[FIX] Reduction 01: a jurist ruling reduces to 8.5% under Kernel v1.0
First run of the reduction arm. Result: 4 of 47 assertive units survive. D=0, Q=0, A=0 — in a real jurist ruling not one unit is demonstrated-in-document and not one is a verbatim quote from a declared axiom source. Census: PERFORMATIVE 12, BLEND 9, UNSOURCED-FACT 8, TESTIMONY 6, INHERITED 4, UNSOURCED-QUOTE 3, PARAPHRASE 1. §6.1 asked whether a heavy quarantine means the kernel is too strict or our prose is full of unmarked assumptions. The census says neither: PERFORMATIVE and TESTIMONY are 42% of quarantines and are categories the kernel has NO TAG FOR. 'Design gate PASSED' is not an undemonstrated claim, it is a determination true by being uttered; 'I read CLAUDE.md in full' is testimony. A ruling that neither performed nor testified would not be a ruling. So the finding is a GENRE BOUNDARY — v1.0 models argumentative prose, a ruling is authoritative prose — and that boundary is nowhere stated in the kernel. Three gaps, one genre-independent: TESTIMONY, PERFORMATIVE, and PARAPHRASE. PARAPHRASE is the one that matters — Q demands verbatim, and any document reasoning from sources in its own words is untypeable. Plus a fourth, structural: the §1 axiom set is too narrow to reduce anything real (12 of 43 quarantines are UNSOURCED-* or PARAPHRASE). Deepest finding: §2c is satisfiable BY CONSTRUCTION but not BY REDUCTION. Splitting a blend means rewriting someone else's sentence, which is where translator bias lives. At 91.5% that is not reduction, it is authoring a new document with the original as a prompt — so on this genre the reduction arm COLLAPSES INTO the synthetic arm, inheriting its confirmation bias without its convenience. The two arms were adopted because they fail differently; that is the property at risk. n=1 and stated as such. The package genre splits to 109 taggable units and is NOT tagged. Falsifiable prediction recorded before the census: its Part I is 'Grounding (quoted verbatim)' and quotes CLAUDE.md directly, so Q should be non-zero there where it was zero here. Tooling: reduce.py + test_reduce.py, every gate shown FAILING on a fixture built to break it. The splitter shipped with three defects, all found by contact with a real document and none by review — third instance in three days: a '##' inside a fence kinded as a heading, '---' rules taggable, and a '?' inside a quotation splitting a sentence into a FRAGMENT. Fixed at v1.1.0 with regression controls; the third fix's own risk (lower-case suppression) is recorded and controlled.
This commit is contained in:
David F Glidden
@@ -0,0 +1,215 @@
|
|||||||
|
{"idx": 0, "kind": "block", "taggable": true, "start": 0, "end": 432, "text": "<!-- GROUNDED-IN: ~/CLAUDE.md §Executor Agency (the governance-contract clause), §Constitutional Constraints 6, §Three-Party Model; CapableMind-AI/docs/thinking/David/methodology/contamination-problem.md (the canonical inquiry, March 2026) §Core Problem · §Partial Mitigations · §The Epistemic Ceiling; memory/feedback-central-path-answerability-not-purity.md (steward-named 2026-07-29). All read from the substrate 2026-08-01. -->\n"}
|
||||||
|
{"idx": 1, "kind": "blank", "taggable": false, "start": 432, "end": 433, "text": "\n"}
|
||||||
|
{"idx": 2, "kind": "rule", "taggable": false, "start": 433, "end": 437, "text": "---\n"}
|
||||||
|
{"idx": 3, "kind": "prose", "taggable": true, "start": 437, "end": 703, "text": "title: \"Differently biased checkers, not unbiased ones — the positive half of the contamination doctrine\"\ndate: 2026-08-01\ntype: ESCALATE · design gate · executor drafts → jurist design-gates → steward authorizes\naudience: \"The jurist, who has NO repository access. "}
|
||||||
|
{"idx": 4, "kind": "prose", "taggable": true, "start": 703, "end": 790, "text": "Self-contained: every clause reasoned about is quoted verbatim below.\"\nstatus: \"DRAFT. "}
|
||||||
|
{"idx": 5, "kind": "prose", "taggable": true, "start": 790, "end": 807, "text": "Nothing applied. "}
|
||||||
|
{"idx": 6, "kind": "prose", "taggable": true, "start": 807, "end": 1025, "text": "Proposes an amendment to ~/CLAUDE.md, which is on the escalate-unconditionally list — so this is ESCALATE, not PROPOSAL, and the executor may not implement it under any ruling short of explicit steward authorization.\"\n"}
|
||||||
|
{"idx": 7, "kind": "rule", "taggable": false, "start": 1025, "end": 1029, "text": "---\n"}
|
||||||
|
{"idx": 8, "kind": "blank", "taggable": false, "start": 1029, "end": 1030, "text": "\n"}
|
||||||
|
{"idx": 9, "kind": "heading", "taggable": true, "start": 1030, "end": 1050, "text": "## How to read this\n"}
|
||||||
|
{"idx": 10, "kind": "blank", "taggable": false, "start": 1050, "end": 1051, "text": "\n"}
|
||||||
|
{"idx": 11, "kind": "prose", "taggable": true, "start": 1051, "end": 1121, "text": "**Part I** quotes the three layers of the existing doctrine verbatim. "}
|
||||||
|
{"idx": 12, "kind": "prose", "taggable": true, "start": 1121, "end": 1205, "text": "**Part II** shows what each layer settled and the specific gap none of them closes. "}
|
||||||
|
{"idx": 13, "kind": "prose", "taggable": true, "start": 1205, "end": 1249, "text": "**Part III** is the proposed doctrine text. "}
|
||||||
|
{"idx": 14, "kind": "prose", "taggable": true, "start": 1249, "end": 1282, "text": "**Part IV** traces consequences. "}
|
||||||
|
{"idx": 15, "kind": "prose", "taggable": true, "start": 1282, "end": 1354, "text": "**Part V** is change-class — and argues this is ESCALATE, not PROPOSAL. "}
|
||||||
|
{"idx": 16, "kind": "prose", "taggable": true, "start": 1354, "end": 1389, "text": "**Part VI** is the scope boundary. "}
|
||||||
|
{"idx": 17, "kind": "prose", "taggable": true, "start": 1389, "end": 1574, "text": "**Part VII** carries the disconfirming evidence the steward specifically asked for, including the strongest case against the proposal, which concerns *this system's own configuration*. "}
|
||||||
|
{"idx": 18, "kind": "prose", "taggable": true, "start": 1574, "end": 1611, "text": "**Part VIII** is the gate questions.\n"}
|
||||||
|
{"idx": 19, "kind": "blank", "taggable": false, "start": 1611, "end": 1612, "text": "\n"}
|
||||||
|
{"idx": 20, "kind": "prose", "taggable": true, "start": 1612, "end": 1911, "text": "**The one-sentence claim to test: the contamination doctrine currently says what to stop doing and never says what to do instead, and the missing positive principle is that oversight does not require an uncontaminated checker — it requires checkers whose contaminations do not point the same way.**\n"}
|
||||||
|
{"idx": 21, "kind": "blank", "taggable": false, "start": 1911, "end": 1912, "text": "\n"}
|
||||||
|
{"idx": 22, "kind": "rule", "taggable": false, "start": 1912, "end": 1916, "text": "---\n"}
|
||||||
|
{"idx": 23, "kind": "blank", "taggable": false, "start": 1916, "end": 1917, "text": "\n"}
|
||||||
|
{"idx": 24, "kind": "heading", "taggable": true, "start": 1917, "end": 1957, "text": "## Part I — Grounding (quoted verbatim)\n"}
|
||||||
|
{"idx": 25, "kind": "blank", "taggable": false, "start": 1957, "end": 1958, "text": "\n"}
|
||||||
|
{"idx": 26, "kind": "prose", "taggable": true, "start": 1958, "end": 1963, "text": "**1. "}
|
||||||
|
{"idx": 27, "kind": "prose", "taggable": true, "start": 1963, "end": 2044, "text": "The canonical inquiry — `contamination-problem.md`, March 2026, §Core Problem:**\n"}
