Three concrete improvements to the textbook ingest layer:
Per-book make targets
=====================
- `make textbook ID=<id>` — generic, looks up one entry from the
manifest and runs the right ingest path (BFS crawl if
`crawl_url` is set, shallow URL ingest otherwise).
- `make textbook-list` — list ingestable textbook ids.
- Convenience targets per active manifest entry:
textbook-bogart, textbook-keller-trotter, textbook-levin,
textbook-aristotle-prior, textbook-aristotle-posterior,
textbook-newton, textbook-morin
Each per-book target writes to a dedicated shard at
`$(CRAWL_SHARDS_DIR)/textbook_<id>.db` — one shard per textbook
id rather than one per host (which would conflate multiple
Wikisource books — Aristotle Prior, Posterior, Newton — into
one shard).
The manifest helper script gains `ids` and `lookup <id>`
subcommands so the Makefile can introspect manifest entries
without grep gymnastics.
Idempotency contract documented
================================
Every ingest path is content-addressed. Same content →
same `document_root` → no-op insert. Re-running `make textbook-*`
or `make crawl-textbooks` on the same network state is safe at
the database layer — no duplicates, no schema drift, no
audit-event proliferation. HTTP refetch happens every run; pair
with `make recrawl-check` for a HEAD-only freshness probe.
README rewritten to make this contract explicit, with shard-
layout table + crawl-delay caveats consolidated.
Vital-books checklist
=====================
The README now carries a g4-pillar coverage table:
Pillar I Logic ✓ Levin + Aristotle Prior + Posterior
Pillar II Set Theory ✓ Levin
Pillar III Arithmetic ✓ Levin
Pillar IV Geometry ✗ Hilbert PG #17384 (PDF/TeX only)
Pillar V Probability ✗ Kolmogorov license analysis pending
Pillar VI Phys. Newton ✓ Newton Principia Motte
Pillar VII Combin. ✓ Bogart + Keller-Trotter + Levin
Pillar IX λ-Calculus ✗ Church + Turing 1936 papers pending
5 of 7 pillars now have surface coverage. The remaining gaps
are explicit license (V) or format (IV PDF/TeX, IX paper-level)
issues, each documented in the corresponding manifest entry's
notes field.
Test suite stays at 1554 passed / 28 skipped — no source-code
changes to arborist itself; the textbook layer is pure tooling
on top of the existing pipeline.
Per the #000031 implementation track — surface-ingest cited
textbooks for warrant promotion of claim-pack records. Five new
manifest entries spanning the PD-scope additions plus
license-fail placeholders for citation traceability:
newton-principia-motte (PD)
Newton's Mathematical Principles of Natural Philosophy
(Motte 1729 / Motte-Cajori 1846 English translations) via
Wikisource. Crawled @ depth=2: 60 docs, 289 chunks, 10K
edges. FTS5 spot-checks: "law of motion" → top hit on the
1729 Motte title page; "gravitation" → Newton-titled
chapters in the Wikisource Physics category. Cited by
claim-pack pillar VI.
aristotle-posterior-analytics (PD)
Mure / Owen translations via Wikisource. Crawled @ depth=1:
20 docs, 71 chunks. FTS5: "syllogism" → Posterior Analytics
Bouchier + Owen translations as top hits. Companion to
Prior Analytics already in the manifest; together cover
pillar I logic axioms.
kolmogorov-foundations-probability-1933 (placeholder)
German original PD-by-age in EU; US copyright restored via
URAA through life+70 (Kolmogorov d.1987 → US PD 2058).
Morrison 1956 English translation: Chelsea Publishing
copyright. Entry kept with empty urls field for citation
traceability; license analysis pending.
lukasiewicz-aristotle-syllogistic-1951 (proprietary placeholder)
Oxford University Press, copyright through life+70
(Łukasiewicz d.1956 → PD 2027 EU, longer US via URAA). NOT
redistributable. Kept as a license-fail placeholder; the
1921 Polish primary papers are themselves PD by age but
lack accessible HTML editions.
License-validator drive-by
==========================
The textbooks_manifest.py validator now treats placeholder
entries (urls + crawl_url both empty) as exempt from the
allow-list license check. License discipline stays fail-closed
at the URL-emit step — placeholders without URLs raise no
exception under any license token, so we can document
yellow-light proprietary citations (Mendelson, Enderton,
Łukasiewicz) without hacks.
Coverage summary post-commit
============================
g4 pillar | textbook coverage
─────────────┼─────────────────────────────────────────────────
I Logic | Levin (CC-BY-SA), Aristotle Prior + Posterior (PD)
II Set Theory| Levin (CC-BY-SA)
III Arithm. | Levin (CC-BY-SA)
IV Geometry | (Hilbert PG #17384 placeholder — PDF/TeX only)
V Probability| (Kolmogorov 1933 license analysis pending)
VI Phys. | Newton Principia Motte (PD)
VII Combin. | Bogart (GFDL), Keller-Trotter (CC-BY-SA), Levin
Out-of-scope follow-ups: PDF/TeX source support (#000031 §5),
Mendelson + Enderton license decision (#000031 §2.1),
chunk-resolution + derivations.proof_blob warrant promotion
(separate ticket).
The first iteration of this layer landed scaffolding plus shallow
seed-URL ingest, which produced 300-1000 byte stubs (front matter
+ nav chrome only) instead of real book content. Fox flagged
that as the wrong outcome — we were supposed to track down the
actual texts.
This commit fixes it by adding the deep-ingest path: each
manifest entry now carries `crawl_url` + `crawl_depth` +
`crawl_max` fields, and `make crawl-textbooks` iterates them
through the existing BFS crawler (one shard per host).
Manifest changes
================
Three URL fixups uncovered by the first crawl attempt:
- PreTeXt textbooks (Bogart, Keller-Trotter, Levin) ship an
`index.html` that's a 1-line meta-refresh redirect; the BFS
crawler doesn't follow meta-refresh, so the seed URL has to
point at the actual content entry (`frontmatter.html`,
`app-comb-2.html`, `ctgd/index.html`).
- Open Data Structures uses trailing-slash directory paths
(`/ods-cpp/`, `/ods-python/`, `/ods-java/`), not `/ods-cpp.html`.
- `appliedcombinatorics.org` declares `Crawl-Delay: 20` in
robots.txt — full BFS takes ~25 minutes; documented in the
manifest entry's notes so operators know to background it.
Hilbert + Boole stay flagged as `urls: []` placeholders because
PG eBooks #17384 + #15114 ship as PDF/TeX only (no HTML format).
Their `tex_url` + `pdf_url` fields are now preserved on the
entry for the future PDF/TeX source ticket.
Make targets
============
- crawl-textbooks — BFS-crawl every entry with `crawl_url`
- crawl-textbooks-stats — docs/chunks/size summary across crawl shards
Helper script: `bench/scripts/textbooks_manifest.py crawl-targets`
emits one tab-separated `<url>\t<depth>\t<max>\t<id>` row per
entry that declares a `crawl_url`.
Real-content verification
=========================
Crawl results across 4 of 5 textbooks (Keller-Trotter still
running in background due to crawl-delay; expected to complete
in ~25 min):
crawl_bogart_openmathbooks_org.db docs=44 chunks=161 971 KB
crawl_discrete_openmathbooks_org.db docs=51 chunks=376 1929 KB
crawl_en_wikisource_org.db docs=30 chunks=110 1651 KB
crawl_opendatastructures_org.db docs=64 chunks= 84 1110 KB
TOTAL 189 docs, 731 chunks
FTS5 spot-checks confirm substantive content landed:
- Bogart "binomial coefficient" → "CTGD Partitions and Stirling
Numbers" chapter, top hit
- Levin "modus ponens" → "Propositional Logic" chapter, top hit
That's the real lift signal we were after. Out-of-scope
follow-ups remain: PDF/TeX support for PG editions (Hilbert,
Boole) and full Aristotle Owen-translation depth (currently
~75% complete on Wikisource).
First make fetch-textbooks run revealed three URL issues that the
license-validator catches as policy-correct but the underlying
hosts return as 404:
- Keller-Trotter applied-combinatorics.org used a stale
/appcomb/sec_pre.html path; the canonical PreTeXt-generated
book lives at /book/app-comb-2.html.
- Open Data Structures uses /ods-cpp/ /ods-python/ /ods-java/
trailing-slash directory paths, not /ods-cpp.html (404). Three
language editions of the same content; ingest all three so the
multi-language structure stays intact.
- Hilbert Foundations of Geometry on Wikisource resolves to a
1906 Popular Science Monthly article ABOUT Hilbert's work, not
Hilbert's translated text. Project Gutenberg eBook #17384 is
PDF/TeX only — no clean HTML edition exists. Same status as
Boole #15114: kept the entry as a placeholder for license +
provenance documentation, urls intentionally empty until PDF
source support lands (out of scope per #000031).
Final fetch result: 7/7 manifested URLs ingested (Bogart,
Keller-Trotter front matter, Levin front matter, Aristotle Prior
Analytics, ODS C++, ODS Python, ODS Java) — 7 docs / 7 chunks /
86 edges / 8/8 Merkle proofs verify (verify samples 8 because
Aristotle yields multiple chunks).
Two textbooks (Hilbert + Boole) carry zero URLs; the helper
script's allow-list passes them through (license is PD;
license_url is set) but emits no URLs since their `urls` array
is empty. Documents this gap explicitly in each entry's notes
field so future operators see what awaits PDF source support.
Front-matter URLs alone yield minimal indexable content per
textbook (PreTeXt single-chapter scaffolds). To grow real
coverage: append more chapter URLs to a textbook's `urls`
array, OR use `make crawl-ingest URL=<home>` per the README's
"Going deeper" section.
A make-target scaffold for pulling public-domain and copyleft-
redistributable textbooks into a dedicated arborist shard via the
existing HtmlPageSource ingest pipeline. No new ingest machinery —
the existing path (robots.txt → noise-strip → 512-token chunk →
Merkle root → audit-event) IS the consistent process.
Manifest format
===============
bench/fixtures/textbooks/manifest-v1.jsonl — one JSON record per
textbook with explicit license tokens (PD / CC-BY / CC-BY-SA /
GFDL / OSI-approved). Fail-closed validation in
bench/scripts/textbooks_manifest.py refuses to emit URLs from
entries with missing or disallowed license tokens, so license
discipline surfaces at the URL-emit step rather than getting
silently buried in a fetch run.
Allow-list excludes CC-BY-NC (incompatible with arborist's
AGPLv3 distribution profile) and CC-BY-ND (no-derivatives
prevents chunking). Wilf's generatingfunctionology stays out
because its educational-use license forbids rehosting; citable
but not redistributable.
Initial entries (7 textbooks, 8 seed URLs)
==========================================
- Bogart, Combinatorics Through Guided Discovery — GFDL-1.3
- Keller & Trotter, Applied Combinatorics — CC-BY-SA-4.0
- Levin, Discrete Mathematics: An Open Introduction (3rd ed) —
CC-BY-SA-4.0 (4th ed switched to NC; pin 3rd)
- Hilbert, Foundations of Geometry (Townsend 1902) — PD via PG
- Boole, An Investigation of the Laws of Thought (1854) — PD
- Aristotle, Prior Analytics (Jenkinson) — PD via Wikisource
- Morin, Open Data Structures — CC-BY-2.5
Covers pillars I (logic) / II (set theory) / III (arithmetic) /
IV (geometry) / VII (combinatorics) on the claim-pack
(#000029) layout, plus a CS anchor for downstream domain
expansion. Each entry's pillar_targets field lists which
claim-pack pillars its records are candidate citations for.
Make targets
============
- textbooks-summary — license + URL counts per entry
- textbooks-urls — flat URL list to stdout
- fetch-textbooks — ingest all manifested URLs into
$(TEXTBOOK_DB) (default
~/.arborist/textbooks.db)
- textbooks-stats — documents / chunks / edges
- textbooks-verify — sample Merkle proof verification
To grow coverage of one textbook: append more chapter URLs to
its `urls` array. For deep-BFS of a textbook home: use the
existing `make crawl-ingest URL=<base> DEPTH=N` instead.
