This packet continues the user-supplied Master Continuity Charter v1.0. It preserves the primary ternary/alternative-representation question while keeping governance and CI as supporting, separately evaluated work. It does not alter the public runtime contracts or certify any implementation.
Start with the source-grounded lineage, the versioned hypothesis register, and the selected next discriminator. The session handoff records the next single action and stopping criterion.
t81recursion.cweb is byte-identical at Foundation commit fe00396831c8bb249fde06a04186fa02f6018caf (2025-11-22). Its archive was removed from the working tree on 2026-03-06 but remains in history. Hanoi’s public Git repository currently has one import commit on 2026-10-25. This establishes shared source, not the original invention date or the direction of every private transfer.-1 + 0 emits a digit 2; 3 * 1 emits zero. These are failures against the balanced positional-integer contract used by the probe. Componentwise multiplication may instead belong to a vector algebra; no disproof of ternary arithmetic or future hardware follows.Under the ideal fixed-capacity model, let n_b = ceil(log_b M). Ternary storage is cheaper exactly when C3/C2 < n_2/n_3; the asymptotic limit is log2(3). This independently derived comparison assumes the same cost units and no sensing, ECC, control or conversion overhead. It is neither a universal latency threshold nor a manufactured-device result. Joint/block packing changes the capacity being compared and must be handled for both baselines.
Use Python 3, a C11 compiler, and a clone containing Hanoi commit 59205d76db4aca3da7c0bdc521924b8f9bc52cc3:
python3 docs/research/continuity/probes/reproduce_hanoi_arithmetic.py \
--hanoi-root /path/to/Hanoi --output-dir /tmp/t81-hanoi-probe
python3 docs/research/continuity/probes/storage_boundary.py
python3 docs/research/continuity/probes/verify_lineage.py \
--foundation-root /path/to/t81-foundation --hanoi-root /path/to/Hanoi \
--lang-root /path/to/t81-lang
The first command extracts function bodies verbatim from the pinned Git object and compiles only those functions plus an independent bounded integer oracle. It does not tangle CWEB, load kernel modules, run a VM, test a GPU, or validate a complete Hanoi build. Exit zero means the expected historical counterexamples were collected; the JSON explicitly records candidate_passes: false.
The full Hanoi root make -n all independently fails with a missing hvm_loader.c rule. No full-build success is claimed; ctangle was unavailable. Sources and histories were inspected without running installation/migration scripts or rewriting the archives.
Evidence includes source pins, source/blob identities, exact extracted bodies, compiler command/version, result counts, the dry-run failure, an exact finite-state calculation, and current PR snapshots. A/B/C/D/E labels distinguish reproduced observations, source inspection, reports, derivations and proposals. The referenced 34-page PDF, original paper and two historical transcripts were not available under the supplied names; their contents were not inferred or quoted.
PRs #565 and #566 remain open/draft at their existing heads. Their CodeQL checks now report success, but their required quality gates report failure. No merge, closure, new PR, CI repair or branch-protection change was performed in this continuation. Protocol 001 remains NOT READY and was not modified.