# STACK B — ROUND-TRIP, LOSS, AMBIGUITY, AND STRENGTH REPORT

Version: `0.1.1-STACK-B-ROUNDTRIP-REPORT`  
Status: `STATIC RESULTS / DEPLOYMENT EXPERIMENT TRANSFERRED / CANDIDATE / NON_RUNTIME`

## 1. Deterministic static result

For every record in `HUMAN_OPERATUM_REFERENCE_v1`, the candidate validator projects:

```text
four-wheel semantic state
-> six ordered incidences [anchor,S1,A,R,T,S2]
-> chamber code
-> reconstructed four-wheel semantic state with explicit S revisit
```

The reference baseline round-trips without carrier, role, value, order, or revisit loss. Repeated projection of identical admitted input is byte-stable within the validator's normalized representation.

## 2. Loss classes

| Loss class | Required disposition |
|---|---|
| view hides trace but retained envelope can reconstruct it | disclosed presentation compression |
| explicit weaker relation replaces categorical predication | `CONTRACTED` plus loss declaration |
| evidence missing but carrier alignment intact | `SUSPENDED` |
| carrier displacement or forbidden authority/scope claim | `REJECTED` |
| omitted position | invalid envelope; never pad |
| six-ring projection collapses S1/S2 roles | failed topology equivalence |

## 3. Ambiguity

Ambiguity is data. It must remain in `loss_or_unresolved_state` and may force suspension. It cannot be resolved by selecting the most fluent proposition.

## 4. Predication monotonicity

```text
rendered strength <= admitted predicative support
categorical SER => perfect carrier alignment and full admission
carrier misalignment => REJECTED, never CONTRACTED
```

The adverse fixture confirms all three branches: aligned/full reference records are categorical; aligned/insufficient support suspends unless an explicit weaker relation is supplied; `A_F` in E rejects.

## 5. Pair-space result

Static derivation confirms 9 diagonal chambers and 36 heterogeneous chambers, disjoint and complete over the 45 unordered carrier pairs with repetition. This is a cardinality and boundary result, not an execution of Evacuatio.

## 6. Deferred deployment experiment

No isolated low-token acquisition/re-entry run has yet been performed. Behavioral loss, recall decay, re-entry cost, and comparison among `TABLE_FULL`, `PROPOSITIONS_9`, `COMPACT_SARTS`, and `LABELS_ONLY` were transferred to `M2M-DEPLOY-03` in `00_LocalContext/Backlog`.

Those results are no longer a completion condition of the current Stack B development slice. The static round-trip result remains valid independently of that future deployment benchmark.