PROPOSED INTERPRETATION INSERTIONS - NOT APPLIED TO ORIGINALS
Version 1.0 working release candidate, 9 November 2026

These are short editorial additions, not theorem retractions. Verify the
current paragraph anchors before merging; the inspected baseline is the
preserved November 4 source set, with live thinning/acquisition rechecks on
November 9. The refinement and logical-topology anchors below were checked
against preserved text; their November 9 live retrievals failed. A source
hash identifies inspected bytes, not priority.

1. SUCCESSIVE CONCEPT DIFFERENTIATION AS OBSERVATION REFINEMENT
Target: Section 11.1, after the paragraph explaining that finer description
can be topologically conservative while adding local information.
Insert exactly:

"Here 'neutral' is a structural term, not an information-gain score. In a
finite noiseless model, uniform exact-state recovery can improve from 1/n
to 1 while every observation complex remains contractible and every
coordinate-forgetting map is a homotopy equivalence. Even the full homology
diagram therefore need not determine recovery capacity. The actual signature
fibers, prior and target remain operational data. The simultaneous two-bit
addendum gives a different caution: under its explicit height and triangle
restrictions, integer and characteristic-two tests recognize collapse, but
a single odd-characteristic test can miss relative two-torsion. This does
not change the single-observation weighted-height theorem here."

Reference to add: Signed incidence and a nine-state torsion witness in
simultaneous two-bit order thinning, Sections 3-6 (this local addendum;
coordinator assigns a public locator only after publication).
Preserve: Theorems 7.2 and 7.9; the separate interaction-core results;
exclusive K2,2 Question 19.1. This example does not answer that question.

2. OBSERVATION TOPOLOGIES FOR FINITE PROPOSITIONAL SEMANTICS
Target: Section 10, after the paragraph following Proposition 10.1 which
explains that continuity can merge distinguishable points.
Insert exactly:

"Preservation of the positive-observation order is not a guarantee of
unchanged predictive performance. Homotopy type and homology are coarser
summaries still: neither determines the observation fibers. Under noiseless
observation and zero-one loss, optimal target-prediction success is the sum,
over fibers, of the largest target-class probability mass within each fiber.
For uniform identity recovery on n states, this is the number of nonempty
fibers divided by n. Costed acquisition additionally depends on the supplied
tests and their costs; even an identical final observation topology can
arise from libraries with different optimal acquisition costs."

Reference to add: the addendum, Section 6, for the explicit finite examples.
Preserve: Proposition 10.1 and the existing Section 11.1 nerve warning.
Do not replace 'topology' globally with 'homology'; the full observation
topology retains distinctions that its homotopy type discards.

3. FROM OBSERVABLE STATES TO ACQUIRABLE PREDICTIVE MODELS
Target: Section 9, following the scope paragraph listing sensing costs among
features requiring changed models and proofs.
Insert exactly:

"Exact feasibility is distinct from budgeted performance. In extensions with
a prior and query costs, optimum success and acquisition cost depend on the
target, supplied test library and execution contract, not only on the final
observation topology or a homology diagram. For the identity-recovery target
on an antichain of four equiprobable states, with every supplied query costing
one, two coordinate tests and their complements and
four singleton indicators generate the same discrete topology, but have
optimal expected unit-cost exact-acquisition costs 2 and 9/4. This example
uses a different background order from the endpoint-antichain family of
the present classification. Extensions to simultaneous blocks must also
check each proposed sequential execution separately; simultaneous block
admissibility need not authorize every ordering in general posets."

Reference to add: the addendum, Section 6, for the cost and four-state
execution-order examples with their explicit matrices and collapse pairs.
Preserve: the fixed-state R subset E_* criterion, local pre-answer legality,
branch cancellation, supplied-query orientation, moving-state distinctions,
and the express denial of a physical-safety guarantee.

MERGE RULE
These paragraphs do not change proof hypotheses or silently enlarge an
existing theorem's domain. New publication links and bibliography keys are
the coordinator's final integration step, not placeholders to put online.