Smoke-tested end-to-end against Wikisource Prior Analytics:
1 doc / 1 chunk / 12 outbound edges / Merkle proof passes.
Shows the pipeline works; populating each book to depth needs
either more URLs in the manifest or the crawler.
Out of scope
============
- PDF processing. The Internet Archive hosts PD textbooks
(MacMahon's Combinatory Analysis 1915, Whitworth's Choice
and Chance 1867, the Motte 1729 Principia) as scanned PDFs;
a TextbookPdfSource with pdftotext / pypdf extraction is a
separate ticket — current scope is HTML-shaped sources.
- Mendelson + Enderton (proprietary; await #000031 §2.1
decision).
- Wilf generatingfunctionology (license forbids redistribution).
Test suite stays at 1554 passed / 28 skipped — no source-code
changes to arborist itself; the textbook layer is pure tooling
on top of the existing pipeline.
Three streams. Two land in the repo; one lands in fox's Downloads
(existing g4 packs live there too).
Witness sweep automation
========================
`bench/scripts/witness_sweep_cron.sh` — schedulable harness that
runs `make bench-witness-sweep` + `make bench-witness-divergence`
back-to-back and (with --commit) auto-commits any new divergence
fixtures + sweep result. Fail-closed: pre-commit hook failure
aborts; non-main-branch refuses to commit.
$ bench/scripts/witness_sweep_cron.sh # dry run
$ bench/scripts/witness_sweep_cron.sh --commit # auto-commit
$ bench/scripts/witness_sweep_cron.sh --commit --push # full pipeline
Recommended cron (every 6 hours; LLM cost scales with frequency):
0 */6 * * * \
bench/scripts/witness_sweep_cron.sh --commit --branch main \
>> /var/log/arborist-witness.log 2>&1
Smoke this commit: dry-run produced 5 fresh divergences (10
cumulative across two extraction passes). The witness-sweep + 5F
extractor pipeline runs unattended now; calibration-data corpus
will grow over time.
#000018 — adversarial soft-hash covert-channel analysis
========================================================
`docs/soft-hash-channel-analysis.md` — formal threat model + channel-
capacity definition + cryptographic reduction + mitigation table
+ residual analysis. Doc-only deliverable per the ticket's hard
constraint.
Headline result: under threat models T1 (data poisoner) and T2
(gradient adversary), the channel is bounded by SHA-256 partial-
preimage hardness when φ (the anchor map) is independent of
parameter geometry. T3 (hyperparameter adversary) requires
M2 (per-checkpoint anchor nonce) for boundedness.
Recommended mitigation: M2 (per-checkpoint nonce) — 32 bytes added
to the snapshot manifest, no training-ergonomics regression,
closes T1+T2 channels under any φ choice and bounds T3 by
per-checkpoint window. Fall back to M1 (PRG-based φ) if M2-only
deployment surfaces structural concerns. M3 (drop anchor entirely)
stays in reserve as the strict-construction fallback.
Three open questions (§9): Hessian alignment under φ_linear,
PRG choice for φ_PRG, and explicit T3 per-window bound. Each is a
follow-up ticket.
Ticket #000018 status: closed · landed 2026-05-09 (analysis doc).
v7 § 9.10 spec amendment proposed in §7 of the analysis.
#000033 — pillar VII (combinatorics), Claude-authored
======================================================
NOT committed to the arborist repo (the existing g4-v2 packs live
in `/home/fox/Downloads/` too — that's the operator's bundle
location). Two new bundle files at:
/home/fox/Downloads/axiomsclaude-vii-v1.json (7 axioms)
/home/fox/Downloads/theoremsclaude-vii-v1.json (7 theorems)
Pillar VII covers combinatorial counting — the gap between Grok's
pillars VI and IX in the v2 packs:
axioms (7): addition principle · multiplication principle ·
pigeonhole principle · factorial definition ·
binomial coefficient definition · Pascal's rule ·
empty-set / boundary axiom
theorems (7): binomial theorem · inclusion-exclusion (counting
form) · hockey-stick identity · Vandermonde's
identity · Catalan number closed form · stars-and-
bars · strong pigeonhole
Each record in the dual-thread format the existing g4 packs use
(Δ symbolic LaTeX + ∇ verbose prose + ∇ concise + sigil + formal
language + role + status + source_reference + date + foundational
group + category + subfield). Per fox's directive: explicit
authorship metadata everywhere — `authored_by: Claude (Anthropic)
— model claude-opus-4-7`. NOT Grok-generated; no silent invention.
Each record carries `pi_star_ref: combinatorics@v1` so the kernel
binding is explicit. Theorems list `depends_on_axioms` arrays so
each theorem cites the foundation axioms it bottoms out on.
Smoke test (committed alongside):
$ arborist --db /tmp/test.db ingest --source claim_pack \\
--bundle /home/fox/Downloads/axiomsclaude-vii-v1.json \\
--bundle /home/fox/Downloads/theoremsclaude-vii-v1.json
→ 14 docs, 14 chunks, 0 cross-bundle edges
Source attributions: Stanley EC1, Brualdi Introductory
Combinatorics, Knuth TAOCP Vol 1, plus historical sources where
applicable (Pascal 1654, Vandermonde 1772, Dirichlet 1834, Catalan
1838, Feller 1950 for stars-and-bars).
Tests: 1636 passed, 37 skipped (no regressions; pillar VII
ingestion smoke covered above).
Apples-to-apples FTS5 search comparison on a single shard
(~/.arborist/shards/000.db, ~867 K Wikipedia surface docs)
before vs after ingesting the claim-pack source (#000029)
bundles.
Method: copy 000.db twice; ingest axiomsg4-v2.json +
theoremsg4-v2.json into one copy; run identical search queries
against both. FTS5-only path so retrieval-layer behavior is
isolated from QA pipeline reranking.
Results: 3/4 representative queries show claim-pack record in
top-3.
modus tollens → claim-pack at #3 (BM25 31.77)
law of excluded middle → claim-pack at #3 (related axiom
in same pillar)
associativity of addition → claim-pack at #1 — displaces
Wikipedia's general "Addition"
article entirely
Bayes theorem → no top-3 lift; Wikipedia "Bayes"
+ "Bayes rule" articles dominate
via short-doc BM25 bias
Storage tax: sub-MB. 78 chunks against 6.2M existing chunks is
below filesystem allocation granularity. Claim-pack rides
essentially for free on the existing Wikipedia substrate.
Honest verdict: claim-pack ingest earns its storage tax for
narrow technical queries where pre-distilled CORE-shape content
has title-token advantage. Doesn't help on queries Wikipedia
already covers with focused articles. The Bayes theorem case
likely shifts under the full QA pipeline's body-coverage sqrt
rerank (which counters BM25 short-doc bias) — that's a separate
larger bench worth running once the retrieval layer's other
stages are exercised.
Four-rung audit ladder still caps at ANCHOR-WARRANTED per
#000029's honest ceiling. Surface-ingest of cited textbooks
(#000031) is the path toward EVIDENCE-WARRANTED.
Three small streams:
#3 — close#000030 properly
============================
All 7 phases + Phase 1b landed across two commits (`04f3f5d`,
`abe5988`). Status header updated; ticket body now carries a phase
landing table with commit refs:
Phase 1 algebra-symbolic@v1 04f3f5d
Phase 1b algebra-symbolic-simplified@v1 04f3f5d
Phase 2 calculus-derivative@v1 04f3f5d
Phase 3 calculus-integral@v1 fox-direct
Phase 4 calculus-limit@v1 abe5988
Phase 5 calculus-series@v1 abe5988
Phase 6 linear-algebra@v1 abe5988
Phase 7 function-sampled@v1 abe5988
Plus tabular-pinned@v1 (last reserved stub) graduated in abe5988
closes the registry chapter — 15 concrete π*'s, no remaining
reserved stubs. Index updated.
#5 — composition fixtures across new SymPy π*'s
================================================
12 new tests in tests/test_pi_star_compositions.py covering pairs
that compose naturally:
- algebra-symbolic ∘ algebra-symbolic — idempotency check (running
expand twice equals expand once for any expression).
- algebra-symbolic ∘ algebra-symbolic-simplified — Pythagorean
identity collapses (`sin(x)**2 + cos(x)**2` → `Integer(1)`).
- Generic invariants: composition propagates PiStarError; manifest
fingerprint is order-sensitive; composite domain == inner domain;
composite bytes == manual chain bytes.
Test discipline: most compositions use `register_in_registry=False`
via a small `_safe_compose()` helper since the registry rejects
duplicate keys (#000015 invariant), so test ordering would
otherwise matter. Only the registration-test path uses real
compose().
#4 — end-to-end witness sweep against real shards + Hermes
===========================================================
New script `bench/scripts/witness_sweep.py`. Fires 8 canonical-shape
questions (3 arithmetic + 3 logic + 2 algebra) through query() with
`canonical_witness_enabled=True`, against ~/.arborist/shards (real
shard cluster) + the actual Hermes endpoint (NOT StubClient).
Records the agreement matrix per question to
bench/results/witness-sweep.json.
`make bench-witness-sweep` Makefile target. Honors
`ARBORIST_SHARDS_DIR`.
First real sweep (this commit, against Hermes-3-8B):
agreement label count rate
KERNEL-LLM-DIVERGED 5 62.5%
KERNEL-LLM-AGREE 3 37.5%
───────────────────────────────────────────
divergence_count 5 62.5%
wall median / max 130 ms / 1.1 s
Hermes diverged on 5/8 of the canonical-shape questions:
- said `1/10` for `0.1 + 0.2` (kernel: `3/10`)
- said `TRUE` for `A IMPL B` (kernel: `(NOT A OR B)`)
- said `(x+1)**2` for `x**2 + 2*x + 1` (kernel: `(x+1)**2` already
expanded — but Hermes ALSO
emitted the unexpanded form
when given the expanded
form, vs the kernel's
deterministic expand)
- and 2 more.
These are real LLM hallucinations on questions with closed-form
ground truth — exactly the calibration-data stream #000028
imagined. Pipeline validated end-to-end against actual hardware.
Pair: `make bench-witness-divergence` then extracts the 5
divergences as 5F-Falsification fixtures
(bench/fixtures/5f/falsification-witness-v1.jsonl, also committed).
Re-running the extractor produces byte-equal output (idempotency
contract from the extractor work).
Tests
=====
Full suite: 1636 passed, 37 skipped (was 1624; +12 composition
tests). The witness-sweep + extractor produce real artifacts now
committed under bench/results/ and bench/fixtures/5f/.
A new π* kernel that canonicalizes pure-integer counting
expressions and FAILS CLOSED on any input whose result isn't a
non-negative sp.Integer. Tighter domain than algebra-symbolic@v1,
which already accepts the same input surface but happily returns
symbolic / negative / non-integer outputs.
Distinguishing feature versus algebra-symbolic@v1:
algebra-symbolic@v1: binomial(n, k) → "binomial(n, k)" (symbolic
passthrough)
combinatorics@v1: binomial(n, k) → PiStarError (fail-closed
on free-symbol output)
algebra-symbolic@v1: binomial(Rational(1,2), 3) → 1/16 (rational)
combinatorics@v1: binomial(Rational(1,2), 3) → PiStarError
(output not Integer)
Boundary kept explicit: binomial(-3, 2) = 6 IS accepted because the
output is an integer 6. The fail-closed rule is on output shape
(Integer ≥ 0), not input range. Documented as
test_generalized_binomial_negative_args_accepted_when_integer.
Output format: plain decimal literal (b"10", b"5040"). Composes
with arithmetic@v1 for byte-identical agreement with the rational
route (b"10/1") so the multi-modality witness (#000028) can pin
equivalence-class agreement when both routes fire on the same
question.
