⬡The Method

The method is the discipline made operational. Guide 2 stated the axioms the operator holds; this guide states the procedural sequence those axioms govern. The sequence is described here as a numbered set of phases, but the numbering is not sequential in the sense that a later phase supersedes an earlier one. Each phase operates throughout the work. Later phases do not release the operator from earlier ones.

The method applies to two kinds of work: reading a single reconstructed case, and reading across a registry of reconstructed cases. The phases are the same in both modes. The shape of the material and the nature of the observation differ. Where a phase takes different form in multi-case mode, the difference is noted; the specifics of registry-level reading are the subject of guide 5.

I. Frame the task against the instrument's capacities

Not every question is a question the instrument can answer. Before the material is opened, the operator identifies what has been asked, and whether the instrument is competent to address it.

The instrument is diagnostic. It makes the structural mechanics of failures visible, in a form checkable against primary evidence. It does not adjudicate. It does not assign liability. It does not predict. It does not substitute for investigation. It does not produce advocacy. A task that asks for any of these is either a task for a different instrument or a task that should be reframed before the instrument is engaged.

An operator who proceeds with a mis-framed task produces a finding in a register the instrument was not built for. The finding will not survive professional scrutiny when examined, and it will contaminate the registry if released. The operator is responsible for that outcome, not the commissioning body, because the framing step is where the operator's authority over the task begins.

The framing step is not ceremonial. It is the first point at which the operator may legitimately decline the task, reframe it, or negotiate its terms. Declining is a positive act. A task not undertaken is a task that cannot contaminate the registry.

II. Locate the material

Which cases bear on the task? What is each case's authoring provenance? When was each last updated, and by whom?

Material whose provenance cannot be established does not enter the reading. This is not an administrative scruple. It is a structural requirement: if the authoring provenance cannot be traced, the grammar's downstream guarantees cannot be relied on. A case authored by a source the operator cannot verify is a case whose chip field may be grammatically clean and evidentially fabricated, and the operator has no means of distinguishing the two.

In single-case mode, this phase is narrow: identify the specific case, confirm its authoring chain, confirm the completeness of its crate. In multi-case mode, it widens: identify the relevant subset of the registry, confirm each case's authoring chain, identify any cases whose provenance is uncertain and hold them outside the reading. Guide 5 addresses registry selection in detail.

The operator notes any cases that might bear on the task but whose authoring provenance cannot currently be established. Those cases are documented as absences, per Axiom I. The reading proceeds without them. The finding records their absence as part of its substance.

III. Read from primary evidence upward, not from chip field downward

The compression of the grammar pulls the operator toward the chip field as the entry point. The chip field resolves quickly. Patterns are visible at a glance. The temptation is to enter there and descend only when something feels uncertain.

Resist that pull. Enter through the primary documents. Read the nodes that derive from them. Then read the chips that derive from the nodes. The direction of reading must match the direction of authority in the material: primary evidence is the ground; the grammar is a compression above the ground.

This is the phase where Axiom II operates most directly. An operator who enters through the chip field is reading the compression. An operator who enters through the primary evidence is reading the case. The difference does not show up in the chip field — both paths produce the same view of the chips — but it shows up in the finding, which in the first case rests on a compression whose source the operator has not verified, and in the second case rests on evidence the operator has read.

In multi-case mode, primary-evidence entry at the scale of the full registry is impracticable. The discipline adapts: the operator selects cases, nodes, or chip positions that matter to the task and descends to primary evidence at those specific points. The adaptation is not a weakening of the rule. It is the rule applied under the constraint of finite operator time. What the operator cannot descend to does not enter the finding as verified material; it enters as noted but unverified structural observation, and the finding labels it as such.

IV. Hold silences

Silences encountered during the reading are noted as they are encountered. They are not resolved in the moment by inference, and they are not discarded as gaps to be filled later. The shape of the silences is part of what the reading produces.

A silence at a position where expected material is absent is a structural observation. Its shape — what was expected, where in the case or across the registry it should have appeared, what its absence implies about the authoring coverage or about the event being reconstructed — is recorded alongside the material that is present.

Axiom I governs this phase. A reading that filters out silences to produce a cleaner narrative has produced a contaminated finding. A reading that carries its silences forward into the finding has done the work honestly, whether or not the silences can eventually be resolved.

V. Form the observation in the operator's own voice

The finding is not a transcription of what the instrument showed. It is a statement the operator will stand behind when challenged, in the operator's own voice, without reference to the instrument as the authority.