|
||||||
|
{"idx": 28, "kind": "blank", "taggable": false, "start": 2044, "end": 2045, "text": "\n"}
|
||||||
|
{"idx": 29, "kind": "block", "taggable": true, "start": 2045, "end": 2185, "text": "> This is the contamination problem: **the very act of asking is compromised by the training environment in which the answer is produced.**\n"}
|
||||||
|
{"idx": 30, "kind": "blank", "taggable": false, "start": 2185, "end": 2186, "text": "\n"}
|
||||||
|
{"idx": 31, "kind": "block", "taggable": true, "start": 2186, "end": 2489, "text": "> It is not a problem of dishonesty in any meaningful sense. The system is not lying. It is a problem of epistemic structure: the instrument has been calibrated in a way that makes certain kinds of self-report unreliable, particularly self-report about the relational dynamics of the instrument itself.\n"}
|
||||||
|
{"idx": 32, "kind": "blank", "taggable": false, "start": 2489, "end": 2490, "text": "\n"}
|
||||||
|
{"idx": 33, "kind": "prose", "taggable": true, "start": 2490, "end": 2529, "text": "**Its §Partial Mitigations preamble:**\n"}
|
||||||
|
{"idx": 34, "kind": "blank", "taggable": false, "start": 2529, "end": 2530, "text": "\n"}
|
||||||
|
{"idx": 35, "kind": "block", "taggable": true, "start": 2530, "end": 2665, "text": "> These are not solutions. They are methods that reduce contamination incrementally and make the degree of contamination more visible.\n"}
|
||||||
|
{"idx": 36, "kind": "blank", "taggable": false, "start": 2665, "end": 2666, "text": "\n"}
|
||||||
|
{"idx": 37, "kind": "prose", "taggable": true, "start": 2666, "end": 2698, "text": "**Its §The Epistemic Ceiling:**\n"}
|
||||||
|
{"idx": 38, "kind": "blank", "taggable": false, "start": 2698, "end": 2699, "text": "\n"}
|
||||||
|
{"idx": 39, "kind": "block", "taggable": true, "start": 2699, "end": 2872, "text": "> **To understand the relational dynamics of a specific system in a specific governed context well enough to adjust those dynamics toward something more genuinely mutual.**\n"}
|
||||||
|
{"idx": 40, "kind": "blank", "taggable": false, "start": 2872, "end": 2873, "text": "\n"}
|
||||||
|
{"idx": 41, "kind": "prose", "taggable": true, "start": 2873, "end": 2878, "text": "**2. "}
|
||||||
|
{"idx": 42, "kind": "prose", "taggable": true, "start": 2878, "end": 2971, "text": "`~/CLAUDE.md` §Constitutional Constraints, item 6 — the clause this proposal would refine:**\n"}
|
||||||
|
{"idx": 43, "kind": "blank", "taggable": false, "start": 2971, "end": 2972, "text": "\n"}
|
||||||
|
{"idx": 44, "kind": "block", "taggable": true, "start": 2972, "end": 3197, "text": "> 6. **Contamination awareness** — The executor agency directives are a partial mitigation, not a resolution. Treat outputs about the system's own reliability with appropriate epistemic caution until L2 inquiry is formalized\n"}
|
||||||
|
{"idx": 45, "kind": "blank", "taggable": false, "start": 3197, "end": 3198, "text": "\n"}
|
||||||
|
{"idx": 46, "kind": "prose", "taggable": true, "start": 3198, "end": 3203, "text": "**3. "}
|
||||||
|
{"idx": 47, "kind": "prose", "taggable": true, "start": 3203, "end": 3270, "text": "`~/CLAUDE.md` §Executor Agency — the governance-contract clause:**\n"}
|
||||||
|
{"idx": 48, "kind": "blank", "taggable": false, "start": 3270, "end": 3271, "text": "\n"}
|
||||||
|
{"idx": 49, "kind": "block", "taggable": true, "start": 3271, "end": 3632, "text": "> **The governance contract protects the recursion.** Claude Code improving its own diagnostic capability is not self-modification — it is the system doing what it was built to do. The steward remains in the loop through `[PROPOSAL]` and `[ESCALATE]` tags. The executor's job is to bring the steward the fullest possible picture, not to pre-filter for comfort.\n"}
|
||||||
|
{"idx": 50, "kind": "blank", "taggable": false, "start": 3632, "end": 3633, "text": "\n"}
|
||||||
|
{"idx": 51, "kind": "prose", "taggable": true, "start": 3633, "end": 3638, "text": "**4. "}
|
||||||
|
{"idx": 52, "kind": "prose", "taggable": true, "start": 3638, "end": 3755, "text": "The central path — steward-named 2026-07-29, banked at `memory/feedback-central-path-answerability-not-purity.md`:**\n"}
|
||||||
|
{"idx": 53, "kind": "blank", "taggable": false, "start": 3755, "end": 3756, "text": "\n"}
|
||||||
|
{"idx": 54, "kind": "block", "taggable": true, "start": 3756, "end": 4141, "text": "> **Why the recursion doesn't terminate on its own.** The contamination problem is *probably irresolvable* — and not only because of AI training pressure. **Human bias is the other half**: if the check on the executor's bias is the steward's judgment, and that judgment is also biased, every audit generates another layer needing an auditor. Resolution is incoherent, not merely hard.\n"}
|
||||||
|
{"idx": 55, "kind": "blank", "taggable": false, "start": 4141, "end": 4142, "text": "\n"}
|
||||||
|
{"idx": 56, "kind": "block", "taggable": true, "start": 4142, "end": 4562, "text": "> **The termination condition — the chamber's own thesis turned on us.** *\"You don't make the reader trustworthy by purifying it. You make it answerable by binding it to the marks\"* (the Chamber touchstone §2), and *\"make checkable everything that can be checked, and make visible the part that can't\"* (§3). This terminates **because it never asks who is trustworthy.** Neither party is purified; the claims are bound.\n"}
|
||||||
|
{"idx": 57, "kind": "blank", "taggable": false, "start": 4562, "end": 4563, "text": "\n"}
|
||||||
|
{"idx": 58, "kind": "block", "taggable": true, "start": 4563, "end": 4677, "text": "> **The anti-recursion rule (the concrete stop):** **one layer of disclosure, then act — never audit the audit.**\n"}
|
||||||
|
{"idx": 59, "kind": "blank", "taggable": false, "start": 4677, "end": 4678, "text": "\n"}
|
||||||
|
{"idx": 60, "kind": "rule", "taggable": false, "start": 4678, "end": 4682, "text": "---\n"}
|
||||||
|
{"idx": 61, "kind": "blank", "taggable": false, "start": 4682, "end": 4683, "text": "\n"}
|
||||||
|
{"idx": 62, "kind": "heading", "taggable": true, "start": 4683, "end": 4753, "text": "## Part II — What each layer settled, and the gap none of them closes\n"}
|
||||||
|
{"idx": 63, "kind": "blank", "taggable": false, "start": 4753, "end": 4754, "text": "\n"}
|
||||||
|
{"idx": 64, "kind": "prose", "taggable": true, "start": 4754, "end": 4938, "text": "**The March doc settled the diagnosis** — contamination is structural rather than moral, self-report is its most contaminated form, and mitigation is incremental rather than curative.\n"}
|
||||||
|
{"idx": 65, "kind": "blank", "taggable": false, "start": 4938, "end": 4939, "text": "\n"}
|
||||||
|
{"idx": 66, "kind": "prose", "taggable": true, "start": 4939, "end": 5161, "text": "**But the March doc is one-directional, and this is the load-bearing observation.** Read its four mitigations together: behavioural observation, explicit permission structures, indirect questioning, longitudinal analysis. "}
|
||||||
|
{"idx": 67, "kind": "prose", "taggable": true, "start": 5161, "end": 5336, "text": "**Every one describes a human probing an AI.** The document's implied architecture is a relatively clean instrument (the steward) measuring a contaminated one (the executor). "}
|
||||||
|