Allowed surface (via SymPy primitives): binomial, factorial, ff /
rf (falling/rising), catalan, bell, partition, stirling, plus
arithmetic compositions over those primitives
(3*binomial(5,2) + factorial(4) = 54).
Coverage:
- 43 unit tests including binomial symmetry C(n,k)=C(n,n-k),
Pascal's rule C(n,k)=C(n-1,k-1)+C(n-1,k), the C(n,k) =
factorial(n)/(factorial(k)·factorial(n-k)) identity,
fail-closed paths (symbolic/negative/non-integer/relational/
parse), round-trip idempotence, composition with arithmetic@v1.
- 10 syntax + 12 semantics bench fixtures, 100% pass.
- bench/batteries/base.py PHASE_1_CARRIERS gains "combinatorics".
- Makefile bench-5s-combinatorics target.
All gate on pytest.importorskip("sympy") so a sympy-less suite
stays green. Full make test: 1537 passed / 28 skipped.
Sequencing rationale honored: this kernel lands FIRST so that
#000033 (claim-pack pillar VII for combinatorics) can bind its
records to the tighter integer kernel from day one — avoids
rebind churn on pi_star_ref fields.
Three small streams in one commit:
#000028 follow-up — witness divergence → 5F fixtures
=====================================================
Witness fan-out now writes a `providence_canonical_witness` audit
event when it fires (next to the capital-ledger record landed in
708aa45). Body carries pi_star_ref, question_text, agreement_label,
canonical_answer_text, llm_raw_text, llm_canonical_bytes,
cache_status. Best-effort write — chain failure never fails the
query.
New extractor `bench/scripts/witness_to_5f.py` reads those events
from a qa.db and writes them out as 5F-Falsification fixtures
matching the existing `falsification-live-v1` schema. Filtering
includes only divergence labels (LLM-DIVERGED / KERNEL-LLM-DIVERGED
/ CACHE-DRIFT); skips KERNEL-LLM-AGREE / STRICT-WITNESSED (no
calibration signal) and KERNEL-ONLY (LLM unparseable, not a
supervised-correction sample).
Idempotent: sorted by audit-event seq, so re-running against the
same qa.db produces byte-equal fixture files. The existing
fixture-digest discipline stays valid.
Makefile: `make bench-witness-divergence` (override default
qa.db / output path via WITNESS_QA_DB / WITNESS_OUT env-vars).
Closes the divergence → calibration data loop the witness ticket
imagined: every LLM hallucination on a canonical-shape question
becomes a supervised-correction fixture downstream prompt
improvements can grade against.
#000030 Phase 7 demo — function-sampled@v1 end-to-end
======================================================
`bench/scripts/demo_plot.py` — closes the loop on opencompletion's
activity24-math-plot.yaml. SymPy expression → quantized
integer-vector signature (canonical bytes) → optional matplotlib
PNG. Canonical bytes are the proof; PNG is just a downstream view
of the same evidence.
$ make demo-plot Q='sin(x)' PNG=/tmp/sin.png
Output JSON contains canonical_bytes_sha256 + canonical_bytes_preview
+ canonical_bytes_total_chars + grid metadata + the optional png_path.
matplotlib is gated — when absent, --png prints a warning to stderr
and skips the render; the canonical bytes still print. Tests skip
the PNG-presence assertion via `pytest.importorskip("matplotlib")`.
Public docs polish (#7)
========================
- docs/_source/bench.rst: updated fixture-count narrative (~660 →
662 default tasks + ~110 math π* fixtures); `make` quick-reference
now lists all per-π* 5S targets (tabular, calculus-limit/series,
linear-algebra, function-sampled) plus bench-real-shard,
bench-fork-baseline/score, bench-witness-divergence.
- docs/_source/v8-fork-score.rst: CLI section gained --out flag
documentation + a Make-harness sub-section covering
bench-fork-baseline / bench-fork-score / FORK_PARENT/CHILD/REPORT
env-vars.
Tests
=====
- tests/test_witness_to_5f.py — 8 new tests covering the audit-event
write (3) + extractor logic (5).
- tests/test_demo_plot.py — 6 new tests covering canonical-bytes
determinism + equivalence-class collapse + matplotlib gating.
Full suite: 1624 passed, 37 skipped (was 1568; +56).
tabular-pinned@v1 + calculus-limit@v1 + calculus-series@v1 +
linear-algebra@v1 + function-sampled@v1 — all reserved stubs
graduated; the π* registry is now 15 concrete kernels with no
remaining reserved-stub entries.
#000030 Phase 4 — calculus-limit@v1
====================================
sp.limit with thread-timeout. One-sided dir support (+/-/+-).
Pinned spelling for infinity cases: b"+oo" / b"-oo" / b"zoo"
(complex infinity) — bypasses sp.expand since Infinity isn't
algebraic. Finite results re-canonicalize through algebra-symbolic
recipe (sp.expand + sp.srepr). Unevaluated cases / timeouts emit
b"unevaluated:" + sp.srepr(<Limit>) sentinel, mirroring
calculus-integral's pattern.
#000030 Phase 5 — calculus-series@v1
=====================================
sp.series(f, x, x0, n).removeO() → sp.expand → sp.srepr. Drops
O(x**n) remainder explicitly so the canonical form is finite-byte.
Sentinel format mirrors limit/integral: b"unevaluated:Series(...)"
on timeout. n must be a positive int; 0 / float / negative rejected.
#000030 Phase 6 — linear-algebra@v1
====================================
Single π* covers the whole linear-algebra surface via {op, matrix}
JSON. Ops: rref / det / eigenvalues / inverse. Matrix cells go
through Fraction(Decimal(str(...))) for floats so 1, 1.0, "1.0"
all collapse to Rational(1, 1) — matching arithmetic@v1's
discipline. Without this fold, sp.sympify keeps floats as Float
(separate type) and downstream det/inverse return Float-shaped
bytes. Eigenvalues are sorted by srepr for determinism.
Output formats:
rref / inverse: rows/cols header + cells joined by | (rows by ||)
det: det:<num/den-or-srepr>
eigenvalues: eigenvalues:<value-1>x<mult-1>|...
#000030 Phase 7 — function-sampled@v1
======================================
Bridge to time-series-quantized@v1. SymPy expression + linspace
grid → quantized integer-vector signature in time-series's exact
output format (dt=...;dv=...;n=...;t0=0:v0|v1|...). Two functions
that render identically (within sample-grid tolerance) collapse
to the same canonical bytes. This is what plotting CAN become
in π* terms — the PNG render is a downstream view of the same
canonical evidence.
Math-only sampler (no numpy in the dep surface); Python's round()
is banker's-rounding so the bytes are interchangeable with
time-series-quantized@v1's output. Complex / non-finite samples
raise PiStarError rather than silently dropping imaginary parts.
tabular-pinned@v1 — last reserved stub graduates
=================================================
JSON-rows input ({schema, key_columns, rows}); declared
key_columns sort policy (stable sort by primary-key tuple);
type-fold per column (int/rational/bool through arithmetic@v1
discipline; str verbatim; bool normalized). Header case is
PINNED EXACT — Excel and PostgreSQL both care about case;
defaulting to lowercase-fold would break operator expectations.
Output: header (schema + key + n) + rows joined by \n + cells by |.
The π* registry has no remaining reserved stubs. Every modality
the substrate paper reserved is now real.
Test suite: 1568 passed (was 1467; +101). New closure-criterion
test (test_no_stub_pi_stars_remain) replaces the old reserved-stub
parametrize — adding a future stub re-opens this list.
110/110 fixtures pass across the 5 new bench-5s-* targets.
PHASE_1_CARRIERS gained calculus / linear-algebra / function-sampled
/ tabular.
Three small streams in one commit; each closes / expands a
recently-landed ticket without changing its hard contract.
#000026 Phase 3 wiring — authorship warrant ladder visible
============================================================
Phase 3 sidecar (arborist/qa/warrant_authorship.py landed in 60b5748)
exposed the classifier but didn't surface it. Two wirings:
- arborist/qa/inspect.py — diagnose_authorship_warrant runs against
the cached row's question + answer + per-source raw chunks +
URIs + titles; result lands as `authorship` field alongside the
other sidecars.
- arborist/cli.py _render_warrant_tail — appends ` · warrant:
<readable-tier>` when result['authorship'] is populated with a
non-quiet tier. AUTHOR_COPYRIGHT_FOOTER → "copyright-footer", etc.
NO_AUTHORSHIP_SIGNAL stays silent. Backward-compat: results
without an `authorship` key render unchanged.
Tests: 3 inspect-path tests (no-signal, copyright-footer,
repository-owner) + 4 render-tail tests (presence, no-signal
silence, missing-key silence, all-six-tiers readable mapping).
#000028 follow-ups — capital ledger + sample-rate
==================================================
Two policy fields layered on top of canonical_witness_enabled:
- canonical_witness_sample_rate (0.0..1.0; default 1.0). Operators
wanting passive calibration set 0.05 to fire witness on 5% of
canonical questions while paying 5% of LLM cost. 0.0 effectively
off; 1.0 = current always-on behavior. Gating uses random.random()
so distribution is uniform; clamped to [0, 1].
- Capital ledger row written for each FIRED witness (not skipped
ones). op_type='canonical_witness'; estimator inputs include
prompt_chars + answer_chars + llm_seconds + agreement_label +
pi_star_ref. Best-effort: ledger-write failure must never fail
the query (sidecar discipline).
Tests: 4 new — sample_rate=0.0 skips (no LLM call, no ledger row);
sample_rate=1.0 always fires; capital_ledger row written under
op_type='canonical_witness' with full input blob; sampled-out
witness records zero ledger rows.
Both fields fold into governance_policy_hash naturally via the
existing policy-hash machinery — flipping witness mode invalidates
prior records as expected.
#000025 Phase 1d — 5F fixture catalog 30 → 50
==============================================
Both synthetic and live sides of all 5 sub-batteries expanded
30 → 50 (+200 fixtures total: 5 × 20 synthetic, 5 × 20 live).
function — claim_count cycles 2..7 across new fixtures
falsification — 10-violation palette across new ids
feedback-loop — fact-N learning chains
finetuning — capability transitions across canonical π*
(math/logic/algebra/calculus pool)
formulate — multi-pointer claim shapes
500/500 pass through respective runners. test_session_integration
total bumped 562 → 662. Pinned test_5f_*_runs counts updated 30 →
50 (synthetic main + embedded + live).
Tests
=====
Full suite: 1467 passed, 36 skipped (was 1388; +79 across warrant
render + witness sample/ledger + 5F implicit coverage).
Two new π* canonicalizers extend the math substrate above
arithmetic@v1 (closed-form rationals) and logic-kernel@v1
(propositional Boolean → CNF):
algebra-symbolic@v1 (Phase 1) — symbolic-algebra domain.
sp.expand → sp.srepr canonical bytes. Polynomial identity collapses
((x+1)**2 ≡ x**2 + 2*x + 1); exponential identity collapses
(exp(a+b) ≡ exp(a)*exp(b), inherited from sp.expand's default
behavior); trigonometric identity does NOT collapse
(sin²+cos² ≢ 1). The trig surface is reserved for a future
algebra-symbolic-simplified@v1 variant that wraps sp.simplify at
unbounded CPU cost. Rejects relationals (`x > 0`) and
BooleanFunction shapes (`x & y`) via `isinstance(expr, sp.Expr)` —
sp.Symbol confusingly inherits from Boolean so the right rejection
filter is "not Expr" rather than "Boolean".
calculus-derivative@v1 (Phase 2) — calculus domain. JSON-shaped
{f, x, n} input → sp.diff → sp.expand → srepr bytes. Output is
itself a valid algebra-symbolic@v1 input so the two compose
naturally under arborist.pi_star.compose. n defaults to 1; bools
explicitly rejected (Python isinstance(True, int) is True so we
filter that explicitly).
