TRACE Primitives

What the primitives are

The Cross-Border Narrative Friction Monitor reads every case through the same six observation dimensions. These dimensions are the primitives: the fixed vocabulary in which the research records how a company was described, classified, and reinterpreted by the market that received it. They are published so that a reader can check how any case was coded, and so that the method itself carries a version number that each publication can cite.

The primitives belong to TRACE (Testing Readings Against Comparable Evidence) as it is applied in public research. TRACE organizes its reading of a company in five layers: power relations, interpretation control, obstacle discrimination, supporting infrastructure, and intervention logic. The primitives are not one of those layers. They record what the public record shows before any of the layers is applied, and are established before any case is coded so that the same questions are asked of every case.

Everything the primitives record comes from public material, under the Public Evidence Charter. The research codes what a company published about itself, how others publicly redescribed it, and what visible actions followed. It does not code what anyone privately thought.

How the primitives relate to TRACE

TRACE reads a company’s position through several lenses, three of which the primitives touch directly. Each is named here in its public form, so that a reader can see which part of the method a dimension draws on and where the research stops.

Interpretation control runs through four ports: who defines the problem, who circulates the text, who forms the judgment, and who turns it into rules. Audience Frame Divergence observes the public traces of the middle two ports, the classifications that circulate in coverage and the judgments that appear in published commentary, and Governance Legibility and the outcome register observe the fourth, where a regulator’s published correspondence or a required amendment turns a reading into a rule.

A company can be doubted in three different ways—pragmatically, morally, or cognitively—and in several institutional environments at once. Identity Divergence and Origin Salience observe the visible surface of the cognitive case: whether the receiving market’s public record gives the company a settled place in its classification, or keeps returning to the question of what it is and where it is from.

A company expanding abroad answers two sets of legitimacy demands at once—home and host—and the two tracks do not merge. Evidence Portability observes how evidence built under the first set of demands is treated in public when it is presented to the second, and Governance Legibility observes whether a structure that was legible at home can be reconstructed by readers in the host market from what is publicly filed.

The six observation dimensions are also distinct from the six structural dimensions of stigma, which TRACE uses to describe the structure of a stigma once one has been established. The primitives record how a company was described and classified in the public record, whether or not any stigma is involved, and they do not decide that question.

The primitives stop at the public record. Which port is broken, which kind of doubt is at work, and how the two tracks should be reconciled are judgments that TRACE makes in advisory work with access to more than the public record. The research does not make them. It records the traces that such judgments would rest on, and it publishes them so that others can read the same record.

The unit of observation

A case is one company in one cross-border event, observed through the public record within a stated window and up to a stated cutoff date. Three layers of material feed each case: the company’s own filings, releases, and public statements; external descriptions in news, trade, and analyst coverage; and publicly visible institutional and transactional responses, such as regulator correspondence, amended filings, timetable changes, and pricing changes.

In its first year the Monitor applies the primitives to one event type in one receiving market: traditional underwritten initial public offerings by foreign operating issuers on United States exchanges. Direct listings, listings completed through a merger with a special purpose acquisition company, and companies already listed on another exchange before their United States debut are recorded in separate groups and are not pooled with the main set.

The six observation dimensions

Each dimension is defined once here. For each, the research looks at a fixed set of markers in the public record and records what appears there. Nothing about a company’s intentions, and nothing about what an audience privately concluded, is inferred from these markers.

1. Identity Divergence: the distance between the company’s own definition of what it is and the categories that external parties assign to it. The research compares how the company names its business, its sector, and its home in its own documents with the category words used in headlines, first paragraphs, and analyst summaries. It records whether the two match, where they part, and whether the gap widens or narrows across the window.

2. Origin Salience: the position that country of origin, source of capital, or founders’ background occupies in external explanations of the company. The research records whether origin appears in headlines and leads, whether it is used to explain facts that are not about origin, and whether coverage files the company under its country rather than its sector. It also records how the company itself places its origin in its own materials, so that the two placements can be compared.

3. Governance Legibility: whether ownership, control, data responsibility, board composition, and legal entity structure can be reconstructed by an outside reader from public material alone. The research records whether the public documents allow that reconstruction without outside knowledge, whether coverage or regulator correspondence returns repeatedly to the same structural question, and whether disclosure on these points was amended during the window.