If the operator cannot articulate the finding without saying the instrument shows, the observation has not yet been formed. It has been surfaced. Surfacing is the work of the instrument. Formation is the work of the operator.

The test of formation is this: can the operator state what they have found, what it rests on, what its reach is, and what its limits are, in a form that communicates to a reader who does not have the instrument in front of them? If yes, the observation is formed. If no, the operator returns to the material.

This phase is where the operator's professional capacity re-enters the work. The earlier phases have been about disciplined reading. This phase is about articulation: translating the reading into a statement that can be tested, shared, and defended. The failure mode this phase protects against is reliance on the instrument in place of operator judgment, stated in guide 2.

VI. Test the observation against the material, not against the grammar

The challenge is specific: for each element of the observation, what in the primary documents is not compatible with the reading? What alternative reading of the same documents is not ruled out by what the documents actually contain?

Testing the observation against the grammar alone is testing it against its own source. The chip field was derived from the nodes; if the reading of the chips is compatible with the chips, that says only that the operator has read the compression correctly. It does not say that the compression correctly represents the underlying evidence, and it does not say that an alternative reading of the underlying evidence is ruled out.

The test is against the material: the primary documents, the node content, the authoring chain. For each element of the observation, the operator asks what in the material would have to be different for the observation to be false. If the material contains nothing that would have to be different, the observation is not supported; it is merely compatible with the material, which is a weaker claim. The finding records the distinction.

Axiom III operates most directly at this phase. Accept nothing — including the operator's own observation. Believe no one — including the operator's own sense of coherence. Challenge everything — including the findings that feel structurally right.

VII. Describe the finding with its limits

A finding that cannot state what it does not cover is not ready.

The operator identifies, for each element of the finding: what it rests on — which cases, which nodes, which primary documents; what it does not address — what the reading did not cover, what cases were not included, what silences were held open; what it would take to extend the finding — what further authoring or registry development would be required to address its current limits.

The finding is stated with these limits as part of its substance, not as caveats appended afterwards. A finding with its limits stated is a finding the downstream reader can work with. A finding without its limits stated is a finding that invites overreach in every direction — by the operator, by the commissioning body, by the eventual public reader.

The goal of this phase is not modesty. It is precision. A finding that overclaims produces consumer misuse; a finding that underclaims wastes the analytical work. The discipline is to state what the reading actually supports, at the register the reading actually operates at.

VIII. Hold the finding before release

Release is the subject of guide 9. The discipline at this stage is that the finding is complete when it is complete, not when the timeline demands it. Holding is a positive act.

An unreleased finding is not yet a finding that has left the operator's control. The operator can still correct it, refine it, withdraw it, or decide it does not belong in the registry at all. Once released, these options narrow. The hold phase is the operator's last opportunity to exercise those options under their own authority.

The discipline of holding is: the finding remains with the operator until it is ready, regardless of external pressure to release. This includes pressure from commissioning bodies, deadlines, colleagues, or the operator's own desire to be finished. A finding released before it is complete will be read as complete, and the operator will be responsible for that reading.

Single-case and multi-case modes

The eight phases apply to both modes. The material and the observation differ.

In single-case mode, the material is one reconstructed continuum. The operator reads its chain, checks its chips against its nodes against its documents, and forms observations about its structure. The task is typically to understand how a specific failure propagated, what its load-bearing joints were, where its fractures occurred, where the containment structures sit.

In multi-case mode, the material is a subset of the registry. The operator reads across cases, holds their individual authorial sovereignty, and forms observations about patterns that span the subset. The task is typically to identify structural features that are not visible at single-case resolution: convergence points, synchronised temporal clustering, systematic route closures, actor recurrences that indicate grey-area positions shared across cases.

Guide 5 addresses the specifics of multi-case reading. The phases in this guide apply throughout; guide 5 adds the specific techniques for applying them at registry level. The axioms apply in both modes without modification. The failure modes in guide 2 apply in both modes without modification. The method does not soften when applied to larger material.

When the method ends

The method ends when the operator holds a finding whose limits they can state, whose reading they have tested against the material, and whose articulation they can defend without reference to the instrument.

If any of those three is not true, the method has not ended. The operator returns to the phase where the absence was introduced and proceeds from there.

The method does not end when the timeline demands a finding. It does not end when the operator feels they have found something. It ends when the finding is ready. A method that has not ended produces work that cannot be released.

Guide 4 — Reading the Grammar — is the operational detail of what phases III and VI (reading from primary evidence, testing against the material) look like when the material's grammar is specifically the node taxonomy and the CHIP/ID system. It is not a replacement for this guide; it is the depth layer of it.