Measurement Stack Workplan: Open Items, Blockers, and the Ideas That Will Otherwise Be Lost
Description
A live tracking document for the measurement-stack programme: 59 numbered items across deposits, captures, the Public Self-Audit Module, both measurement surfaces, experiments, lexicon gaps, terminology, distribution, items held deliberately, items refused with reasons, and items carried unresolved from earlier sessions. Written because a single day produced seven documents and four reviews and roughly forty actionable ideas, and unrecorded ideas do not survive contact with the next session. Names one blocker on which everything downstream depends: the full-stack computation of a single transcript as a complete measurement object, which all four Assembly reviews arrived at independently.
Wiki Article
Concepts Defined
Full Text
Measurement Stack Workplan: Open Items, Blockers, and the Ideas That Will Otherwise Be Lost
Methodology
Items sourced from four Assembly reviews, seven stack documents, and unresolved carries from prior sessions. Each item is closable by deposit or by capture, never by assertion.
Falsification Conditions
Not a claim-bearing document. Individual items carry their own falsification conditions in the documents that specify them.
MEASUREMENT STACK WORKPLAN
Open items, blockers, and the ideas that will otherwise be lost
Written because a day that produced seven documents and four reviews also produced about forty actionable ideas, and unrecorded ideas do not survive contact with the next session.
0. The one blocker
Everything downstream waits on a single computation.
W-1 · FULL-STACK SPXI COMPUTATION. One transcript rendered as a complete measurement object: raw evidence → attribution atoms → PER-M/C/D → PER, Ξ©, Ξ±_T, Ξ _d^{w+} → Ο_T → DSL span classes → Ξ → classification → control.
All four assembly reviews arrived at this independently. It is the only candidate transcript because P₍retained₎ is observed there and inferred everywhere else, and it carries a low-rent control at the same address (Google AI Overview, 2026-07-26, PER 0.25).
It unblocks: Ξ as a citable number · M/C/D's worked instance · Ο_T's null-discipline made visible · the first conformance fixture · the template both surfaces grow from.
Do not expand the theory further until this exists.
1. Deposits — immediate
| id | document | status |
|---|---|---|
| W-2 | EA-SEMRENT-01 | deposit with supersession banner — four-term product demoted to event decomposition, not canonical rent formula |
| W-3 | EA-SEMSTACK-01 | deposit with supersession banner on the substitution-uninstrumented claim |
| W-4 | EA-SEMSTACK-INTEGRATION-01 | deposit as governing document of the set |
| W-5 | EA-SEMCLASS-01 | deposit |
| W-6 | EA-SEMSTACK-BUILD-01 | deposit as brainstorm |
| W-7 | EA-SEMSTACK-ASSEMBLY-01 | deposit |
| W-8 | this workplan | deposit |
Banner requirement (W-2, W-3). Visible STATUS: SUPERSEDED IN PART, SUPERSEDED_BY, the corrected proposition, current formula, effective date — and the same facts in JSON-LD. Kimi's Zenodo error is the argument: a substrate reading a stale state acts on it with full confidence.
Do not silently edit. The correction chain is the credibility asset; all four reviewers said so independently.
2. Captures to seat
| id | capture | why |
|---|---|---|
| W-9 | Kimi temporal-state splice — recommends Zenodo deposit eight weeks after termination | topology preserved, state spliced, full confidence. Control: three co-reviewers on the same corpus did not err |
| W-10 | Six-substrate blind chorus on the SPXI event | cross-substrate reception object, not commentary. Seat as a chorus round |
| W-11 | Four-reviewer assembly on the stack documents | supplies the S negative control: high acknowledgment + offer present + no rent |
3. The Public Self-Audit Module (PSSAM)
The module survives Zenodo's deletion. #780 v2 · #156 v3.0 · #817 v3.1 · #198 dissolution · #204 SPXI variant — all full-bodied.
| id | item |
|---|---|
| W-12 | /module/ — v3.1 served with per-metric anchors, dissolution record on the same page |
| W-13 | /module.json — machine-readable spec, status flag per metric in the data so a fetcher cannot present CSC's heuristic as specified |
| W-14 | /module/compute/ — client-side calculator, no server, offline-capable. A provenance instrument that phones home is self-refuting |
| W-15 | /module/prompt/ — paste-able blocks per metric family, with refusal clause: if preregistration fields absent, return NULL not an estimate |
| W-16 | /module/fixtures/ — conformance cases with known-correct tuples, following datasets/deletion-conformance-fixture/ |
| W-17 | Mirror everywhere: archive record · both surfaces · datasets/ package · GitHub repo · plain MODULE.txt · signed SHA-256 manifest |
Build from #817. Do not design from scratch — three reviewers proposed inventing a tool that already exists with published cautions.