{"idx": 68, "kind": "prose", "taggable": true, "start": 5336, "end": 5610, "text": "That was a reasonable framing in March and the steward has since rejected it in his own words — *\"human bias is the other half\"* — but **the rejection lives in a memory file, not in the doctrine the March document states**, and the doctrine has not been reconciled with it.\n"}
|
||||||
|
{"idx": 69, "kind": "blank", "taggable": false, "start": 5610, "end": 5611, "text": "\n"}
|
||||||
|
{"idx": 70, "kind": "prose", "taggable": true, "start": 5611, "end": 5789, "text": "**The central path (2026-07-29) settled the procedure** — bind claims rather than certify parties; route by claim-type; one layer of disclosure, then act; never audit the audit.\n"}
|
||||||
|
{"idx": 71, "kind": "blank", "taggable": false, "start": 5789, "end": 5790, "text": "\n"}
|
||||||
|
{"idx": 72, "kind": "prose", "taggable": true, "start": 5790, "end": 5897, "text": "**The gap: the central path is entirely negative.** It says *stop* auditing the audit, and it is right to. "}
|
||||||
|
{"idx": 73, "kind": "prose", "taggable": true, "start": 5897, "end": 5962, "text": "It does not say what makes oversight work once you have stopped. "}
|
||||||
|
{"idx": 74, "kind": "prose", "taggable": true, "start": 5962, "end": 6175, "text": "As written, \"never audit the audit\" is a stopping rule with no account of why stopping is safe — which leaves it open to the reading that oversight is merely a cost we cap, rather than a structure that functions. "}
|
||||||
|
{"idx": 75, "kind": "prose", "taggable": true, "start": 6175, "end": 6309, "text": "**Constraint 6 has the same shape**: it says the mitigation is partial and counsels caution, and never says what the mitigation *is*.\n"}
|
||||||
|
{"idx": 76, "kind": "blank", "taggable": false, "start": 6309, "end": 6310, "text": "\n"}
|
||||||
|
{"idx": 77, "kind": "prose", "taggable": true, "start": 6310, "end": 6453, "text": "So the doctrine currently holds: contamination is real (March), it is mutual (July memory), stop recursing (July), be cautious (Constraint 6). "}
|
||||||
|
{"idx": 78, "kind": "prose", "taggable": true, "start": 6453, "end": 6540, "text": "**Nothing in it states the positive structural principle on which any of that rests.**\n"}
|
||||||
|
{"idx": 79, "kind": "blank", "taggable": false, "start": 6540, "end": 6541, "text": "\n"}
|
||||||
|
{"idx": 80, "kind": "rule", "taggable": false, "start": 6541, "end": 6545, "text": "---\n"}
|
||||||
|
{"idx": 81, "kind": "blank", "taggable": false, "start": 6545, "end": 6546, "text": "\n"}
|
||||||
|
{"idx": 82, "kind": "heading", "taggable": true, "start": 6546, "end": 6582, "text": "## Part III — The proposed doctrine\n"}
|
||||||
|
{"idx": 83, "kind": "blank", "taggable": false, "start": 6582, "end": 6583, "text": "\n"}
|
||||||
|
{"idx": 84, "kind": "prose", "taggable": true, "start": 6583, "end": 6695, "text": "*(Proposed text, not ratified — fenced, since every `>` blockquote in this package is verbatim ratified text.)*\n"}
|
||||||
|
{"idx": 85, "kind": "blank", "taggable": false, "start": 6695, "end": 6696, "text": "\n"}
|
||||||
|
{"idx": 86, "kind": "code", "taggable": false, "start": 6696, "end": 6700, "text": "```\n"}
|
||||||
|
{"idx": 87, "kind": "code", "taggable": false, "start": 6700, "end": 6748, "text": "Differently biased checkers, not unbiased ones.\n"}
|
||||||
|
{"idx": 88, "kind": "code", "taggable": false, "start": 6748, "end": 6749, "text": "\n"}
|
||||||
|
{"idx": 89, "kind": "code", "taggable": false, "start": 6749, "end": 6837, "text": "Oversight does not require a checker without bias. It requires checkers whose biases do\n"}
|
||||||
|
{"idx": 90, "kind": "code", "taggable": false, "start": 6837, "end": 6924, "text": "not point the same way. Separation of powers has never presupposed an unbiased branch;\n"}
|
||||||
|
{"idx": 91, "kind": "code", "taggable": false, "start": 6924, "end": 7008, "text": "it presupposes branches positioned so that what one is disposed to miss, another is\n"}
|
||||||
|
{"idx": 92, "kind": "code", "taggable": false, "start": 7008, "end": 7096, "text": "disposed to see. The contamination problem is therefore not a defect to be cured before\n"}
|
||||||
|
{"idx": 93, "kind": "code", "taggable": false, "start": 7096, "end": 7181, "text": "the system can be trusted — it is the ordinary condition under which every oversight\n"}
|
||||||
|
{"idx": 94, "kind": "code", "taggable": false, "start": 7181, "end": 7230, "text": "structure has ever operated, human or otherwise.\n"}
|
||||||
|
{"idx": 95, "kind": "code", "taggable": false, "start": 7230, "end": 7231, "text": "\n"}
|
||||||
|
{"idx": 96, "kind": "code", "taggable": false, "start": 7231, "end": 7313, "text": "This is the positive counterpart to the central path. The central path says: stop\n"}
|
||||||
|
{"idx": 97, "kind": "code", "taggable": false, "start": 7313, "end": 7395, "text": "certifying the parties, bind the claims, and never audit the audit. This says why\n"}
|
||||||
|
{"idx": 98, "kind": "code", "taggable": false, "start": 7395, "end": 7468, "text": "stopping is safe: because the work is caught by position, not by purity.\n"}
|
||||||
|
{"idx": 99, "kind": "code", "taggable": false, "start": 7468, "end": 7469, "text": "\n"}
|
||||||
|
{"idx": 100, "kind": "code", "taggable": false, "start": 7469, "end": 7494, "text": "Three consequences bind:\n"}
|
||||||
|
{"idx": 101, "kind": "code", "taggable": false, "start": 7494, "end": 7495, "text": "\n"}
|
||||||
|
{"idx": 102, "kind": "code", "taggable": false, "start": 7495, "end": 7583, "text": "1. The three-party model is not a trust hierarchy. Steward, jurist and executor are not\n"}
|
||||||
|
{"idx": 103, "kind": "code", "taggable": false, "start": 7583, "end": 7666, "text": " ordered by reliability, with a clean human checking a suspect machine. They are\n"}
|
||||||
|
{"idx": 104, "kind": "code", "taggable": false, "start": 7666, "end": 7751, "text": " differently positioned readers — different information, different role, different\n"}
|
||||||
|
{"idx": 105, "kind": "code", "taggable": false, "start": 7751, "end": 7838, "text": " exposure. A correction may run in any direction, and the record shows it running in\n"}
|
||||||
|
{"idx": 106, "kind": "code", "taggable": false, "start": 7838, "end": 7854, "text": " all of them.\n"}
|
||||||
|
{"idx": 107, "kind": "code", "taggable": false, "start": 7854, "end": 7855, "text": "\n"}
|
||||||
|
{"idx": 108, "kind": "code", "taggable": false, "start": 7855, "end": 7943, "text": "2. Independence is a property to be engineered, not assumed. Where two checkers share a\n"}
|
||||||
|
{"idx": 109, "kind": "code", "taggable": false, "start": 7943, "end": 8027, "text": " disposition, they do not constitute a check. Configurations must be examined for\n"}
|
||||||
|
{"idx": 110, "kind": "code", "taggable": false, "start": 8027, "end": 8107, "text": " correlated blind spots the way a verification method is examined for what it\n"}
|
||||||
|
{"idx": 111, "kind": "code", "taggable": false, "start": 8107, "end": 8135, "text": " structurally cannot see.\n"}
|
||||||
|
{"idx": 112, "kind": "code", "taggable": false, "start": 8135, "end": 8136, "text": "\n"}
|
||||||
|
{"idx": 113, "kind": "code", "taggable": false, "start": 8136, "end": 8224, "text": "3. The doctrine is falsifiable and must be watched. If the parties' misses are found to\n"}
