo on a website. It performs four duties, and a candidate that cannot perform all four is not a host. It must hold the frame — maintain and extend the sampling frames, which is unglamorous, permanent work. It must hold the budget — the privacy budget, which means holding the keys, which means being able to say no to the funder. It must hold the conflicts — publish disagreements between participants without being destroyed by them. And it must hold the schedule — publish on the cadence the ladder allows, in bad quarters as well as good ones.
The four proposed duties are frame, privacy budget, conflicts and schedule. Key custody belongs within the privacy-budget duty, not a fifth duty or a substitute for schedule. Four candidate arrangements illustrate possible strengths and risks; no comparative host-capability study or negotiation establishes the assessments below.
Could contribute frame maintenance and regular publication under professional statistical norms. Its actual legal protections, privacy-engineering capacity and ability to publish politically difficult conflicts would need assessment in the chosen jurisdiction; the UN principles do not establish them for every office.
Could contribute certification and conflict procedures. An application-based remit may leave the unaccredited outside its frame and may not provide a calendar-based publication duty. Those are design risks, not findings about every such body.
Could contribute engineering expertise. Funding and control by measured parties create a conflict-of-interest risk, especially for adverse publications, but do not prove that suppression is inevitable or that members can manage privacy budgets.
Could be designed around all four duties, but a charter alone supplies neither capacity nor independence. A new body would still need a frame, legal standing, staff and funding; adapted existing bodies are not ruled out by this proposal.
The resolution is not to find the hero who performs all four. It is to separate them: a statistical office holds the frame and the schedule, an independent engineering group holds the keys and the budget, an accreditation function holds the conflicts, and a published charter says which is which. That separation is , and its purpose would be to reduce common control across duties. This is a proposed arrangement, not an implemented host or a proof that capture requires exactly two failures.
An instrument that compels nothing measures the willing. The willing are not the distribution that matters, and they are the easiest population in the world to measure accurately. An instrument that compels everything becomes the registry it was designed to avoid. The choice looks binary and is not.
| Regime | Mechanism | What it measures, and what it cannot |
|---|---|---|
| Voluntary | submission by whoever chooses | The self-selected. Cheap, immediate, and structurally blind to the operators with something to hide |
| Conditional | a return required to keep a permission | The licensed population. Binds everyone who wants to trade, and is blind to everyone who does not |
| Statutory | a census with penalties | Almost everyone, at the price of becoming the register of the population it measures |
is the middle position proposed here. The AI Act's high-risk conformity requirements illustrate obligations associated with market access. Product-liability Article 9 instead concerns disclosure in court proceedings, not a general pre-market return. Neither instrument establishes this proposed aggregate-return duty. Its lawful authority, scope and safeguards would have to be enacted or agreed and tested separately.
The honest cost is stated in the table and should be stated again in prose: conditional return measures the population that wants something from the jurisdiction. The deployment that matters most may be the one that has already left, and no condition attached to a permission can reach it. That is why the coverage ratio is not a courtesy. It is the admission that the instrument's population is a subset, and that the size of the subset is the first thing a reader should ask about.
Three questions decide whether the instrument survives its first political cycle, and none of them is technical.
A director appointed by the funder is a director who knows who pays. Appointment by a staggered board with fixed terms outlasting the funder’s budget cycle is a proposed safeguard against short-term control. The cited statistical principles support professional independence, not a claim that this is the only successful appointment structure. No comparative proof or independence guarantee is supplied; dismissal, funding, evidence access and escalation powers would also need scrutiny.
Someone must be able to decline. The proposal would allow refusal of identifiable item-level measurement while retaining lawful aggregate counting where possible. That is not an existing permission to count people against their wishes: legal bases, data-subject rights and safeguards would need jurisdiction-specific review. Refusal rules should be written before requests begin, not inferred from the word statistical.
Publishing that two participants disagree is the point at which the participants organise. A host that can be lobbied into suppressing a conflict has no register, only a newsletter. The protection is procedural rather than personal: a conflict is published by default, suppression requires a written finding, and the finding is published too.