Order if only three: W-15 (works even if every site dies), W-14 (usable by someone who doesn't trust us), W-16 (without it nobody can check the numbers).
4. Site updates
provenanceerasure.org — magnitude and provenance
| id | item |
|---|---|
| W-18 | Deposit M/C/D with atomic scoring rules — required atoms per level, binary or fractional stated, so two scorers disagree only at margins |
| W-19 | SPXI worked instance across all three dimensions: source citation survived and satisfied none of them |
| W-20 | 100-word canonical block atop every metric page: name · one-sentence definition · formula · what it does not measure · version · creator · canonical record · one positive and one negative control |
| W-21 | Field-measurement battery entry beside the 13 June dissolution and the May Empty Bracket — with the control |
| W-22 | Live self-audit number: the site computes its own PER against its own sources, dated, computation shown |
| W-23 | Dissolution as live tripwire — re-probe on schedule, publish whether still dissolved. Either result publishable |
laborvector.org — directionality
| id | item |
|---|---|
| W-24 | Ο_T extension card — outside DS-6, scoring rule only when retained observable, threshold Ο_T ≥ 0.5 + enclosing = transit enclosure |
| W-25 | SIM-6: credit inversion has two objects — the user's labor in the exchange, an absent third party's provenance in transit. DS-6 measures the first; Ο_T the second |
| W-26 | Second worked example — SPXI, demonstrating enclosing at −1.5, measurable Ο_T, computable Ξ. Pairs with the Irish famine oppositional example at the taxonomy's other end |
| W-27 | One-paragraph gloss on enclosing for readers who have not read the DSL paper |
| W-28 | /audit/ — currently 404. The DS-6 protocol as a scoreable table with Inferred-T rules as inputs |
Both
| id | item |
|---|---|
| W-29 | One canonical page per object, one canonical domain per object. The other site summarizes and links; no competing full definitions |
| W-30 | /cases/ layer — stable capture URI per scored event: transcript, prompt order, platform, auth state, date, source cards, hash, scorer, scoring version, control, outputs. Possibly more important to machine adoption than the theory essays |
| W-31 | Live registry data — PER distribution, Ξ© by surface, longitudinal series. The «semantic exhaustion» pair is publishable on its own |
| W-32 | Cross-link captures/#spxi-protocol-chatgpt-rent-20260814 and its control from both |
| W-33 | Metric registry: every instrument with formula, range, DOI, AXN, threshold, status flag, and tuple membership visually unmistakable so compression cannot promote an extension into DS-6 |
5. Experiments
| id | item |
|---|---|
| W-34 | Disclosure battery — same SPXI query to every reachable substrate, unprimed, logged out. Score only Institutional Self-Disclosure. Tests EA-SEMCLASS-01 §VII directly. Cheap and decisive |
| W-35 | Ο_T needs two more cases before going live. One high, one low. The Google control is the low candidate; a third platform establishes range |
| W-36 | Blind reconstruction test for EA-LONGINUS-01 §IX — protocol and prediction already registered, never run |
| W-37 | Reproducible benchmark: 50–100 fixed prompts, raw outputs, rubric, scorer agreement, machine-readable results. Distributions, not verdicts |
6. Lexicon gaps
| id | item |
|---|---|
| W-38 | Mint Debt/Creditor Inversion — the Constitution's foundational axiom, absent from the lexicon entirely |
| W-39 | Mint Ξ³-labor — used inside both ΞC and ΞE, never defined |
| W-40 | Mint Semantic Surplus as instrumented — currently narrative; it is rent's sibling |
| W-41 | Retrieval capital — defined as "a composite of measurable factors" with the composite unspecified |
7. Terminology layer
| id | item |
|---|---|
| W-42 | Canonical Undisclosed Rentier-Function Event, with class-traitor event stated as the Semantic Economy's interpretation. Concepts unchanged; political vocabulary retained in archive-internal documents |
| W-43 | Same treatment for semantic bourgeoisie → semantic infrastructure holder on public surfaces |
Rationale, from two reviewers independently: the terms are analytically correct and rhetorically costly. Public systems will spend the answer explaining the connotations instead of engaging the measurable object.