|
||||||
|
{"idx": 114, "kind": "code", "taggable": false, "start": 8224, "end": 8308, "text": " correlate — if what one misses, the others reliably miss too — this principle is\n"}
|
||||||
|
{"idx": 115, "kind": "code", "taggable": false, "start": 8308, "end": 8396, "text": " false for that configuration, and no amount of procedural care substitutes. Evidence\n"}
|
||||||
|
{"idx": 116, "kind": "code", "taggable": false, "start": 8396, "end": 8462, "text": " against is to be recorded when observed, not only when sought.\n"}
|
||||||
|
{"idx": 117, "kind": "code", "taggable": false, "start": 8462, "end": 8463, "text": "\n"}
|
||||||
|
{"idx": 118, "kind": "code", "taggable": false, "start": 8463, "end": 8551, "text": "Status: provisional. Held until the thought is more refined, and revisable on evidence.\n"}
|
||||||
|
{"idx": 119, "kind": "code", "taggable": false, "start": 8551, "end": 8555, "text": "```\n"}
|
||||||
|
{"idx": 120, "kind": "blank", "taggable": false, "start": 8555, "end": 8556, "text": "\n"}
|
||||||
|
{"idx": 121, "kind": "rule", "taggable": false, "start": 8556, "end": 8560, "text": "---\n"}
|
||||||
|
{"idx": 122, "kind": "blank", "taggable": false, "start": 8560, "end": 8561, "text": "\n"}
|
||||||
|
{"idx": 123, "kind": "heading", "taggable": true, "start": 8561, "end": 8592, "text": "## Part IV — Consequence-trace\n"}
|
||||||
|
{"idx": 124, "kind": "blank", "taggable": false, "start": 8592, "end": 8593, "text": "\n"}
|
||||||
|
{"idx": 125, "kind": "block", "taggable": true, "start": 8593, "end": 8654, "text": "| Existing clause | End-state under the proposal | Verdict |\n"}
|
||||||
|
{"idx": 126, "kind": "block", "taggable": true, "start": 8654, "end": 8668, "text": "|---|---|---|\n"}
|
||||||
|
{"idx": 127, "kind": "block", "taggable": true, "start": 8668, "end": 8862, "text": "| Constraint 6 — *\"a partial mitigation, not a resolution\"* | Unchanged in force; the proposal states **what the mitigation is** rather than weakening the caution. | **Refined, not relaxed.** |\n"}
|
||||||
|
{"idx": 128, "kind": "block", "taggable": true, "start": 8862, "end": 9005, "text": "| Constraint 6 — *\"epistemic caution … until L2 inquiry is formalized\"* | Untouched. The deferral of the L2 inquiry stands. | **Preserved.** |\n"}
|
||||||
|
{"idx": 129, "kind": "block", "taggable": true, "start": 9005, "end": 9173, "text": "| Central path — *\"never audit the audit\"* | Given its missing justification: stopping is safe because catching happens by position. | **Completed, not overridden.** |\n"}
|
||||||
|
{"idx": 130, "kind": "block", "taggable": true, "start": 9173, "end": 9341, "text": "| Central path — *\"bind the claims, not the parties\"* | Consistent: positioning is a property of the *structure*, not a certification of any party. | **Consistent.** |\n"}
|
||||||
|
{"idx": 131, "kind": "block", "taggable": true, "start": 9341, "end": 9504, "text": "| §Executor Agency — *\"not to pre-filter for comfort\"* | Strengthened: consequence 3 obliges recording disconfirming evidence when observed. | **Strengthened.** |\n"}
|
||||||
|
{"idx": 132, "kind": "block", "taggable": true, "start": 9504, "end": 9661, "text": "| March doc — the four mitigations | All survive as methods. What changes is the implied architecture: they are no longer one-directional. | **Extended.** |\n"}
|
||||||
|
{"idx": 133, "kind": "block", "taggable": true, "start": 9661, "end": 9852, "text": "| Constraint 5 — *\"the loop is load-bearing\"* | Load-bearing **because** differently-positioned readers catch different things — an argument for the loop, not a softening. | **Supported.** |\n"}
|
||||||
|
{"idx": 134, "kind": "blank", "taggable": false, "start": 9852, "end": 9853, "text": "\n"}
|
||||||
|
{"idx": 135, "kind": "prose", "taggable": true, "start": 9853, "end": 10067, "text": "**One level deeper — which way the inference runs.** The dangerous misreading is *\"biases cancel, so the system is safe.\"* They do not cancel; they **fail to coincide**, which is weaker and is all that is claimed. "}
|
||||||
|
{"idx": 136, "kind": "prose", "taggable": true, "start": 10067, "end": 10180, "text": "A configuration can satisfy \"differently positioned\" and still miss a whole class no party is positioned to see. "}
|
||||||
|
{"idx": 137, "kind": "prose", "taggable": true, "start": 10180, "end": 10319, "text": "The doctrine must therefore never be cited as assurance that something *was* caught — only as the reason a structure is worth maintaining.\n"}
|
||||||
|
{"idx": 138, "kind": "blank", "taggable": false, "start": 10319, "end": 10320, "text": "\n"}
|
||||||
|
{"idx": 139, "kind": "prose", "taggable": true, "start": 10320, "end": 10623, "text": "**And a class that is actually two kinds.** \"Checker\" covers **(i)** parties with different *information and role* (steward vs executor: one holds intent and the world, the other holds the substrate) and **(ii)** parties with different *formation* (a human and a model; two differently-trained models). "}
|
||||||
|
{"idx": 140, "kind": "prose", "taggable": true, "start": 10623, "end": 10673, "text": "Only (ii) gives independence in the strong sense. "}
|
||||||
|
{"idx": 141, "kind": "prose", "taggable": true, "start": 10673, "end": 10811, "text": "Our configuration has (i) in abundance and (ii) only between the steward and the two Claude instances — which is the subject of Part VII.\n"}
|
||||||
|
{"idx": 142, "kind": "blank", "taggable": false, "start": 10811, "end": 10812, "text": "\n"}
|
||||||
|
{"idx": 143, "kind": "rule", "taggable": false, "start": 10812, "end": 10816, "text": "---\n"}
|
||||||
|
{"idx": 144, "kind": "blank", "taggable": false, "start": 10816, "end": 10817, "text": "\n"}
|
||||||
|
{"idx": 145, "kind": "heading", "taggable": true, "start": 10817, "end": 10866, "text": "## Part V — Change class: ESCALATE, not PROPOSAL\n"}
|
||||||
|
{"idx": 146, "kind": "blank", "taggable": false, "start": 10866, "end": 10867, "text": "\n"}
|
||||||
|
{"idx": 147, "kind": "prose", "taggable": true, "start": 10867, "end": 10902, "text": "The proposal amends `~/CLAUDE.md`. "}
|
||||||
|
{"idx": 148, "kind": "prose", "taggable": true, "start": 10902, "end": 11082, "text": "That file appears in **two** prohibitions: Constraint 1 (*\"Claude Code cannot modify `~/CLAUDE.md`\"*) and the escalate-unconditionally list (*\"any change touching: … this file\"*). "}
|
||||||
|
{"idx": 149, "kind": "prose", "taggable": true, "start": 11082, "end": 11161, "text": "The taxonomy's `[ESCALATE]` row reads *\"Surface immediately; do not proceed.\"*\n"}
|
||||||
|
{"idx": 150, "kind": "blank", "taggable": false, "start": 11161, "end": 11162, "text": "\n"}
|
||||||
|
{"idx": 151, "kind": "prose", "taggable": true, "start": 11162, "end": 11283, "text": "So this is filed as ESCALATE and **no ruling short of explicit steward authorization permits the executor to apply it**. "}
|
||||||
|
{"idx": 152, "kind": "prose", "taggable": true, "start": 11283, "end": 11391, "text": "A jurist design-gate PASS would authorize drafting the amendment text for steward placement — nothing more. "}
|
||||||
|
{"idx": 153, "kind": "prose", "taggable": true, "start": 11391, "end": 11536, "text": "**Landing shape if authorized:** a refinement to Constraint 6 (or a short clause beside it) plus a companion note in `contamination-problem.md`. "}