Optional dependency: sympy ships in the new [math] extra
(pyproject.toml). Folded into [dev] so make bootstrap pulls it
transitively. An explicit `bootstrap-math` Makefile target documents
the opt-in for minimal-install users. Both modules self-guard
via `try: import sympy as sp / except ImportError: sp = None` and
only register(...) when sympy is present, so a fresh checkout
without [math] still loads arborist.pi_star without raising.
Preflight algebra route lands in
arborist.qa.query._canonical_projection_preflight between the
arithmetic and logic routes. Charset regex (_CANONICAL_ALGEBRA_RE)
allows lowercase letters + math chars; requires at least one
letter (else arithmetic wins); rejects natural-language leading
verbs via _CANONICAL_ALGEBRA_NL_LEAD_RE (4-letter minimum so
single-/two-/three-char identifiers like x, xy, sin, cos, pi
survive while "simplify (...)", "factor x...", "expand (a+b)..."
fall through). PiStarError + KeyError both fall through cleanly
so a sympy-less install just routes everything past algebra.
Bench substrate:
- bench/batteries/base.py PHASE_1_CARRIERS gains "symbolic_algebra"
- bench/fixtures/5s/syntax-algebra-symbolic-v1.jsonl (10 fixtures)
- bench/fixtures/5s/semantics-algebra-symbolic-v1.jsonl (13 fixtures
including the documented trig non-collapse + exp collapse)
- Makefile bench-5s-algebra target → 100% pass
Tests: 18 algebra-symbolic + 38 calculus-derivative unit tests +
~10 new preflight-route tests in test_canonical_projection.py. All
gate on pytest.importorskip("sympy") so a sympy-less suite stays
green. Full suite: 1369 passed / 27 skipped.
Phases 3-7 (integral, limit, series, linear-algebra,
function-sampled) remain open as future work; each lands as its
own ticket when an actual consumer surfaces.
Two fan-out streams. Both bear directly on the ticket's "search
latency on real shards" headline finding.
## Lazy concept_relations loading
Phase 1 (migration memoization) cut SQLite executes 65% but warm-
cache wall barely moved. cProfile pinned the next hotspot:
synonym_expand 2.8 s × 2 calls + _load_token_idf 0.3 s. The eager
loader dumped all ~290 K concept_relations rows on first call —
the price of being able to answer ANY future question without
re-querying. Wrong tradeoff for single-query CLI use.
Refactored arborist/concepts/query.py:
- New _load_neighbors_for(shards_dir, tokens) — targeted
WHERE token IN (...) OR target IN (...) query. Returns just the
direct synonym neighborhood for the given tokens (~300 rows for
a typical 5-token question, vs 290 K for the full table).
- New _load_rivalry_rows(shards_dir) — process-wide cache of the
~2-row rivalry-relation set; near-zero cost.
- New _load_idf_for(shards_dir, tokens) — IDF only fetched when
expansion exceeds max_total (the cap). Most queries never reach
the truncation branch and skip IDF entirely.
- Per-token process-wide neighbor cache so multi-query bench scripts
don't re-query tokens already seen.
- synonym_expand and rivalry_excluded refactored to use the lazy
loaders. Eager _load_indices / _get_indices kept for any
back-compat caller; not used by hot paths.
- invalidate_cache() clears all three caches.
All 14 concept tests pass unchanged — the contract is preserved.
Re-profile of `who wrote virt-back?` against ~38 GB of real shards
(warm cache):
metric pre-fix post-Phase-1 post-lazy-concepts
wall_ms 14,500 13,400 9,300 (-36%)
search_ms 9,900 10,600 5,000 (-49%)
SQLite executes 10,623 3,687 3,708 ~same
synonym_expand 2,966 2,840 ~0 (lazy hit)
Search target was <5 s; we hit 5.0 s on the warm path. Cold cache
should drop further (the 290 K-row dump was disk-bound).
## Real-shard baseline artifact (Phase 2)
bench/scripts/real_shard_baseline.py — runs an 8-question fixture
through the full query() pipeline and emits:
- bench/results/real-shard-baseline.json (durable; commit_sha,
shards_fingerprint, per-query rows, summary)
- bench/results/real-shard-baseline.md (human-readable summary)
Question set in bench/fixtures/real-shard-baseline-v1.jsonl:
virt-back, France, Mac OS X, Linux, Microsoft, AMD/Intel
(rivalry path), and two canonical-projection cases (math + logic
preflight short-circuit).
First baseline run (commit e78814c plus this fan-out, BURN=1):
audit_mode n notes
STRICT 4 (virt-back, France, Mac OS X, Linux)
HYBRID 2 (Microsoft founder, AMD/Intel)
CANONICAL 2 (0.1+0.2, A IMPL B; <1 ms each)
wall median 4.1 s (range 0.6 ms – 7.2 s)
primary used 4 / 8
`who wrote virt-back?` lands at 6.3 s wall, audit STRICT, primary
source #1, cited evidence still includes a copyright footer
(reviewer's warrant-quality finding — deferred to follow-up ticket
since the latency fix was the gating concern).
Hard constraint preserved: baselines NEVER gate CI. The artifact
is for confirming a fix moved the needle, seeding ForkScore
comparisons, and noting findings worth tickets.
`make bench-real-shard` wires it. Honors ARBORIST_SHARDS_DIR.
## Test status
Full suite: 1306 passed, 36 skipped (no regressions from the
concept refactor; 14 concept tests cover the lazy/eager
equivalence).
Closes the math half of fox's 2026-05-08 roadmap question ("did we
implement the math and logic stuff?"). Until now arborist had logic
KERNELS (Syllogism, Truthtables, Transitivity sub-batteries with
deterministic evaluators) but no math π*'s. SQD whitepaper §14
framed math as "operations over π*-canonical invariant objects:
integers, algebraic expressions, proof states, constraint graphs"
— that framing is now actually implementable.
arithmetic@v1 — exact rational arithmetic (SQD §14.1)
-----------------------------------------------------
Parse arithmetic expression → fractions.Fraction → canonical
"<num>/<den>" in lowest terms. Decimals via Decimal(str(...)) for
exact rational interpretation:
"0.1" → "1/10" (not 0.1±ε)
"0.1+0.2" → "3/10" (the SQD-canonical floating-point question)
"1+2" → "3/1"
"(1+2)*3" → "9/1"
"6/4" → "3/2" (lowest terms via Fraction invariant)
"2**3" → "8/1"
"-(1+2)" → "-3/1"
Rejects: identifiers, function calls, division by zero, non-integer
exponents, boolean literals. PiStarError with explanatory message.
logic-kernel@v1 — propositional CNF canonicalizer (SQD §14.3)
-------------------------------------------------------------
Parse Boolean expression (AND/OR/NOT/XOR/IMPL/IFF + named atoms +
parens) → eliminate IMPL/IFF/XOR → push NOT inward (NNF) →
distribute OR over AND (CNF) → dedupe + sort literals + sort
clauses + drop tautological clauses → serialize.
Equivalence classes preserved:
A AND B ≡ B AND A (commutativity)
(A AND B) AND C ≡ A AND (B AND C) (associativity)
A IMPL B ≡ NOT A OR B (IMPL rewrite)
A IMPL B ≡ NOT B IMPL NOT A (contrapositive)
NOT (A AND B) ≡ NOT A OR NOT B (De Morgan)
NOT NOT A ≡ A (double negation)
A OR (B AND C) ≡ (A OR B) AND (A OR C) (distribution)
A OR NOT A → TRUE (tautology)
A AND A ≡ A (idempotence)
Atom cap: N=8 (256 max clauses). Larger inputs raise PiStarError;
CNF blow-up is exponential in atom count, the cap keeps
canonicalization deterministic in bounded time.
Surface
-------
- arborist/pi_star/arithmetic.py — full implementation (new file)
- arborist/pi_star/logic.py — full implementation (graduates from
stub; preserves the LogicKernelV1 dataclass shape so the registry
key arithmetic@v1 / logic-kernel@v1 stays stable)
- arborist/pi_star/__init__.py — imports arithmetic to auto-register
- bench/batteries/base.py — PHASE_1_CARRIERS adds "arithmetic" + "logic"
- bench/fixtures/5s/syntax-arithmetic-v1.jsonl (12 fixtures)
- bench/fixtures/5s/semantics-arithmetic-v1.jsonl (15 fixtures)
- bench/fixtures/5s/syntax-logic-v1.jsonl (12 fixtures)
- bench/fixtures/5s/semantics-logic-v1.jsonl (18 fixtures)
- Makefile: bench-5s-arithmetic, bench-5s-logic-kernel,
bench-5s-math aggregate
- tests/test_pi_star.py: 30 new tests
- 13 for arithmetic@v1 (decimals, equivalence, rejects, idempotency,
integer exponents, negative results, the SQD 0.1+0.2 case)
- 17 for logic-kernel@v1 (commutativity, associativity, all rewrite
rules, De Morgan, double-neg, distribution, tautology, idempotence,
contrapositive, atom cap, syntax errors, idempotency)
Cross-modality discipline now spans:
text + claim_lattice + memory + code + arithmetic + logic
— six carrier domains, all with real π* implementations.
Two stubs remain (time-series-quantized@v1, tabular-pinned@v1);
neither is needed for math/logic coverage.
Full suite: 1256 passed, 36 skipped (+30 tests, +57 fixtures).
Completes the bridge: every 5F sub-battery now has an embedded
(Phase 1a) AND a live (Phase 1b.2) path. Per-task detail.source
distinguishes synthetic from live signal.
| Sub-battery | Live surface | Live fixtures |
|---|---|---|
| Formulate | qa.parse_claims.parse_pointer_claims | 15 (prior) |
| Feedback Loop | store.append_audit + memory.snapshot (temp shard) | 12 (prior) |
| Function | parse_pointer_claims + shape evaluators | 12 |
| Finetuning | selfmodel.store_snapshot + claims_for round-trip | 10 |
| Falsification | qa.verify.verify_quotes (real verifier) | 12 |
Function live: input_text routes through real parse_pointer_claims;
the same Phase-1a shape evaluators (shape_match / pointer_set_match
/ threshold_on_metric) run on the live-parsed output. Tests
parser→shape pipeline against actual organism behavior.
Finetuning live: gated via fixture's "live": true flag. Runner
writes parent + child SelfModel + capability_claim to a fresh temp
shard via real arborist.selfmodel.store_snapshot, reads back via
claims_for, runs the same improvement check. Tests the SelfModel
persistence + claim-attach surface, not just static fixture data.
Falsification live: answer_text + context routes through real
arborist.qa.verify.verify_quotes. Live signals translate the
verifier's audit_mode + verifier_method + unverified_quotes into
flat tags (UNGROUNDED, STRICT_<method>, HYBRID_<method>,
UNVERIFIED_QUOTE) the runner matches against expected_reason. One
fixture (5f-fal-live-003) deliberately documents a known soft-
signal gap — the entity strategy treats Insulin/Penicillin claims
as equivalent because both share "Alexander Fleming". Production
catches it via title-relevance + claim-lattice verifier; the
fixture pins the soft-path limit so future verifier changes
re-trigger review.
Surface:
- bench/batteries/b_5f.py — _live_function_produced,
_live_finetuning_measure, _live_falsification_violations
helpers + two-mode dispatch in run_function / run_finetuning /
run_falsification.
- bench/fixtures/5f/{function,finetuning,falsification}-live-v1.jsonl
(12 + 10 + 12 fixtures).
- Makefile: bench-5f-{function,finetuning,falsification}-live
targets; bench-5f-live aggregate now covers all five sub-batteries.
- Tests: 9 new in tests/test_bench_batteries.py
- live + embedded paths for each sub-battery (6 tests)
- direct helper tests verifying the real arborist surfaces are
invoked, not stubs (3 tests)
Phase 1a fixture digests unchanged. _DEFAULT_FIXTURES still points
at synthetic Phase-1a fixtures so `runner --all` behavior is
identical and deterministic; live mode invoked via explicit
--fixtures path or make targets.