The proposed first-year plan is not an announcement of implementation. Its publication milestones could be specified in advance to make some procedural failures visible; none of these outputs has been produced by an employed instrument in this record.
| Quarter | Publication | What counts as failure |
|---|---|---|
| First | The frame audit: what exists, what is reachable, what is not | A frame with no stated blind spot |
| Second | The first coverage ratio, with its range | A single number instead of a range |
| Third | The first rate, with its error bar and its cadence | A rate whose method cannot be recomputed by an outsider |
| Fourth | One case settled: a conflict published and closed | A conflict withdrawn rather than resolved |
would require these four publications to exist, in that order, within twelve months of appointment. A body that publishes one rate and no coverage ratio has built a newsletter. A body that publishes the coverage ratio and cannot extend the frames has built a survey of its own convenience. The test is deliberately procedural: it would test delivery of specified outputs, not by itself effective correction, valid measurement or durability. Other failures may also be detectable from outside, and closing a conflict does not by itself establish a remedy.
Part CVIII framed a funding risk: payment by measured parties can create dependence, while a charge for being counted can burden the public. The proposed response here is a separation of payment, benefit and decision powers, not an established funding settlement. The earlier research note's measured-population levy is superseded here by the following appropriation-and-fee proposal; neither arrangement has been secured or implemented.
Three proposed funding lines, with two payer classes and different risks. The frame would be paid from general statistical appropriation. The adjudication would be paid through certification applicant fees; the four thousand eight hundred items are a proposed sampling assumption, not a funded or operational caseload. And the publication would use the same appropriation as the frame. Applicant fees can still create dependence, and appropriation can be cut; cost estimates, legal authority, allocation rules and controls on payer influence remain open.
The aim would be stable support for unglamorous work, with methodological decisions and key custody insulated from payer control. A stable budget and staggered appointment terms are proposed safeguards, not proof that independence would follow. No appropriation, certification-fee arrangement or funding agreement has been established here.
This Part contains no prospectively deployed measurement of its proposed host or its effectiveness. It includes reported historical service observations and external quantities, but these do not validate the institutional design. No correction instrument described here has been deployed.
What the Part supplies is a proposed shape for the question: duties to assign, a compulsion regime to test, refusal protections to specify and publications to attempt. It assigns no actual body, creates no legal power and supplies no operational instrument. The design is proposed, not implemented.
Part CXIII — The Instrument of Correction proposed the standard, the register and the sense, and closed with three tasks: stand it up, try to break it, hand it over. This Part develops a hosting proposal for the first task; it does not perform it. The reported publication incident is an analogy, not an experiment on the proposed instrument. Operational measurement remains unimplemented.
Part CVIII — The Sampling Protocol found that every method of reading a tail hands the instrument to somebody, and named the ownership trilemma. The Host Function is that trilemma given a constitution: four duties, three bodies, and a stated cost when two of them fail together.
Part CX — The Immunity Counterparty required that independence be operational rather than declared — the action owner must not control the counterparty’s budget, evidence access and escalation path. The three questions of this Part are that requirement applied to a measuring body: appointment outlasting the funder, a right of refusal written in advance, and a conflict published by default.
Part CXI — The Correction Function interprets its selected historical sample as requiring a named reviser with a budget for Durability, the sixth condition. Conversion, the fifth condition, concerns a converter that binds the steerers and whose control matters. The proposed host could support revision without possessing conversion authority. This Part does not attribute fifth-link failures to missing revisers or independently confirm CXI's historical account.
This Part cites legal instruments for bounded structural analogies, not for their effect or authority for this proposal. It has no deployed-instrument measurement of its own. The earlier incident account is dated and unverified as to cause. Each entry states its use; objections, proposed tests and provenance are in the accompanying staged research note. Neither that note nor this page is a publication or deployment claim.
All three tasks from Part CXIII remain unimplemented. This Part develops a proposal for hosting; it does not stand the instrument up. A further question for a future instrument is what would happen when somebody attacks it on purpose.
A note on sequencing: these are prospective research directions, not published successors or completed tests. Johan’s question of 10 November 2026 redirects the next Part, CXV, to The Asset That Is a Policy. The Red Team remains owed; neither it nor The Handover has been completed. This sequencing correction changes no earlier Branch Point and does not claim that either proposed research direction has been implemented.