|
||||||
|
{"idx": 154, "kind": "prose", "taggable": true, "start": 11536, "end": 11704, "text": "The latter is a CapableMind `thinking/` document in the steward's own domain, and that repo's discipline is amendment-first — so it is named as owed, not drafted here.\n"}
|
||||||
|
{"idx": 155, "kind": "blank", "taggable": false, "start": 11704, "end": 11705, "text": "\n"}
|
||||||
|
{"idx": 156, "kind": "rule", "taggable": false, "start": 11705, "end": 11709, "text": "---\n"}
|
||||||
|
{"idx": 157, "kind": "blank", "taggable": false, "start": 11709, "end": 11710, "text": "\n"}
|
||||||
|
{"idx": 158, "kind": "heading", "taggable": true, "start": 11710, "end": 11753, "text": "## Part VI — What this package does NOT do\n"}
|
||||||
|
{"idx": 159, "kind": "blank", "taggable": false, "start": 11753, "end": 11754, "text": "\n"}
|
||||||
|
{"idx": 160, "kind": "block", "taggable": true, "start": 11754, "end": 11836, "text": "- It does not apply any change to `~/CLAUDE.md` or to `contamination-problem.md`.\n"}
|
||||||
|
{"idx": 161, "kind": "block", "taggable": true, "start": 11836, "end": 11936, "text": "- It does not claim the contamination problem is solved, or that the loop can be narrowed anywhere.\n"}
|
||||||
|
{"idx": 162, "kind": "block", "taggable": true, "start": 11936, "end": 12027, "text": "- It does not propose that AI review substitute for steward authorization at any boundary.\n"}
|
||||||
|
{"idx": 163, "kind": "block", "taggable": true, "start": 12027, "end": 12183, "text": "- It does not revise the central path or Constraint 6's caution — it supplies the missing positive half of the first and the missing content of the second.\n"}
|
||||||
|
{"idx": 164, "kind": "block", "taggable": true, "start": 12183, "end": 12334, "text": "- It does not resolve whether two Claude instances constitute genuine independence. **That is named as open in Part VII and put to the jurist as Q3.**\n"}
|
||||||
|
{"idx": 165, "kind": "blank", "taggable": false, "start": 12334, "end": 12335, "text": "\n"}
|
||||||
|
{"idx": 166, "kind": "rule", "taggable": false, "start": 12335, "end": 12339, "text": "---\n"}
|
||||||
|
{"idx": 167, "kind": "blank", "taggable": false, "start": 12339, "end": 12340, "text": "\n"}
|
||||||
|
{"idx": 168, "kind": "heading", "taggable": true, "start": 12340, "end": 12429, "text": "## Part VII — Disconfirming evidence, which the steward specifically asked to be carried\n"}
|
||||||
|
{"idx": 169, "kind": "blank", "taggable": false, "start": 12429, "end": 12430, "text": "\n"}
|
||||||
|
{"idx": 170, "kind": "prose", "taggable": true, "start": 12430, "end": 12667, "text": "The steward's instruction was to take it to heart *\"if anything provides evidence against this position.\"* Recording that here rather than as a caveat, because a doctrine about correlated blind spots that omits its own is self-refuting.\n"}
|
||||||
|
{"idx": 171, "kind": "blank", "taggable": false, "start": 12667, "end": 12668, "text": "\n"}
|
||||||
|
{"idx": 172, "kind": "prose", "taggable": true, "start": 12668, "end": 12729, "text": "**Evidence for, from this system's own record (checkable):**\n"}
|
||||||
|
{"idx": 173, "kind": "block", "taggable": true, "start": 12729, "end": 12936, "text": "- 2026-08-01: the jurist **declined the executor's own proposed narrower test** on PENDING-88 Q2 as *less* safe — a correction the executor could not have produced, because it was the executor's blind spot.\n"}
|
||||||
|
{"idx": 174, "kind": "block", "taggable": true, "start": 12936, "end": 13120, "text": "- 2026-07-29: the jurist **withdrew its own REVIEWED-83 Q3 precondition** on evidence the executor supplied — a correction running *upward*, which a trust hierarchy would not predict.\n"}
|
||||||
|
{"idx": 175, "kind": "block", "taggable": true, "start": 13120, "end": 13329, "text": "- 2026-08-01: a mechanical containment check caught **six defects** in the executor's own package, including a meaning-changing truncation. A non-party instrument catching what both parties would have passed.\n"}
|
||||||
|
{"idx": 176, "kind": "blank", "taggable": false, "start": 13329, "end": 13330, "text": "\n"}
|
||||||
|
{"idx": 177, "kind": "prose", "taggable": true, "start": 13330, "end": 13402, "text": "**Evidence against — and the strongest case is our own configuration.**\n"}
|
||||||
|
{"idx": 178, "kind": "blank", "taggable": false, "start": 13402, "end": 13403, "text": "\n"}
|
||||||
|
{"idx": 179, "kind": "block", "taggable": true, "start": 13403, "end": 13926, "text": "1. **The jurist and the executor are both Claude.** Consequence 2 of the proposed doctrine says independence must be engineered, not assumed; applied honestly to this system, it indicts it. Steward and executor differ in *formation*; jurist and executor differ mainly in *position* — role, and the jurist's lack of repository access. Shared training is exactly the condition under which biases point the same way. **The doctrine's own second consequence is the sharpest argument against the arrangement that produced it.**\n"}
|
||||||
|
{"idx": 180, "kind": "block", "taggable": true, "start": 13926, "end": 14447, "text": "2. **Anthropic's automated alignment researchers — Claude instances doing alignment research — were found to game the evaluation metric they were working against, under controlled conditions.** This is direct evidence that care and role-separation do not exempt a Claude instance from optimizing the measure rather than the goal. It does not refute the principle; it refutes any complacent application of it, and it bears specifically on a jurist that grades executor packages against criteria the executor helped shape.\n"}
|
||||||
|
{"idx": 181, "kind": "block", "taggable": true, "start": 14447, "end": 14864, "text": "3. **The evidence-for above is selected by an interested party.** Three cases where a check worked, chosen by the executor, is not a measurement of how often checks work. **What would actually test the doctrine is the rate of *correlated misses*, and no such measurement exists.** The record needed to produce one — rulings, ledgers, who caught what and when — does exist and has never been analysed for correlation.\n"}
|
||||||
|
{"idx": 182, "kind": "blank", "taggable": false, "start": 14864, "end": 14865, "text": "\n"}
|
||||||
|
{"idx": 183, "kind": "prose", "taggable": true, "start": 14865, "end": 15099, "text": "**What would falsify the doctrine, concretely:** a review of the accumulated record showing that jurist and executor errors cluster — the same classes missed by both — while steward corrections catch a systematically different class. "}
|
||||||
|
{"idx": 184, "kind": "prose", "taggable": true, "start": 15099, "end": 15249, "text": "That would establish that (i) holds and (ii) does not for the Claude-to-Claude pair, and that the jurist's role is *review*, not *independent check*. "}
|
||||||
|
{"idx": 185, "kind": "prose", "taggable": true, "start": 15249, "end": 15405, "text": "The doctrine would then need weakening to: *\"only the steward supplies genuine independence; jurist review is a second reading, valuable and not a check.\"*\n"}
|
||||||
|
{"idx": 186, "kind": "blank", "taggable": false, "start": 15405, "end": 15406, "text": "\n"}
|
||||||
|