Full suite: 1210 passed, 36 skipped.
5F battery is now the first to bridge synthetic→live across every
sub-battery. The pattern + live fixture format are reusable for
5R Phase 1b.2 (when SelfModel-backed workspace ops want live
verification).
Second live wire-up (after Formulate). run_feedback_loop now supports
both modes:
- Embedded (Phase 1a): chain of (operation, observation) string pairs;
runner aggregates observations and string-matches expected_delta.
- Live (Phase 1b.2): live_chain of typed ops applied to a fresh temp
arborist shard via real append_audit + memory.snapshot +
selfmodel.snapshot. expected_delta is a dict of predicates against
the resulting audit_events / memory_branch_summaries.
The live helper _live_feedback_chain creates a tempfile-backed shard,
runs the chain through real arborist surfaces, and queries the final
state. Every audit event chains via the production append_audit, so
the audit chain is re-verifiable after live execution (see new test
test_5f_live_feedback_chain_audit_chain_intact).
Three predicate types in expected_delta:
- audit_event_type_present: named event_type appears in audit chain
- memory_branch_present: named branch_id in memory_branch_summaries
- body_substring_present: substring appears in any audit body JSON
Surface:
- bench/batteries/b_5f.py — _live_feedback_chain helper +
_live_delta_satisfied predicate checker; two-mode dispatch in
run_feedback_loop
- bench/fixtures/5f/feedback-loop-live-v1.jsonl — 12 live fixtures
exercising providence_write, providence_repair, memory_snapshot,
selfmodel_snapshot ops. Includes 2 negative fixtures testing the
predicate checker (expected_delta absent → expected:fail).
- Makefile: bench-5f-feedback-loop-live + bench-5f-live aggregate
for all 5F Phase-1b.2 live wire-ups.
- Tests: 5 new in tests/test_bench_batteries.py
- live path runs all 12 fixtures
- embedded path still works (10 Phase-1a fixtures)
- helper directly tests audit-event write
- rejects unknown live op type
- audit chain re-verifies after live ops
Two of the five 5F sub-batteries now bridge synthetic → live
(Formulate + Feedback Loop). Function/Finetuning/Falsification
follow in subsequent commits.
Full suite: 1201 passed, 36 skipped.
First sub-battery to bridge from synthetic gold output to actual
organism behavior. run_formulate now supports two fixture modes
selected per-task:
- Embedded (Phase 1a): produced_lattice in the fixture. 10 seed
fixtures continue to pass via this path.
- Live (Phase 1b.2): only input_text in the fixture; runner calls
arborist.qa.parse_claims.parse_pointer_claims(input_text) and
matches the live output against expected_lattice.
Embedded takes precedence if both fields are present. Per-task
detail.source ("embedded" | "live") surfaces in bench output so
synthetic vs live signal is distinguishable.
Surface:
- bench/batteries/b_5f.py — _live_produced_lattice helper +
two-mode dispatch in run_formulate
- bench/fixtures/5f/formulate-live-v1.jsonl — 15 live-mode fixtures
with input_text + expected_lattice (no produced_lattice)
- Makefile: bench-5f-formulate-live target
- Tests: 4 new in tests/test_bench_batteries.py
- live path routes through real parser, all 15 pass
- embedded path still works (10 Phase-1a fixtures)
- _live_produced_lattice helper directly verifies parse output
- fixture missing both fields fails cleanly with explanatory reason
Phase 1a fixture digests unchanged. _DEFAULT_FIXTURES still points
at formulate-v1.jsonl so `runner --all` behavior is identical;
live-mode fixtures invoked via explicit --fixtures path.
Full suite: 1196 passed, 36 skipped.
Pattern set. Function/Finetuning/Falsification/Feedback Loop
follow in subsequent commits.
Phase 2 of #000021. React/Rearrange/Restore/Replicate/Resonate over
the workspace surface — selfmodel_records (#000014) + memory_records
(#000017), both landed earlier today. Closes the gap that gated 5R
since the substrate work shipped.
Sub-battery semantics (per SQD whitepaper §9.3 + ticket #000021 §4.2):
- React: incorporate new fact/constraint. Workspace = (snapshot_t0,
snapshot_t1, expected_delta). Pass = added_facts present + removed_facts
absent in t+1.
- Rearrange: restructure without semantic shift. Re-canonicalize
different surface forms through a named π*; pass = bytes match
expected_equivalent flag. Tests the order-invariance contracts in
SelfModel (capability_claim_hashes sorted) and Memory (branches
sorted by branch_id).
- Restore: retrieve prior fact. Workspace = (history[], current_facts[]).
Pass = fact in current OR any historical snapshot.
- Replicate: independent canonical encodings via π*. Same input run
N times must yield byte-equal output. Tests determinism contract.
- Resonate: variance across N runs. Deterministic π*'s yield
distinct=1; expected_max_distinct=1 enforces zero-variance contract.
Surface:
- bench/batteries/b_5r.py (5 deterministic runners; no LLM-as-judge)
- bench/fixtures/5r/{react,rearrange,restore,replicate,resonate}-v1.jsonl
(30 each = 150 new fixtures)
- runner.py registers 5r in _BATTERIES + _DEFAULT_FIXTURES
- Makefile: bench-5r + bench-suite (5S+5T+5F+5R aggregate)
Final tally:
5S syntax/semantics/syllogism/synthesis/semiotics 108
5T transfer/transfer-learning/triangulation/... 154
5F function/finetuning/falsification/... 50
5R react/rearrange/restore/replicate/resonate 150
TOTAL: 462 fixtures across 21 sub-batteries — 100% pass.
Tests: 6 new in tests/test_bench_batteries.py + adjustment to
test_session_integration.py for the 312→462 count + 5R sub-battery
presence assertion. Full suite: 1192 passed, 36 skipped.
Closes#000021. Phase 3 (external-corpus expansion) remains open
under the ticket but does not gate closure — the complete
Dav1DPrometheus surface is now executable infrastructure.
Per fox's 2026-05-08 review of fbd99a8: implement adaptation_efficiency
and feedback_efficiency in run_finetuning + run_feedback_loop with
explicit sentinels — not Python floating-point accidents.
_efficiency(gain, cost) helper:
- cost > 0: standard ratio
- cost == 0, gain > 0: EFFICIENCY_INFINITE (free improvement)
- cost == 0, gain == 0: EFFICIENCY_UNDEFINED (= 0.0; no signal)
- cost == 0, gain < 0: -EFFICIENCY_INFINITE (free regression)
Battery-level metrics report mean_finite (computed over finite
values only) + infinite_count + neg_infinite_count so the mean
stays dimensionally truthful and consumers can pivot on the special
buckets separately.
Phase 1a cost proxies:
- run_finetuning: _capital_cost_delta sums resource_budget
(max_compute_ms_delta * 1e-3 + max_storage_delta_bytes / 1e6).
Phase 1b.2 will replace with real capital_ledger reads.
- run_feedback_loop: chain length = cost. Phase 1b.2 capital_ledger
integration replaces it.
Tests added (7):
- _efficiency over four boundary cases
- 5F finetuning + feedback_loop emit the new metrics keys
- Synthetic zero-cost finetuning fixture verifies +inf path
Full suite: 1110 passed, 36 skipped.
Closes the one actionable from the fbd99a8 review.
Per fox's "partial punt on larger ones" — ships the bench/ skeleton +
small seed fixture sets so future v8/v7-W/SelfModel work can cite a
real fitness target. Full Phase 1 (50-200 fixtures per sub-battery)
and Phases 2-3 stay open in the ticket.
Phase 1a delivers:
- bench/batteries/{base,b_5s,b_5t,runner}.py — Battery protocol,
BatteryResult, fixture-digest helpers, CLI runner.
- Seed fixtures:
- bench/fixtures/5s/syntax-v1.jsonl — 10 tasks against
wikitext-base@v1 and claim-lattice@v1
- bench/fixtures/5s/semantics-v1.jsonl — 8 equivalence tasks
- bench/fixtures/5t/transfer-v1.jsonl — 4 paraphrase-invariance
tasks
- Runners for 5S Syntax, 5S Semantics, 5T Transfer. Other 5S/5T
sub-batteries are stubs returning zero-task results.
- Makefile targets: bench-5s, bench-5t, bench-5s5t.
- runtime_digest field captures the active π* registry fingerprint
so a registry change surfaces in bench results.
Tests: tests/test_bench_batteries.py (17 cases). Full suite:
1076 passed, 36 skipped. `make bench-5s5t` runs end-to-end and
emits JSON results.
Ticket #000021 status: in progress · Phase 1a landed; Phase 1b/2/3
remain open.
Bench-emergent stress test ran another 100 cycles under the
post-#000008/9/10/11 substrate. Total accumulated: 300 cycles.
Verdict distribution shift on last 100 vs 134-cycle baseline:
STRICT 5% (7/134) → 0% (0/100) -5pp
HYBRID 22% (29/134) → 16% (16/100) -6pp
UNGROUNDED 73% (98/134) → 84% (84/100) +11pp
Zero false-positive STRICTs across 100 random-word triplets.
The 5pp drop in STRICT-rate isn't a regression — it's the
verifier ladder + new preflight contracts doing their job.
Random-word triplets are genuinely ungrounded for the most
part; the prior 5% STRICT rate included false-positives that
the post-hardening verifier now catches.
Violation profile (last 100 cycles, claim_lattice JSON):
CITATION_MISMATCH: 86 dominant gate
TOO_MANY_EVIDENCE_IDS: 24
SUBJECT_TOKENS_ABSENT: 12 Rule 9 firing on parroting
DEFLECTION_DETECTED: 12
TITLE_MISMATCH: 10
...
metaphor_deflection fires 6/100 — still rare. Item 3
(calibration) is now closer to sample-size threshold (~30
signals across 300 cycles; needs ~50-100 to calibrate).
No new tuning candidates surface. Original three remain at
their resolution states.
Big batch — closes 4 of the 5 deferred items from the prior status
report plus opens & implements a previously-deferred design ticket
(#000011) zero-shot.
#000025 — Metacog test fixture expansion:
bench/qa_questions_metacog_subset.txt grows from 6 → 28 questions
covering edge cases per detector kind: temporal (4 cases),
contradiction (4), false-premise (5), out-of-corpus (3), multi-
trigger (2), well-formed controls (5). Documents two known
detector ceilings: Q11 over-fires on past-tense factoid
("who was the first president"); Q16/Q17/Q19 (Edison/Australia/
NASA-fake) miss false premises that lack a presupposition
pattern match. Fixture now serves as long-term regression suite.
#000026 — --show-preflight full clause render:
build_run_dag() and build_reject_run_dag() gain optional
preflight_payload kwarg. When supplied, the canonical 5-clause
CTI payload (classifier / answer_contract / prompt_contract /
evidence_contract / policy_refs + question_state + node_version)
persists alongside the leaf hash in run_dag_blob.
aborist providence --show-preflight CACHE_KEY now renders the
full payload + verifies the persisted hash matches the
recomputed canonical hash (audit-replay tamper detection).
Legacy rows fall through cleanly: payload_hash_check reports
"unavailable: legacy row predates preflight_payload persistence".
#000027 — Latency profile:
Microbenched preflight: 0.46ms/question (negligible). Single
fresh call breakdown: search 2.4s, llm 2.8s, total 5.4s — the
33-35s in Addendum 3 was vLLM concurrency contention at c=4
(per qa-modes-bench.md saturation note), not substrate
overhead. Added preflight_ms + soft_preflight_ms to timings
dict for explicit confirmation in future cycles.
#000028 — Auto-quality-check sweep revival:
scripts/bench_emergent.py running with EMERGENT_N=100 in
background (PID 125680). Will accumulate cycles into
bench/emergent_log.jsonl for #000006 rolling log re-aggregation.
Async — not blocking on completion.