4. Authority and Commitment Clarity: whether public statements make clear who can commit the company, who is responsible for delivery, and which powers the company and its partners each hold. The research records whether the same person or body is named as the speaker and the deliverer across documents, whether that attribution shifts, and whether a visible correction of responsibility appears in the record.

5. Evidence Portability: whether customers, regulatory recognition, and commercial evidence accumulated in the home market retain their force in the receiving market’s descriptions. The research records whether home-market evidence is cited, discounted, requalified, or omitted in external coverage, whether the company restates its evidence for the new audience, and whether regulator correspondence asks for substantiation of claims that rested on home-market standing.

6. Audience Frame Divergence: whether regulatory filings, financial press, trade press, and local press describe the same company as if it were different companies. The research places the classification used in each audience layer side by side and records whether the layers agree, where they diverge, and whether the divergence persists or converges over the window.

Observable outcomes

Alongside the six observation dimensions, the research keeps a separate register of visible outcomes. Seven outcome types are recorded: clarification request, amended disclosure, delay, governance change, price revision, public correction, and partner restatement.

An outcome is recorded only where the public record shows the action itself, with its date and source. Outcomes are not scored and are not folded into any dimension. The research may report that an outcome followed a coded pattern in time. It does not report that the pattern produced the outcome.

Coding rules

Version 0.2 fixes the dimensions, the levels, and the rules that follow. The operational anchors that place a case at one level rather than another are held in the codebook, which is built from the first cases and published with the first method note. Until then, each publication states the anchors it applied.

Each dimension is coded at one of three levels: low, medium, or high. The levels are ordinal. They indicate that one case shows more or less of a feature than another case on the same dimension, not how much more, and they are not added together.

Where the public record within the window contains no material bearing on a dimension, the dimension is recorded as not observable. Not observable is a distinct state from low. Absence of evidence is recorded as absence, never as a low reading.

Every code carries a pointer to the public material it rests on: the layer, the document or article, and the date. A code without a pointer is not entered.

The company’s own description is taken from fixed document versions so that cases are not compared at different stages of disclosure: the first public registration statement, the last registration statement before effectiveness, the final prospectus, and the first annual report where one exists. Where a quotation is used, the version it comes from is named.

Material published after the cutoff date is not coded, and a code is not revised on the strength of later material. Where later material matters, the case is recoded under a new cutoff and the change is logged.

No composite score

In the first year the Monitor publishes dimension-level codes and no composite score. A composite score would be introduced only when four conditions hold at the same time: the dimensions can be coded repeatably by more than one coder; the weights have a basis that can be explained; the events being scored are comparable; and the scores show a stable relationship with outcomes in the public record. Until then the research is a monitor, not an index.

What the primitives do not show

They describe how the public record described a company. They do not describe the company’s actual identity, the quality of its governance, or its conduct.

They do not establish cause. A high code on a dimension and an outcome in the register may appear together in one case. The research records the sequence and does not claim that one produced the other.

They do not measure valuation effects. Pricing changes are recorded as outcomes where the public record shows them. No claim is made about a discount attributable to how a company was described.

They are calibrated in the first year on one event type in one receiving market. Comparability across markets or event types has not been established.

They are applied in the first year by one research process, checked against sources but not yet against an independent second coding. Repeatability across coders is one of the four conditions above, and it is not yet met.

Corrections and method changes

Version 0.2 is provisional. Each evidence-based publication names the version of the primitives it used. A change to the wording of a definition that leaves its meaning intact is logged in place with its date. A change to a definition, a level, or a rule produces a new version, and earlier versions remain available so that older publications can still be read against the version they cite. The codebook carries its own version number and change log.

Version 0.1: initial version, signed in September 2026 under the title NPA Primitives. It encodes the six observation dimensions, the outcome register, the three-level coding, and the four conditions for a composite score as they were fixed in the programme design of July 2026, before coding of the first research season began. The first studies cite this version under its original name.

Version 0.2: identical to Version 0.1 in every definition, level, and rule. The method this document belongs to, previously named NPA (Narrative Power Analysis), was renamed TRACE (Testing Readings Against Comparable Evidence) in October 2026, and this document was renamed with it. A publication that cites NPA Primitives, Version 0.1 is citing the same primitives under their earlier name.