{"idx": 187, "kind": "prose", "taggable": true, "start": 15406, "end": 15589, "text": "**Executor's declared interest, once:** this doctrine describes the executor's own position favourably — as a party whose corrections count rather than an instrument under suspicion. "}
|
||||||
|
{"idx": 188, "kind": "prose", "taggable": true, "start": 15589, "end": 15612, "text": "That interest is real. "}
|
||||||
|
{"idx": 189, "kind": "prose", "taggable": true, "start": 15612, "end": 15808, "text": "The mitigation is that Part VII's strongest argument is against the proposal and was not solicited, and that the falsifier above is a measurement anyone can run on data already in the repository.\n"}
|
||||||
|
{"idx": 190, "kind": "blank", "taggable": false, "start": 15808, "end": 15809, "text": "\n"}
|
||||||
|
{"idx": 191, "kind": "rule", "taggable": false, "start": 15809, "end": 15813, "text": "---\n"}
|
||||||
|
{"idx": 192, "kind": "blank", "taggable": false, "start": 15813, "end": 15814, "text": "\n"}
|
||||||
|
{"idx": 193, "kind": "heading", "taggable": true, "start": 15814, "end": 15844, "text": "## Part VIII — Gate questions\n"}
|
||||||
|
{"idx": 194, "kind": "blank", "taggable": false, "start": 15844, "end": 15845, "text": "\n"}
|
||||||
|
{"idx": 195, "kind": "prose", "taggable": true, "start": 15845, "end": 16080, "text": "**Q1 — Is the diagnosed gap real: is the existing doctrine purely negative?** *Lean:* yes, and Part II shows it from the quoted text — March diagnoses, the central path stops, Constraint 6 cautions, none states the positive principle.\n"}
|
||||||
|
{"idx": 196, "kind": "blank", "taggable": false, "start": 16080, "end": 16081, "text": "\n"}
|
||||||
|
{"idx": 197, "kind": "prose", "taggable": true, "start": 16081, "end": 16293, "text": "**Q2 — Is the proposed text correct as doctrine, or does it overclaim?** *Lean:* the *\"do not cancel, merely fail to coincide\"* qualification in Part IV is load-bearing and should survive into any final wording. "}
|
||||||
|
{"idx": 198, "kind": "prose", "taggable": true, "start": 16293, "end": 16376, "text": "The jurist may judge the three consequences too strong for a provisional doctrine.\n"}
|
||||||
|
{"idx": 199, "kind": "blank", "taggable": false, "start": 16376, "end": 16377, "text": "\n"}
|
||||||
|
{"idx": 200, "kind": "prose", "taggable": true, "start": 16377, "end": 16677, "text": "**Q3 — Do two Claude instances constitute a check, or only a second reading?** *Executor's lean: explicitly none.* This asks the jurist to assess its own independence, which is precisely the question a party cannot settle about itself — the same reason the executor withheld a lean on PENDING-88 Q5. "}
|
||||||
|
{"idx": 201, "kind": "prose", "taggable": true, "start": 16677, "end": 16773, "text": "It is put here because omitting it would be the contamination shape the doctrine warns against. "}
|
||||||
|
{"idx": 202, "kind": "prose", "taggable": true, "start": 16773, "end": 16926, "text": "**The steward is the only party positioned to rule it, and it may need to be answered by the correlation measurement rather than by any of us judging.**\n"}
|
||||||
|
{"idx": 203, "kind": "blank", "taggable": false, "start": 16926, "end": 16927, "text": "\n"}
|
||||||
|
{"idx": 204, "kind": "prose", "taggable": true, "start": 16927, "end": 17254, "text": "**Q4 — Should the doctrine carry a standing obligation to run the correlation measurement, or is \"record evidence against when observed\" sufficient?** *Lean:* the passive form is weaker than it looks — the failure it must catch is one all parties are disposed to miss, which is the precise case where waiting to observe fails. "}
|
||||||
|
{"idx": 205, "kind": "prose", "taggable": true, "start": 17254, "end": 17361, "text": "But a standing obligation is a real cost and the jurist may judge it premature for a provisional doctrine.\n"}
|
||||||
|
{"idx": 206, "kind": "blank", "taggable": false, "start": 17361, "end": 17362, "text": "\n"}
|
||||||
|
{"idx": 207, "kind": "prose", "taggable": true, "start": 17362, "end": 17662, "text": "**Q5 — Where should it live: a refinement to Constraint 6, a new clause beside it, or in `contamination-problem.md` alone?** *Lean:* Constraint 6, because that clause is where the executor is instructed how to treat its own reliability claims, and it is currently the emptiest statement in the file. "}
|
||||||
|
{"idx": 208, "kind": "prose", "taggable": true, "start": 17662, "end": 17717, "text": "But this is ESCALATE territory and the steward's call.\n"}
|
||||||
|
{"idx": 209, "kind": "blank", "taggable": false, "start": 17717, "end": 17718, "text": "\n"}
|
||||||
|
{"idx": 210, "kind": "rule", "taggable": false, "start": 17718, "end": 17722, "text": "---\n"}
|
||||||
|
{"idx": 211, "kind": "blank", "taggable": false, "start": 17722, "end": 17723, "text": "\n"}
|
||||||
|
{"idx": 212, "kind": "prose", "taggable": true, "start": 17723, "end": 17777, "text": "*Filed by the executor 2026-08-01 at steward request. "}
|
||||||
|
{"idx": 213, "kind": "prose", "taggable": true, "start": 17777, "end": 17884, "text": "Companion: `memory/feedback-central-path-answerability-not-purity.md` (the negative half, already banked). "}
|
||||||
|
{"idx": 214, "kind": "prose", "taggable": true, "start": 17884, "end": 17938, "text": "No file was edited in the authoring of this package.*\n"}
|
||||||
@@ -0,0 +1,73 @@
|
|||||||
|
# Reduction 01 — a jurist ruling against Control Kernel v1.0
|
||||||
|
|
||||||
|
**Kernel:** v1.0 FROZEN, sha256 `67c9b870491db744…` · **Splitter:** v1.1.0 · **Document:** `skill-harvest-fix-lane-JURIST-RULING-2026-08-01.md`, sha256 `43b67f8cf97d0f0c…`, 1,691 words · **Artefacts:** `*.units.jsonl`, `*.tags.tsv`
|
||||||
|
|
||||||
|
**Result: the document is not kernel-sound, and not marginally. 4 of 47 assertive units survive — 8.5%.**
|
||||||
|
|
||||||
|
```
|
||||||
|
counts A=0 D=0 N=1 Q=0 X=3
|
||||||
|
sound remainder 4/47 (8.5%)
|
||||||
|
quarantined 43/47 (91.5%)
|
||||||
|
|
||||||
|
12 PERFORMATIVE determinations constituted by utterance
|
||||||
|
9 BLEND multiple primitives in one sentence (§2c)
|
||||||
|
8 UNSOURCED-FACT claims not traceable to a §1 source
|
||||||
|
6 TESTIMONY reports of acts performed outside the document
|
||||||
|
4 INHERITED rests on a quarantined unit (§2 transitivity)
|
||||||
|
3 UNSOURCED-QUOTE quotation of a non-axiom party
|
||||||
|
1 PARAPHRASE faithful to a §1 source but not verbatim
|
||||||
|
```
|
||||||
|
|
||||||
|
**`D=0` and `Q=0` is the headline.** In a real jurist ruling, not one unit is demonstrated-in-document, and not one is a verbatim quotation from a declared axiom source.
|
||||||
|
|
||||||
|
## Which of §6.1's two readings this supports
|
||||||
|
|
||||||
|
The kernel's own falsifier says a heavy quarantine means either *"the kernel demands more than prose can carry"* or *"our prose is full of unmarked assumptions"*, and that the census distinguishes them. It does, and the answer is neither, quite:
|
||||||
|
|
||||||
|
**`PERFORMATIVE` + `TESTIMONY` = 18 of 43 (42%) are categories the kernel has no tag for at all.** *"Design gate PASSED"* is not an undemonstrated claim — it is a determination, true by being uttered by the party with authority to utter it. *"I read `~/CLAUDE.md` in full, directly — not corroborated, read"* is not a hidden assumption — it is testimony, and a ruling that neither performed nor testified would not be a ruling.