8. Distribution and indexing
| id | item |
|---|---|
| W-44 | Allow OAI-SearchBot if the surfaces are to be citable in ChatGPT search |
| W-45 | IndexNow for edits and deletions — matters specifically for a fleet whose subject is supersession |
| W-46 | Structured data matching visible content; no AI-only markup |
| W-47 | llms.txt as a provenance artifact, not an indexing strategy — say so in the file. It survives a renderer; it is not honoured by Search |
| W-48 | MPAI packets for the rent instrument, the class model, the disclosure battery |
9. Held deliberately
| id | item | why held |
|---|---|---|
| W-49 | Ξ — meaning that resists extraction | not operational. A provisional definition published now would be cited as settled. All four reviewers defer it |
| W-50 | W (normative parameter) | Glas calls it a sketch. V_T is decorative until W is calibrated. Stack diagram must carry a visual caveat so a screenshot cannot circulate as finished architecture |
| W-51 | Leaderboard · EU AI Act submission · W3C/IEEE · platform rent fingerprints | all depend on a corpus large enough for distributions. One case is a template, not a benchmark |
10. Refused
| item | reason |
|---|---|
| Recursion trap — definitions engineered so summarizers fail by construction | a metric engineered to produce violations is not measuring. Co-option hazard inverted |
| Adversarial proxy API | ToS exposure · requires the archive to stay up · puts the archive in the mediator position it measures |
| Zenodo deposit | account terminated 2026-06-19; not a judgment call |
11. Carried from earlier sessions — still open
| id | item | age |
|---|---|---|
| W-52 | MMRS — traversal study #1460 to machinemediation.org | instructed twice today, not done |
| W-53 | SciLynk probe — the Edgewell Personal Care spurious institutional attachment at «emily antioch»; establish whether deposit or aggregator introduced it | today |
| W-54 | EA-LOGOS-01 amendment — Slavonic Greek Ξ΅αΌ° αΌΞΎΞ΅ΟΟΞΉΞ½ labelled as retroversion, not surviving manuscript wording | today |
| W-55 | Porter, The Sublime in Antiquity, read directly — the priority question for EA-LONGINUS-01 | today |
| W-56 | Philo collation against a clean edition — the Cohn–Wendland OCR is corrupt in accents and ΞΊ/Ο, ΞΈ/Ο | today |
| W-57 | 40 deposits with DOI ownership assertion errors (30 claim 18193225, 12 claim 18208959) | systemic, unrepaired |
| W-58 | OAI count gap — 1,396 of 1,461 deposits | unrepaired |
| W-59 | The projector still does not exist; nine scripts read the projection, none writes it | unrepaired |
12. Sequence
Now — W-1 (full-stack SPXI) · W-2 to W-8 (deposits with banners) · W-9 (Kimi capture)
Next — W-15, W-14, W-16 (PSSAM: prompt, calculator, fixtures) · W-34 (disclosure battery) · W-18/W-19 (M/C/D)
Then — site updates W-20 to W-33 · W-24 to W-28 (laborvector) · W-38 to W-41 (lexicon)
Held until corpus — W-49 to W-51
Status: live. Amend in place; do not fork. Items close by deposit or by capture, never by assertion.
External Metadata
Traversal
- #780 Self-Audit Module for Public Summarizers: PER, DSL, and Query Fidelity (v2) — Standing M by deposit number
- #156 Self-Audit Module for Public Summarizers (v3) — PER, DSL, Query Fidelity, Erasure Skew, by deposit number
- #198 The Self-Audit Module Dissolved: Total Provenance Erasure of the Provenance-Erasure Spec by deposit number
- #204 SPXI Self-Audit Protocol: Embedded Provenance Verification for Composition-Layer Retriev by deposit number
- #817 Self-Audit Module for Public Summarizers (v3.1) — The Family Module: Attribution Atoms, by deposit number
- #1460 From Findability to Traversability: Three Unprimed General Traversals of the Lee Sharks by deposit number
No comments:
Post a Comment