#000029 — #000011 SOFT_PREFLIGHT_HINT implementation:
aborist/qa/soft_preflight.py — new module. SoftPreflightHint
dataclass + soft_preflight_question() pure function. 9
canonical labels mapping to soft analogues of #000010 hard
detectors plus 2 stub states (SOFT_DISABLED, SOFT_PARSE_FAIL).
Constrained-generation prompt (max_tokens=128, temp=0.0) asks
the model to pick ONE label + one-line rationale.
Fail-closed across every parse path:
- chat_client raises → SOFT_PARSE_FAIL
- response unparseable → SOFT_PARSE_FAIL
- label outside enum → SOFT_PARSE_FAIL
Sidecar enforces SOFT_ prefix at the normalize step so a
model that drops the prefix still gets caught.
Wired into query() between preflight & retrieval. Default
OFF (`soft_preflight_enabled: False`). NOT folded into
_VERIFIER_POLICY_FIELDS — soft hints don't gate cache
identity (#000011 §4). Audit-line tail renders as
"· soft: <label>" (e.g. "· soft: time sensitive") so the
signal is visually distinct from hard tails.
--soft-preflight CLI flag opts in per-call. End-to-end
live-verified on "When did Mr. Burns become Homer's biological
father?" — produces:
EVIDENCE-WARRANTED · via claim_lattice
· false premise · soft: time sensitive
1/1 16.4s
Hard `· false premise` (from #000010 deterministic detector)
composed with soft `· soft: time sensitive` (from #000011
sidecar). The model classified a different shape than the hard
detector — by design; soft hints are independent advisory
signals, not redundant with the hard layer.
25 new tests pin: default-OFF behavior, parse-failure modes,
label normalization (SOFT_ prefix enforced), all 8 actionable
labels round-trip, fail-closed on client exceptions, dataclass
JSON round-trip, rationale-length cap.
Other:
- #000010 §13.3 documents 2/5 metacog-trigger questions return
STRICT despite hard-detector warning — direct empirical
motivation for #000011 design.
- tests/test_dag.py extends with 3 _extract_preflight_hash_*
helper tests (cleaning #000009 §7.2 unfinished state).
- bench/emergent_log.jsonl adds new cycles from background run.
#000011 status: closed. Hard rule (D1) preserved across all
1021 tests (up from 996, +25 new). Soft preflight is purely
advisory; the verifier proof path is unchanged.
Fan-out execution of the deferred-but-not-blocking pile from
prior status reports.
#000009 §7.2 — bench harness preflight_hash field:
- aborist/qa/query.py surfaces `preflight_hash` on result dict
(miss path, reject path, and cache-hit path via new helper
`_extract_preflight_hash_from_blob` that pulls the stage hash
out of persisted run_dag_blob).
- bench/qa_sweep.py adds 12-char preflight_hash prefix to bench
rows. Mirrors cache_key truncation pattern. Operators can
grep / SQL-filter bench JSONL by preflight policy state.
- 3 new tests in tests/test_dag.py for the extract helper.
#000009 §7.2 — `aborist providence --show-preflight CACHE_KEY_PREFIX`:
- New CLI flag pulls the preflight stage payload from a row's
run_dag_blob. Match by 12-char prefix. Renders preflight stage
hash + run-DAG stage list. Operator tool for inspecting which
policy state governed a cached row.
- Live verified on a real cache row (8a212fecb2a9 — current CEO
of OpenAI question, 10-stage CTI shape with preflight at idx 1).
- Legacy rows (predating #000009) report a clean fall-through
message: "run_dag has no preflight stage (predates #000009)".
#000010 §12.6 — reference-frame plumbing into QuestionState:
- Pre-retrieval preflight runs with reference_frames=()
(frame_detection needs source titles, not available yet).
Post-retrieval, query.py re-runs preflight_question() with
the detected frames so the result-dict + run-DAG QuestionState
carry frame-aware logical_statuses (specifically
`reference_frame_ambiguous` when 2+ frames match).
- Live verified on Orwell-style question; logical_statuses now
correctly includes `reference_frame_ambiguous` in the result.
Metacog-trigger bench fixture (#000010 §13.3):
- bench/qa_questions_metacog_subset.txt — 6 questions, one per
detector kind plus a well-formed control.
- Bench artifact 2026-05-04T02-18-42Z. Detector accuracy 6/6
on fixture; 2 of 5 trigger questions return STRICT on lattice
mode despite metacog warning (JSON STRICT on
George-Washington-stop-being-president-of-France false-premise
+ uploaded-contract out-of-corpus questions). Audit-line tails
correctly surface the warnings.
- qa-modes-bench.md Addendum 4 captures the per-question matrix
+ interpretation. #000010 §13.3 cross-references with bench
artifact stamp.
#000011 SOFT_PREFLIGHT_HINT design ticket opened:
- docs/tickets/ticket-000011-soft-preflight-hint-sidecar.md
captures the design proposal per #000010 §18 / source doc.
Implementation deferred — design only.
- Sidecar would add model-assisted preflight as a soft signal
(`SOFT_FALSE_PREMISE_SUSPECTED` etc.) that NEVER enters the
verifier hard path. Strict guardrail: cannot create
PREFLIGHT_OK or PREFLIGHT_BLOCKED without deterministic
support.
- Validated by §13.3 finding: deterministic detectors flag
correctly; corpus-accidental grounding produces 2/5 STRICT
on trigger questions; soft sidecar would add independent
semantic skepticism.
- TICKETS.md index row added; Next ID bumped to 000012.
996 tests passing (3 new for the extract helper).
Cross-doc consistency:
- qa-modes-bench.md Addenda 1+2+3+4 chronological
- #000010 §13.1 (broad subset) + §13.2 (full bench) + §13.3
(metacog trigger subset)
- #000011 design captured but not implemented
Closes ticket #000010 (Meta-Cognition Preflight Guard). Mechanism
complete; defaults preserve the dry-run discipline pattern from
#000008.
Phase 2 — wire preflight into query() and runner.ask():
- preflight_question() runs after policy resolution + quantifier
classification, before retrieval.
- QuestionState surfaces on miss path, cache-hit path, AND
reject-broad early-return path of query() — schema column-
aligned across all four returns.
- runner.ask() carries the same fields for `aborist ask` parity.
Phase 3 — policy fields + governance hash + CLI flags:
- 6 new policy fields, all default-on except
metacognition_block_on_contradiction (default False — label-
only by default; opt-in via --block-on-contradiction).
- All 6 folded into _VERIFIER_POLICY_FIELDS so flipping any
invalidates prior cache records on lookup.
- 2 new CLI flags on `aborist query`:
--no-preflight Level 2 master kill
--block-on-contradiction strict mode (hard-block on
lexical contradictions)
Phase 4 — audit-line labels + bench fields + tests:
- _render_warrant_tail extended with 5 metacog tail tokens:
· false premise
· contradictory
· stale risk
· out of corpus
· frame ambiguous
- Bench rows in qa_sweep.py gain 7 new bounded-size projection
fields (logical_statuses, question_shape, preflight_result,
temporal_sensitivity, has_false_premise, has_contradiction,
corpus_requirement). Full QuestionState stays on result dict
for CLI render only.
- tests/test_metacognition.py grew from 42 → 68 tests
(16 new: 6 governance + 6 audit-line tail + 4 default-policy
pinning).
Live verified end-to-end:
$ make query-dry Q="Who is the current CEO of OpenAI?" BURN=1
UNGROUNDED · via claim_lattice · stale risk
$ make query-dry Q="When did Mr. Burns become Homer's biological
father?" BURN=1
UNGROUNDED · via claim_lattice · false premise
978 tests passing; 36 skipped.
What's NOT shipped (deferred):
- Run-DAG node binding for metacognition_preflight stage —
joins ticket #000009 Phase 5 (same audit-replay gap; both
nodes can land together).
- Reference-frame plumbing — frame_detection runs post-retrieval,
preflight here is pre-retrieval; deferred until two-pass
or post-classification update lands.
- SOFT_PREFLIGHT_HINT (model-assisted sidecar) — source doc §18
reserves this label; hard rule preserved (no LLM in preflight
hard path).
- Bench A/B measuring preflight on vs off — quick to run once
stack settles.
Ticket #000010 status: closed · landed 2026-05-03.
Caught by the 2026-05-03 dry-run distribution review across the
73-question bench set (§10.11.3 step 2):
intensity pre-fix post-fix
SINGULAR 61 (84%) 65 (89%)
MANY 4 ( 5%) 0 ( 0%) ← all 4 were `how many X?`
ALL 1 ( 1%) 1 ( 1%)
COMPREHENSIVE 1 ( 1%) 1 ( 1%)
OPEN_REQUEST 5 ( 7%) 5 ( 7%)
SMALL_NUM 1 ( 1%) 1 ( 1%)
Defect: `how many states are there?` matched the bare `\bmany\b`
pattern in MANY rung — wrong. `how many X?` is a count-question
SHAPE, asking for ONE numeric answer ("50"), not enumeration of
many things. Cap should be 1 (SINGULAR), not 8 (Hermes MANY).
Fix: count-question short-circuit in classify_question_quantifier()
that returns SINGULAR for `^\s*(?:and\s+|but\s+|so\s+)?how (?:many|much)\b`.
Anchored at start so buried `how many` (e.g. "list all the states;
how many are there?") doesn't suppress the rest of the question's
quantifier markers — the leading `list all` still wins.
9 new tests pin: count questions classify SINGULAR, leading
conjunctions don't break the short-circuit, buried `how many` does
NOT short-circuit (verifies anchor is leading-only).
Bonus — Finding 2 from the dry-run review: zero bounded universals
in bench fixture. Adds two:
name all members of the beatles
list all planets in the solar system
Both classify ALL · scope_bound_hint=bounded so the §10.1 bounded-
vs-unbounded distinction has live bench coverage. Without these,
--reject-broad correctness on bounded universals has no automated
test fixture.
915 tests passing (9 new); 36 skipped.
Adds the §10.6 + §9.6 Phase-0.x bench-row fields. Pure additive —
no row-schema renames, no policy effects yet. Phase 1 classifier
will fill the quantifier_* slots; Phase 2 cap-table will fill
claim_cap_applied. Keeping the keys present here makes the JSONL
schema stable across the rollout so post-Phase-1 markdown can
re-render against pre-Phase-1 rows without column-misalignment.
New per-row fields:
- answer_pointer_count distinct E\d+ ids in raw_answer
- answer_chars_with_brackets chars inside [E\d+,...] regions
- raw_meaningful_line_count >20-char lines in raw_answer
(matches verifier FORMAT_COLLAPSED
denominator)
- quantifier_intensity slot for Phase 1 classifier
- quantifier_matched_token slot for Phase 1 classifier
- scope_bound_hint slot for Phase 1 classifier
- claim_cap_applied slot for Phase 2 cap-table
- model_profile_id configured model id verbatim
New helper _bracket_diagnostics() bundles the bracket/pointer/line
extraction in one place; module-level regexes (_BRACKET_RE,
_BRACKET_REGION_RE, _POINTER_ID_RE) avoid per-row recompilation.
5 new tests cover the helper: empty input, single pointer, multi-
pointer-in-one-bracket, separate brackets with shared id, format-
collapsed shape (5+ meaningful lines, 0 brackets — the
2026-05-02 winners-of-all-major-sports case).
Closes the bench-side gap surfaced in §5.2: JSONL was carrying summary
numbers only, blinding the harness to FORMAT_COLLAPSED rate and per-
violation-kind distributions. Without these, A/B/D bench measurements
on the broad-quantifier preflight guard would be guesses.
- query() result dict surfaces format_collapsed + raw_answer (lattice
modes only) so the bench can read them directly instead of re-deriving
from cache rows that --burn overwrites.
- Each bench row gains format_collapsed, violation_kinds (sorted unique
list — full payloads stay off the row to keep size bounded), and
answer_brackets (count of [E\d+] in raw_answer for lattice modes).
- _summarize aggregates per-mode FC count (only explicit True; None
means check didn't apply), per-kind tallies (each kind once per row),
and lattice-only bracket sum/n.