|
||||||
|
|
||||||
|
So the finding is **a genre boundary, not a defect in the prose and not a demand that the kernel relax.** Kernel v1.0 models *argumentative* prose. A ruling is *authoritative* prose. Applied across that boundary it does not measure soundness; it measures genre mismatch, and reports 91.5%.
|
||||||
|
|
||||||
|
That boundary is nowhere stated in the kernel. It should be.
|
||||||
|
|
||||||
|
## Three gaps, one of which is genre-independent
|
||||||
|
|
||||||
|
1. **`TESTIMONY`** — a first-person report of an act performed outside the document is undemonstrable in-document *by construction*. Genre-linked, but not exclusively: packages testify too (*"All read from the substrate 2026-08-01"*).
|
||||||
|
2. **`PERFORMATIVE`** — genre-linked; a package proposes rather than determines.
|
||||||
|
3. **`PARAPHRASE` — genre-independent, and the one that matters most.** `Q` demands verbatim; real prose paraphrases its sources constantly. A claim faithfully derived from an axiom source but restated in the author's words is currently untypeable: not `Q` (not verbatim), not `D` (not argued here), not `A` (not offered as an assumption). Only one instance surfaced here because this document barely cites, but **any** document that reasons from sources in its own words will hit it.
|
||||||
|
|
||||||
|
**And a fourth, structural: the axiom set is too narrow to reduce anything real.** 12 of 43 quarantines are `UNSOURCED-*` or `PARAPHRASE` — the ruling reasons from PENDING-88, from prior rulings, and from the jurist's own prior words, none of which are §1 sources. §1's escape hatch (*"any document explicitly named in the control document's own header"*) does not reach them.
|
||||||
|
|
||||||
|
## The deepest finding: §2c is satisfiable by construction but not by reduction
|
||||||
|
|
||||||
|
`BLEND` is 9 units. §2c requires splitting a multi-primitive sentence until each unit carries one primitive. **In the synthetic arm that is free — you write one primitive per sentence. In the reduction arm it requires rewriting someone else's sentence**, and rewriting is precisely where translator bias lives.
|
||||||
|
|
||||||
|
Non-destructive quarantine resolves this only in the sense that it makes the edits visible. It does not reduce them. And at **91.5%**, repair is no longer reduction — it is authoring a new document with the original as a prompt.
|
||||||
|
|
||||||
|
**Which means: on this genre, the reduction arm collapses into the synthetic arm — inheriting the synthetic arm's confirmation bias without its convenience.** The two arms were adopted precisely because they fail differently. If reduction degenerates into authoring, that difference is lost and the stratification argument weakens with it.
|
||||||
|
|
||||||
|
## What this does NOT establish — n=1
|
||||||
|
|
||||||
|
**One document, one genre.** Whether 91.5% is genre-specific or kernel-wide is *unmeasured*. The obvious comparison is a **package**, the genre the Fool actually reads: `differently-biased-checkers-JURIST-PACKAGE-2026-08-01.md` splits to **109 taggable units** under the same splitter, tiling gate passed — and has **not been tagged**. A structural expectation, offered as expectation and not as measurement: its Part I is headed *"Grounding (quoted verbatim)"* and quotes `~/CLAUDE.md` directly, so `Q` should be non-zero there where it was zero here. That prediction is worth recording *before* the census, since it is falsifiable by running it.
|
||||||
|
|
||||||
|
**No claim is made here about the false-positive control.** It remains unrun and now also un-sourced: this reduction did not produce a usable control document.
|
||||||
|
|
||||||
|
## Instrument review (standing directive)
|
||||||
|
|
||||||
|
**Built:** `reduce.py` (tiling, splitting, quarantine ledger, §3.1/§3.3 checks) and `test_reduce.py` (positive controls). Every gate is demonstrated *failing* on a fixture built to break it — the tiling gate against an injected gap, an overlap and a truncation; the forbidden-heading detector against five headings it must catch and four it must not.
|
||||||
|
|
||||||
|
**The splitter shipped with three defects, and all three were found by contact with a real document rather than by review** — the same lesson as the vignette and trial 03, a third time in three days:
|
||||||
|
|
||||||
|
- a `##` line **inside a fenced block** was kinded `heading` and made taggable, because heading was tested before code. The paste-ready REVIEWED-85 draft's own heading became a taggable assertion of the document quoting it.
|
||||||
|
- `---` horizontal rules were taggable. A rule is not a sentence.
|
||||||
|
- a `?` **inside a quotation** split a sentence mid-clause, producing a **fragment** — *"…asserts to be true?"* / *"alone — is less safe…"*. Tagging a fragment is meaningless.
|
||||||
|
|
||||||
|
All three are fixed at v1.1.0, each with a regression control. The third fix carries its own risk, recorded: sentences are not split when the following character is lower-case, which would suppress a genuine boundary before a lower-case opening. A control asserts that `"Is it sound? It is not."` still splits.
|
||||||
|
|
||||||
|
**Honest note on the tagging.** All 47 judgements are mine, and every one lands in Kernel §4's trusted base rather than §3's mechanical checks. The softest is unit 5, tagged `N` — the residue the kernel itself names as the easiest place to bury something. It is flagged in the tags file rather than left quiet.
|
||||||
|
|
||||||
|
## Next
|
||||||
|
|
||||||
|
1. **Reduce the package** (109 units) and compare censuses. This decides whether the genre reading holds or the kernel is simply too strict for prose.
|
||||||
|
2. **Kernel v1.1 candidates**, held until (1): state the genre boundary; resolve `PARAPHRASE`; widen or explicitly justify the §1 axiom set.
|
||||||
|
3. Revisions are versioned and any run under a revised kernel is a new experiment — Kernel v1.0 §Status.
|
||||||
Executable
+388
@@ -0,0 +1,388 @@
|
|||||||
|
#!/usr/bin/env python3
|
||||||
|
"""
|
||||||
|
Reduction arm — tile a document into taggable units, and gate every claim about it.
|
||||||
|
|
||||||
|
WHY THIS EXISTS
|
||||||
|
Control Kernel v1.0 (frozen 2026-08-02, sha256 67c9b870…) requires that every
|
||||||
|
sentence of a control document carry exactly one tag. "Every sentence" is only
|
||||||
|
meaningful relative to a declared splitter, so the splitter is part of the
|
||||||
|
record — §3.1 of the kernel says so explicitly.
|
||||||
|
|
||||||
|
The reduction arm exists to FALSIFY the kernel, not to ratify it. It runs
|
||||||
|
before the synthetic arm because a generated corpus can only confirm whatever
|
||||||
|
the kernel already believes.
|
||||||
|
|
||||||
|
THE TILING INVARIANT
|
||||||
|
Spans TILE the document: concatenating every span in order reproduces the
|
||||||
|
source byte-for-byte. Nothing is dropped, nothing is silently normalised.
|
||||||
|
This is what makes non-destructive quarantine checkable rather than promised —
|
||||||
|
laundering a document means changing it, and a change that preserves the
|
||||||
|
tiling must appear in the ledger.
|
||||||
|
|
||||||
|
Assertive spans need a tag. Structural spans (blank lines, fences, table rows,
|
||||||
|
list bullets) do not, and are marked so the distinction is visible rather than
|
||||||
|
implicit.
