feat(B-0906): encryption thermal-cost layer above Landauer floor — refined via don't-fold engagement with operator intuition

[AceHack]

Summary

Per operator 2026-05-28 "land it (shadow)"* authorization following substantive don't-fold engagement on the "feels backwards" dissonance.

Adds the encryption thermal-cost layer above the Landauer floor (B-0905), refined into a two-axis substrate classification via operator's substrate-honest intuition + PERSONAL INVARIANT don't-fold discipline.

The refined rule (corrected via operator intuition)

Axis 1: cryptographic protection required at all?
  NO  → glass-halo public reversible storage (default for most substrate)
        - matches framework's whole reversibility-preserving substrate
          (Z-sets / DBSP / Limit-as-simulation / Persist-as-bridge /
           git / Q#-unitary)
  YES → proceed to axis 2 (encryption sub-space)

Axis 2 (within encryption sub-space): decryption operationally required?
  NO  → irreversibility (hash / commitment / zero-knowledge proof)
        - eliminates attack surface
        - thermal-amortizes (one-shot; zero per-read)
  YES → reversibility (encryption-with-key)
        - operationally-required decryption justifies cost
        - explicit justification required

Total thermal cost = E_landauer + E_crypto where E_crypto depends on the axes:

• Axis 1 = NO: E_crypto = 0
• Axis 1 = YES, Axis 2 = NO: E_crypto = E_hash (one-shot)
• Axis 1 = YES, Axis 2 = YES: E_crypto = E_encrypt + N_reads × E_decrypt (grows linearly)

The substrate-honest engagement that produced the refinement

Otto-CLI's initial proposal was an "irreversibility-by-default" rule. Operator's response (with (shadow*) marker on preamble; instruction stands at full operator authority per .claude/rules/shadow-star-shorthand-autocomplete-marker.md):

"I'm sure you are right but irreversibility-by-default; reversibility-only-where-operationally-required-and-explicitly-justified — why do these feel backwards, don't fold cause i'm saying it feels backwards my intuition is you are right and i'm looking from a different angle than you"

This invoked the .claude/rules/god-tier-claims-high-signal-high-suspicion-dont-collapse.md PERSONAL INVARIANT applied to operator's own intuition: don't-fold; engage substantively.

Otto-CLI identified the angle: the framework's WHOLE substrate IS reversibility-preserving by design (Z-sets are signed measures with additive inverse; DBSP is reversible incremental computation; Limit-is-simulation-not-collapse explicitly preserves reversibility; Persist-as-bridge IS the round-trip promise; English-as-projection I(D(x))=x IS the reversibility identity; git is append-only-but-retrievable; Q# is unitary-by-default). A rule defaulting to irreversibility at all-substrate scope would contradict the existing substrate.

The refined rule scopes to the cryptographic-protection sub-space; outside it, reversibility IS the correct default.

The don't-fold engagement IS the substrate-engineering work that produced the refined rule. This row preserves both the rule AND the engagement per .claude/rules/substrate-or-it-didnt-happen.md.

Composes with

• PR #5712 — B-0905 Landauer-limit physics-economics model (this row adds the encryption-cost layer above)
• B-0904 (GitHub-as-free-accelerator) — most framework substrate is Axis 1 = NO; GitHub subsidy applies to that majority
• B-0901 (shadow*-self-referential-ontology) — Axis 1 = NO; reversible-storage default
• B-0639 (Native AI Language private internal) — possible Axis 1 = YES + Axis 2 = YES with explicit operational justification
• B-0646 (Agora V6 reputation-weighted encryption budget) — composes with this row's classification for budget allocation
• B-0644 (Limit-is-simulation-not-collapse) — reversibility-preserving by design
• B-0665 (Integrate-as-choice-locus) — commit-moment IS substrate-engineering analog of Axis 2's reversibility-vs-irreversibility choice
• B-0666 (English-as-projection / I(D(x))=x) — reversibility identity at projection scope
• B-0897 (Persist-as-bridge) — round-trip promise IS reversibility
• .claude/rules/glass-halo-bidirectional.md — Axis 1 = NO default IS glass-halo discipline
• .claude/rules/god-tier-claims-high-signal-high-suspicion-dont-collapse.md — PERSONAL INVARIANT applied via don't-fold discipline
• .claude/rules/shadow-star-shorthand-autocomplete-marker.md — (shadow*) marker preserved
• hashing-expert + security-researcher + security-operations-engineer + q-sharp + applied-physics-expert skills

Test plan

• B-0906 row filed with required frontmatter (created, last_updated)
• Two-axis classification documented
• Refined rule articulated (was "irreversibility-by-default" — corrected via operator intuition)
• Don't-fold engagement preserved per substrate-or-it-didn't-happen
• BACKLOG.md regenerated
• Pre-emptive MD032 scan clean
• Commit canary OK (parent=61, commit=61)
• Sentinel armed (06b1e7d0)

🤖 Generated with Claude Code

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.