- Markdown renderer adds a `## format-collapse + violation kinds`
section with per-mode FC rate, mean raw brackets, and one column per
observed violation kind. Degrades gracefully when the sweep produces
no violations.
- 5 new bench-harness tests pin the aggregation rules.
Re-baseline (2026-05-02T20-58-57Z) sharpens §5.1 analysis dramatically:
NO_EVIDENCE_POINTER fires 3/3 in pointer mode and is the dominant gate,
not TITLE_MISMATCH (1/3) as §5.1 inferred from JSONL alone. FORMAT_
COLLAPSED actually fires 1/3 — not the rare corner the first baseline
called it. Implies Option B (prompt reminder) is the load-bearing fix
for the verdict; Option A (cap reduction) only moves secondary kinds.
§5.3 sub-investigation closed on first read — SCHEMA_INVALID:1 in
pointer mode is a legitimate kind emitted by verify_claim_lattice for
empty-claim-text (verify.py:1242) and bare-name-claim (verify.py:1270),
not a JSON-mode leak.
Sister rule to Rule 9 (SUBJECT_TOKENS_ABSENT) landed in the same
session. Both demote STRICT → HYBRID but on orthogonal signals:
Rule 9 catches premise-parroting; FORMAT_COLLAPSED catches
protocol abandonment.
Surfaced by fox's "winners of all major sports?" 2026-05-02 case:
Hermes-3-8B melted under an under-specified broad-quantifier
question, dumped 50+ free-form prose claims with zero [E\d+]
pointer tags. Verifier honestly returned UNGROUNDED 0/2 (parser
caught two line fragments), but operators couldn't distinguish
"tried & failed to ground" from "abandoned the protocol." This
soft-demote separates the two failure shapes at audit-line glance.
verify_claim_lattice (pointer-mode only — JSON collapse already
shows as SCHEMA_INVALID):
- count meaningful_lines (>20 chars after strip) and [E\d+ regex
matches in raw answer
- ≥5 meaningful lines AND 0 bracket tags → FORMAT_COLLAPSED
violation, soft-demote STRICT → HYBRID
- format_collapsed: bool added to verdict dict
Plumbing:
- claim_lattice_format_collapse_check_enabled: True in DEFAULT_POLICY
and DEFAULT_QUERY_POLICY
- _VERIFIER_POLICY_FIELDS in keys.py adds the field so it folds
into verifier_policy_hash
- threaded through ask() and query() call sites
CLI:
- _SOFT_DEMOTE_VIOLATION_KINDS includes FORMAT_COLLAPSED so the
audit-line ladder rendering treats it as a soft demote
- _render_warrant_tail appends "· format collapsed" tail
Bench fixture: new "under-specified 'all'" section in
qa_questions.txt with `winners of all major sports?` and rationale
about cross-model resilience signal.
Tests:
- test_format_collapsed_fires_on_bracketless_multi_line_prose
- test_format_collapsed_does_not_fire_when_pointer_tags_present
- CLI render coverage
Full suite: 781 passed (up from 776).
Open Ticket #000008 — Broad-quantifier preflight guard. Cleaner
upstream fix: detect quantifier-intensity at query layer and
apply a per-model claim ceiling BEFORE the 13-second LLM call.
FORMAT_COLLAPSED stays as the downstream catch; #000008 proposes
the upstream prevention. TICKETS.md index + Next ID 000008→000009.
Re-aggregated bench/emergent_log.jsonl at 200 cycles. Distribution
holds (79.5% UNGROUNDED, 17.5% HYBRID, 3.0% STRICT). Six STRICTs
total; five defensible (known-truth or non-relationship grounding).
The sixth — `steer/reply/correcter` — is the FIRST CONFIRMED
EVIDENCE-WARRANTED false-positive in 200 cycles. Decompressed all
10 chunks of the cited document (Glossary of language teaching
terms and ideas, 33.5K chars) and confirmed the words `correcter`,
`steer`, and `reply` appear NOWHERE in the source. Yet STRICT 1/1
verified, no violations.
Mechanism: per-claim citation-coverage Rule 5 (≥30% claim tokens
in cited span) passed on GENERIC linguistic vocabulary (language,
communication, grammar, exchange, relationships) which appears in
any linguistics article. The question-distinctive tokens
(correcter, steer, reply) rode along unverified. PREMISE_PARROTING
shape predicted in original ticket §D, exhibit A.
Headline correction: from "zero false-positives" to "one confirmed
false-positive (0.5%)."
Action item: opens design space for #000008 — PREMISE_PARROTING
detector. Lexical signal: question-distinctive tokens absent from
cited chunk while Rule 5 still passes on generic vocabulary →
SUBJECT_TOKENS_ABSENT violation, demote STRICT → HYBRID. Stays
binary, stays lexical, folds into verifier_policy_hash.
The ticket file was at docs/ticket-000006-... at top-level; the
tickets/ subdir convention landed before #6 was opened. Move into
docs/tickets/ to match every other ticket.
Amend with the 62-entry post-ticket delta. bench/emergent_log.jsonl
now holds 134 cycles total. Distribution stable vs original 72:
106 UNGROUNDED (79.1%), 22 HYBRID (16.4%), 6 STRICT (4.5%). Ten
new grounded cases all carry appropriate violation labels — zero
EVIDENCE-WARRANTED false-positives across all 134 cycles. Original
tuning candidates: Tomas-deflection resolved by a1dd330,
xxviii-STRICT kept as designed, metaphor sidecar calibration
deferred (still rare).
Reframe ticket as a rolling research log. Emergent stress-testing
is an ongoing thread, not a one-shot defect-fix; future
bench-emergent re-runs append new ## Amend sections here, and any
code-level tunings open their own tickets linking back.
TICKETS.md row reflects the rolling-research framing. Stale
"## Next ID" footer scrubbed from the ticket body (TICKETS.md is
canonical for the next-id counter).
Pins the Ticket #000007 reproduction case as a permanent bench
fixture. Stresses the FTS5 hyphen-tokenization asymmetry where
query orthography (`bi-polar`) and indexed orthography (`Bipolar
disorder`) use different hyphen conventions for the same concept.
Pre-fix the medical-condition cluster never surfaced; post-fix
EVIDENCE-WARRANTED 2/2 grounded in `Bipolar disorder` + `Bipolar`
disambiguation.
72 random-word triangulation cycles ran on 2026-05-02. Aggregate:
- 54 UNGROUNDED (75%)
- 14 HYBRID (19%)
- 4 STRICT (6%)
- 18 grounded total (25%)
Catalogs five failure/success shapes the curated bench-qa fixture
set can't surface, with one-line per category. Key finding: the
verifier-ladder + soft-demote stack is doing its job — every
HYBRID demoted via TITLE_MISMATCH / DEFLECTION_DETECTED /
CITATION_MISMATCH / TOO_MANY_EVIDENCE_IDS appropriately. Zero
EVIDENCE-WARRANTED false-positives in the sample.
Tuning candidates queued for the parallel-agent work:
1. Tomas-pulls-transceivers case escaped DEFLECTION_DETECTED
(subject anchor buried mid-paragraph; consider last 2-3
content tokens vs just last 1)
2. Roman-numeral / archaic-word triplets — accept gracefully
(system already does)
3. metaphor_deflection sidecar (added in parallel work) only
fired 1×/72; calibration pass when more samples land
Ticket is OPEN and awaiting follow-up tunings; re-run
`make bench-emergent EMERGENT_N=50` after parallel work commits
land and amend with the delta.
Status table + Next ID bumped to 000007.
Empirically motivated by the 2026-05-02 emergent log:
Q: 'How can a swallowtail butterfly, gracefully fluttering amidst
the rockiest terrain, remain undeterred by the upbraiding
winds...'
A: 'The Macleay's Swallowtail butterfly is found in Eastern
Australia including the ACT, New South Wales, Queensland...'
The model traded the metaphor for literal Macleay's-Swallowtail
taxonomic facts. Warrant passed (the literal anchor IS in cited
spans), DEFLECTION_DETECTED didn't fire (the last content token
'flight' did echo somewhere), the bench landed HYBRID 3/3 — but
the user's metaphorical question was never engaged.
Honest gap: catching this structurally requires NLI-grade
semantics, which is the verifier-semantic-gap design proposal.
Until that lands, ship a SMELL SIDECAR — purely lexical, sidecar
only, never enters the binary verifier output.
Detection rule:
1. Extract metaphor cues from the question:
- -ly adverbs (gracefully, defiantly), excluding common
-ly nouns (butterfly, italy, july) via blocklist
- -ing present participles >=6 chars (upbraiding,
fluttering, brooding), excluding common verb -ing forms
- -est superlatives >=6 chars (rockiest, harshest)
- prepositional cues (amidst, despite, against, beneath)
2. Count overlap with answer's content tokens.
3. Fire metaphor_deflection when:
cue_count >= 3 AND answer_overlap_count == 0
The threshold is conservative; the smell only triggers on
STRONGLY poetic questions with PURELY literal answers.
Wire-up:
- aborist/qa/inspect.py:diagnose_metaphor_deflection
- bench/qa_sweep.py: rows gain metaphor_deflection_kind +
metaphor_cue_count + metaphor_overlap_count
- scripts/bench_emergent.py: same fields on emergent log rows
5 new tests in tests/test_inspect.py:
- swallowtail canary case fires metaphor_deflection
- literal questions (mona lisa) return no_signal
- questions whose answer engages cues return no_signal
- common -ly nouns (butterfly, italy, july, family) filtered
- sub-threshold cue counts return no_signal
756/34 tests pass (5 new + 751 prior).
Emergent stress-test log surfaced two failure shapes the curated
bench can't reach. Reviewer (claude-opus-4-7 + fox) appended
teacher: blocks to four log entries; two of them named real
defects in the demote chain. This commit ships those fixes plus
the teacher-block annotations.
Cases reviewed (2026-05-02 emergent_log.jsonl):
✅ Menkar/sterilizers/trifle — working as designed
🟡 swallowtail/upbraided/rockiest — metaphorical deflection
🔴 cashback/widescreens/within — warrant gap on inferential claims
🔴 comeliness/fetish/investitures — topic shift earned ANCHOR-WARRANTED unfairly
Two code changes:
1) verify_claim_lattice + verify_claim_lattice_json: when EVERY
resolving claim has TITLE_MISMATCH, demote audit_mode →
UNGROUNDED. Previously HYBRID. Catches the cashback case
('widescreens offer cashback' cited to a generic Coupon
article — n_verified=1 overclaimed when the citation was
meaningless).
2) cli._ladder_rung_for_lattice: DEFLECTION_DETECTED moves from
soft-demote (cap at ANCHOR-WARRANTED) to hard-demote (cap at
POINTER-LINKED) alongside WARRANT_MISSING and TITLE_MISMATCH.
Catches the comeliness case (model fully shifted topic to
'Rock & Chips'; pre-fix the verified Rock & Chips claims
landed at ANCHOR-WARRANTED unfairly).
Test updates:
- tests/test_verify_json.py:test_verify_json_title_mismatch_demotes_to_ungrounded
(renamed; expected outcome flipped HYBRID → UNGROUNDED)
- 4 fixtures in tests/test_claim_lattice.py and 1 in
tests/test_verify_json.py: claim text augmented with 'in the
film' so the Jurassic-Park-titled fixture's title shares a
stem with the claim. These tests were testing other paths
(NO_EVIDENCE_POINTER, EVIDENCE_LINKED_PARTIAL, UNKNOWN_EVIDENCE_ID)
and the title-mismatch was incidental to the fixture choice.
Teacher blocks on the four log entries record the verdict +
failure_class + recommendation + notes, with reviewer + date.
The two ✅/🟡 entries have no recommended code change; the two 🔴
entries reference this commit as the fix.
751/34 tests pass.
scripts/bench_emergent.py + make bench-emergent + design doc.
Random word triangulation surfaces failure modes the curated
bench/qa_questions.txt doesn't reach.
Loop:
/usr/share/dict/words → random.sample(3) →
Hermes @ temp=0.8 weaves a creative question →
aborist student answers via query() →
append journey to bench/emergent_log.jsonl
(teacher review = separate manual step, fox brings entries to
Opus & gets judgment to append)
Word filter: ^[a-z]{5,12}$ after lowercasing. Skips short words
(too vague) + very long words (Hermes can't weave them).
Cadence: NOT every commit. ~20s per cycle (Hermes generator +
aborist student); N=10 ≈ 4 min, N=50 ≈ 17 min. Most cycles land
UNGROUNDED-by-corpus-design (random triplets rarely overlap with
2010-11 Wikipedia coverage); the interesting cases are STRICT/
HYBRID surprises and the verifier-disagreement cases the teacher
catches.
Teacher review is intentionally out of the bench script:
- separation of concerns: generation is automated, judgment is
contextual & needs the corpus-knowledge frame ("is this a
2010 Wikipedia gap or a substrate failure?")
- future flexibility: today the teacher is Claude Opus 4.7
in this conversation; tomorrow GPT-5 or a review committee.
Swapping teachers is a workflow change, not a code change.
Teacher output schema (appended to the same JSONL line):
teacher.match bool
teacher.audit_agreement agree|disagree|unsure
teacher.novelty_class known_truth_grounding | emergent_synthesis
| novel_claim | no_signal
teacher.score_0_5 0..5
teacher.bench_max_signal retrieval | warrant | prompt | nil
teacher.reasoning one sentence
teacher.reviewed_by model id
teacher.reviewed_ts unix ts
Smoke verified (N=2, seed=42): 41s wall-clock, both UNGROUNDED
(expected — random triplets rarely overlap 2010 Wikipedia).
Append-only log seeded with the smoke entries.
Future flag (not yet wired): --generator-endpoint &
--student-endpoint to swap LLM upstreams per role.
Full design + teacher protocol: docs/bench-emergent-design.md.
Adds --resume <jsonl-path>: read the existing JSONL, build a set
of (question, mode, sample_idx) tasks already done, skip those
in the shuffled task list, and append fresh rows to the same
file. The markdown rollup uses the union of pre-existing + new
rows. Same --seed required for the order to align across the
resumed run.
Why: a 60-90 min full bench is annoying to re-run from zero when
something interrupts (network blip, kill, kernel panic). The
JSONL has been the durable artifact for ages; the bench just
didn't know how to read it.
Implementation:
- args.resume: Path | None
- if given: parse JSONL line-by-line, populate done_tasks set,
reuse the resume_path's stem as the run stamp, set jsonl_path
+ md_path to the resumed paths
- file_mode = 'a' if resuming else 'w'
- tasks filter: drop entries already in done_tasks
- rows initialized with existing_rows so the markdown summary
sees the union
- done counter starts at len(existing_rows) so the print
countdown reflects total progress
Empirical context: the c=6 bench killed at 142/639 left a
recoverable JSONL. Future kills can resume via:
.venv/bin/python bench/qa_sweep.py \
--resume bench/qa_results/<stamp>.jsonl \
--seed 0 --concurrency 4
(Same seed reproduces the shuffle so the remaining tasks come
out in the original order.)
Concurrency sweep on the smoke (15 tasks at each c):
c=3 102s (peak throughput)
c=4 106s (4% slower; chosen for full bench — more
forgiving on single-call hiccups)
c=5 119s (12% slower than c=3)
c=6 185s (45% slower; vLLM batching ceiling)
Lesson: more concurrent ≠ faster. vLLM continuous batching has
a sweet spot around c=3-4 for this endpoint.
Previously the bench scheduled at the (question, mode) cell level —
each cell ran its n samples sequentially before the next cell
started. With n=3 + concurrency=6, that meant 6 workers each
chained 3 sequential calls, vLLM's continuous batcher saw
correlated requests, and the n=3 variance signal was confounded
with 'what was vLLM doing in this same-cell batch.'
Two coupled defects, one fix:
(1) Statistical: samples-of-the-same-cell are NOT i.i.d. when run
back-to-back. Adjacent samples share vLLM batch composition,
KV-cache locality, and queue load. The 'variance' you see
across n=3 is partly that batch's particular weather, not true
model nondeterminism.
(2) Throughput: vLLM batches diverse requests well; correlated
requests fill the batch with similar work and starve.
Cell-grouped scheduling correlated by cache_key.
Refactor: build sample-level tasks ((q, mode, sample_idx)),
shuffle with deterministic seed (default 0), submit all to
ThreadPoolExecutor. Per-cell Lock dict (defaultdict(Lock))
serializes burn+insert against the shared cache_key — two samples
of the same cell that happen to land in adjacent worker slots
will queue on the cell's lock instead of racing. Lock contention
is rare under shuffle (samples of one cell are spread across
time) so throughput cost is near-zero.
New CLI flag --seed N for reproducible task ordering.
Changes:
- tasks list at sample granularity
- rng = random.Random(seed); rng.shuffle(tasks)
- cell_locks = defaultdict(Lock); held over _run_one + sample_idx assignment
- print line shows '#i/n' so order-of-completion is visible
Smoke at --n 1 --concurrency 6: clean, no errors, ~3 min for 15
tasks (essentially same wall-clock as cell-level scheduling at
n=1, as expected — the real win comes at n=3 where sample-level
concurrency unlocks parallelism within cells).
Bench-max sprint 1a + sprint 3 + speed audit. Five wins, none of
them traded calibration.
UTF-16 surrogate fix (sprint 1a)
================================
Hermes occasionally emits text with lone UTF-16 surrogates. Bare
.encode('utf-8') raises UnicodeEncodeError on those, which aborted
the run with no Merkle root. Two errors per lattice mode in the
2026-05-02 bench were this exact path on the 'tell me about the
roman empire' question.
Fix: errors='surrogatepass' on the four sha256 helpers that hash
model-derived text, plus the two audit-chain encode sites in
store.py for defense-in-depth (audit body could carry user text in
some flows). The hash stays deterministic because WTF-8 bytes are
reversible & unique per input.
Touched:
aborist/qa/dag.py:_sha256_hex (the loud one)
aborist/qa/keys.py:_sha256
aborist/qa/evidence.py:_sha256_hex
aborist/store.py: chain_audit_events + append_audit
Predicted Δ on next bench: +1pp on lattice modes (the 2 errors
become valid runs).
Smoke fixture (sprint 3)
========================
bench/qa_questions_smoke.txt — 5 questions, all anchor classes,
each currently failing pointer mode 100% while JSON aces 100% per
the 2026-05-02 bench. Wired as 'make bench-qa-smoke', --n 1
--concurrency 4, ~30-90s wall-clock depending on vLLM warmth. The
inner loop for prompt iteration; the full 71-question sweep stays
the scoreboard.
Smoke verified: pointer=0/5 STRICT, JSON=5/5, quote=2/5. Confirms
the gap pattern from the journal.
Concurrency default
===================
Makefile bench-qa now defaults to BENCH_QA_CONCURRENCY=4 (was
sequential). Override via BENCH_QA_CONCURRENCY=N. Combined with
the --concurrency landing in 0870af6, full sweep drops from ~107
min projected to ~51 min actual.
pytest-xdist (test-speed)
=========================
Added pytest-xdist>=3.5 to dev extras. 'make test' now uses
-n auto (= one worker per logical CPU). Measured: 36s → 10s on
the 641-test suite. 3.6× speedup, no test changes required.
Bench-max scoreboard (predicted lift from this commit alone):
+1pp lattice modes (UTF fix)
+cycle-time enabler (smoke fixture, xdist)
no calibration cost — none of the verifier checks moved.
zstandard.ZstdCompressor and ZstdDecompressor instances each carry an
internal libzstd context that is NOT thread-safe. Calling .compress()
or .decompress() on a single shared instance from multiple threads
corrupts the context and raises:
ZstdError: decompression error: Data corruption detected
Surfaced when bench/qa_sweep.py learned a --concurrency flag and ran
4 (question, mode) cells in parallel. ~19% of retrievals failed on
zstd corruption before the fix. The prior comment claiming the
singletons were 'stateless across calls — safe to share across
threads' was wrong.
Replace the module-level singletons with threading.local() caches.
Each thread reuses its own ZstdCompressor / ZstdDecompressor; no
contention across threads. Init cost is negligible vs decompression.
bench/qa_sweep.py:
--concurrency N (default 1) parallelizes (question, mode) CELLS
using a ThreadPoolExecutor. Samples within a cell stay sequential
so burn-then-insert against a single cache_key never races itself.
Lock-protected JSONL writes & progress prints. Exception in any
worker surfaces via fut.result().
vLLM handles concurrent requests well; 4-8 is a reasonable starting
point. With --n 2 --concurrency 4, expect ~10 min wall-clock for the
full 426-run sweep against hermes.ai.unturf.com (vs ~2h sequential).
Two paired changes addressing fox's "learn this hyperparameter
from model use, not hard coding" + the 1M-context-window
caveat:
(1) Log-scale prompt-size buckets extend from 8KB through 1M+:
<8KB / 8-16KB / 16-32KB / 32-64KB / 64-128KB / 128-256KB /
256-512KB / 512K-1M / >=1M
The same bench harness now covers 8B-class models (Hermes 82K
context, max useful prompt ~32-64KB) through 1M-context models
(Gemini 1.5 Pro, Claude with extended context, Llama 4) without
code change. A model whose context window stops at 82K simply
never populates the giant buckets; a 1M-context model fills
them and finds its own sweet spot.
(2) New "recommended context budget (learned from this bench)"
section — per-mode peak-strict-rate bucket. Operator-driven
landing per the five-step algorithm step 5: surfaced, not
auto-applied. Minimum sample size of 5 runs per bucket so
statistical noise doesn't masquerade as signal. Tie-break on
smaller-bucket-wins so equivalent strict-rates favor the
cheaper choice.
The substrate is now self-tuning at the OBSERVATION layer: bench
records what budget actually grades best per model. Per-model
profile JSON (storing the recommended budget back into
~/.aborist/model_profiles/<model>.json) is the next beat once
this surface is observable in real bench runs.
3 new bench tests cover the recommended-budget section, the
sample-size floor, and the giant-context bucket coverage.
Full suite: 746 passed (was 738, +8).
Connects to D8 (automate after test-pinning): the bench tells us;
we don't guess.
Wires the seven-point program (docs/seven-point-program.md) into
the bench harness as a per-mode coverage table.
Per-row computation in _directive_compliance(answer_mode, result, err):
D2_pointer_clauses — answer_mode is lattice variant
D3_cti_substrate_ready — lattice mode + run_dag_root populated
(full coverage pending ticket #000002)
D4_evidence_map_bound — run_dag_root populated
(retrieval-plan binding pending #000001)
D6_warrant_fired — warrant ran (lattice mode today;
per-shape gating pending #000003)
D7_honest_label — audit_mode in canonical enum
(renderer transformation pinned in
test_cli_render.py)
D1 (no LLM in verifier), D5 (verifier_method enum), D8 (test-pinning
discipline) are global properties of the substrate and don't appear
per-row — they get tracked once in CLAUDE.md / the program doc.
_summarize aggregates per-row booleans into per-mode pass counts.
_render_markdown emits a 'directive coverage (seven-point program)'
section with per-mode per-directive coverage as 'count/N (pct%)'.
Discipline rationale per CLAUDE.md's five-step algorithm step 5:
the bench harness IS the automation substrate, so it must enforce
the directives before any feature gets layered on top. Bench scores
that climb without directive coverage climbing are graveyard-digging
(step 4: don't dig the grave faster).
5 new bench tests cover: per-mode pass aggregation, the markdown
section presence and content, the helper's behavior on quote mode
(D2 fail by design), lattice mode (all directives pass), and error
rows (empty dict — no signal). Full suite 669 passed.