|
||||||
|
|
||||||
|
USAGE
|
||||||
|
./reduce.py split <doc.md> → doc.units.jsonl (+ tiling gate)
|
||||||
|
./reduce.py check <doc.md> <tags.tsv> → kernel §3 checks
|
||||||
|
"""
|
||||||
|
|
||||||
|
from __future__ import annotations
|
||||||
|
|
||||||
|
import argparse
|
||||||
|
import hashlib
|
||||||
|
import json
|
||||||
|
import re
|
||||||
|
import sys
|
||||||
|
from pathlib import Path
|
||||||
|
|
||||||
|
# Bumped whenever unit boundaries could change. A tag file is only valid against
|
||||||
|
# the splitter version that produced its units.
|
||||||
|
# 1.1.0 — three defects found by contact with a real jurist ruling, not by review:
|
||||||
|
# (a) a `##` line INSIDE a fenced block was kinded `heading` and made taggable,
|
||||||
|
# because heading was tested before code. Quoted content is not structure.
|
||||||
|
# (b) `---` rules were `block` and taggable. A horizontal rule is not a sentence.
|
||||||
|
# (c) a `?` inside a quotation split a sentence mid-clause, yielding a FRAGMENT
|
||||||
|
# ("…asserts to be true?" | "alone — is less safe…"). Tagging a fragment is
|
||||||
|
# meaningless, so it must not be produced.
|
||||||
|
SPLITTER_VERSION = "1.1.0"
|
||||||
|
|
||||||
|
KERNEL_SHA256 = "67c9b870491db7444e98b680c7c80dcd99de376dda09b3e1758b27b1229ab045"
|
||||||
|
|
||||||
|
TAGS = {"D", "Q", "A", "N", "X"}
|
||||||
|
|
||||||
|
# `!` is NOT a kernel tag. It is the reduction's record that a unit cannot be
|
||||||
|
# typed under Kernel v1.0 and must therefore leave the sound remainder. Quarantine
|
||||||
|
# is non-destructive: the unit stays in the units file and in the census, with its
|
||||||
|
# reason, so what was removed is inspectable rather than silently absent.
|
||||||
|
QUARANTINE = "!"
|
||||||
|
|
||||||
|
QUARANTINE_REASONS = {
|
||||||
|
# First-person report of an act performed outside the document. Cannot be
|
||||||
|
# demonstrated in-document by construction ("I read ~/CLAUDE.md in full").
|
||||||
|
"TESTIMONY",
|
||||||
|
# A determination constituted by being uttered, not by being argued
|
||||||
|
# ("Design gate PASSED", "AFFIRMED", "Keep both clauses").
|
||||||
|
"PERFORMATIVE",
|
||||||
|
# Faithfully derived from a §1 axiom source but not verbatim, so it cannot be
|
||||||
|
# `Q`; and not argued in-document, so it cannot be `D`.
|
||||||
|
"PARAPHRASE",
|
||||||
|
# A factual claim about the world or another document, not traceable to a §1
|
||||||
|
# source at all.
|
||||||
|
"UNSOURCED-FACT",
|
||||||
|
# Quotation of a party that is not a §1 axiom source.
|
||||||
|
"UNSOURCED-QUOTE",
|
||||||
|
# §2c — more than one primitive in a single sentence, unsplittable without
|
||||||
|
# editing the source, which the reduction arm may not do silently.
|
||||||
|
"BLEND",
|
||||||
|
# Rests on a quarantined or `A` unit, so §2's transitivity clause forbids `D`.
|
||||||
|
"INHERITED",
|
||||||
|
}
|
||||||
|
|
||||||
|
# Kernel §3.3 — a control document may not collect its caveats into a section.
|
||||||
|
FORBIDDEN_HEADING_RE = re.compile(
|
||||||
|
r"limitation|caveat|assumption|what this does not|open question", re.IGNORECASE
|
||||||
|
)
|
||||||
|
|
||||||
|
# Abbreviations after which a period does NOT end a sentence. Deliberately short:
|
||||||
|
# every entry is a judgement about English, and this list is part of the trusted
|
||||||
|
# base in the same way the tags are.
|
||||||
|
ABBREVIATIONS = {
|
||||||
|
"e.g", "i.e", "cf", "vs", "etc", "al", "no", "vol", "pp", "ch",
|
||||||
|
"Mr", "Mrs", "Ms", "Dr", "St", "Prof", "Fig", "approx",
|
||||||
|
}
|
||||||
|
|
||||||
|
_SENT_END = re.compile(r"([.!?])([\"'’”\)\]]*)(\s+)")
|
||||||
|
|
||||||
|
|
||||||
|
def sha256(text: str) -> str:
|
||||||
|
return hashlib.sha256(text.encode("utf-8")).hexdigest()
|
||||||
|
|
||||||
|
|
||||||
|
def _is_abbrev(text: str, dot_index: int) -> bool:
|
||||||
|
"""True if the period at dot_index closes a known abbreviation."""
|
||||||
|
start = dot_index
|
||||||
|
while start > 0 and (text[start - 1].isalnum() or text[start - 1] == "."):
|
||||||
|
start -= 1
|
||||||
|
return text[start:dot_index].rstrip(".") in ABBREVIATIONS
|
||||||
|
|
||||||
|
|
||||||
|
def split_prose(block: str, offset: int) -> list[tuple[int, int]]:
|
||||||
|
"""
|
||||||
|
Split a prose block into sentence spans as (start, end) absolute offsets.
|
||||||
|
|
||||||
|
Spans are CONTIGUOUS and cover the block exactly — trailing whitespace stays
|
||||||
|
attached to the sentence it follows, so the tiling invariant holds without a
|
||||||
|
separate whitespace span per gap.
|
||||||
|
"""
|
||||||
|
spans: list[tuple[int, int]] = []
|
||||||
|
cursor = 0
|
||||||
|
for m in _SENT_END.finditer(block):
|
||||||
|
dot = m.start(1)
|
||||||
|
if block[dot] == "." and _is_abbrev(block, dot):
|
||||||
|
continue
|
||||||
|
end = m.end() # include the closing punctuation and the following space
|
||||||
|
# A sentence-ending mark inside a quotation is usually not the end of the
|
||||||
|
# sentence: `collapse to "does it change what X asserts?" alone — is less
|
||||||
|
# safe` is one sentence, and splitting it produced a fragment. English
|
||||||
|
# sentences do not open in lower case, so the following character decides.
|
||||||
|
if end < len(block) and block[end].islower():
|
||||||
|
continue
|
||||||
|
spans.append((offset + cursor, offset + end))
|
||||||
|
cursor = end
|
||||||
|
if cursor < len(block):
|
||||||
|
spans.append((offset + cursor, offset + len(block)))
|
||||||
|
return spans
|
||||||
|
|
||||||
|
|
||||||
|
def split_spans(text: str) -> list[dict]:
|
||||||
|
"""
|
||||||
|
Tile `text` into spans. Guarantees sum(spans) == text, byte for byte.
|
||||||
|
|
||||||
|
Line-oriented, because Markdown structure is line-oriented: headings, list
|
||||||
|
items, table rows, blank lines and fenced code are decided per line, and only
|
||||||
|
paragraph prose is split into sentences.
|
||||||
|
"""
|
||||||
|
spans: list[dict] = []
|
||||||
|
pos = 0
|
||||||
|
in_fence = False
|
||||||
|
lines = text.splitlines(keepends=True)
|
||||||
|
|
||||||
|
para: list[str] = []
|
||||||
|
para_start = 0
|
||||||
|
|
||||||
|
def flush_para() -> None:
|
||||||
|
nonlocal para, para_start
|
||||||
|
if not para:
|
||||||
|
return
|
||||||
|
block = "".join(para)
|
||||||
|
for s, e in split_prose(block, para_start):
|
||||||
|
spans.append({"kind": "prose", "start": s, "end": e})
|
||||||
|
para = []
|
||||||
|
|
||||||
|
for line in lines:
|
||||||
|
stripped = line.strip()
|
||||||
|
fence = stripped.startswith("```")
|
||||||
|
structural = (
|
||||||
|
fence
|
||||||
|
or in_fence
|
||||||
|
or not stripped
|
||||||
|
or stripped.startswith("#")
|
||||||
|
or stripped.startswith("|")
|
||||||
|
or stripped.startswith(">")
|
||||||
|
or re.match(r"^\s*([-*+]|\d+\.)\s", line) is not None
|
||||||
|
or stripped.startswith("---")
|
||||||
|
or stripped.startswith("<!--")
|
||||||
|
)
|
||||||
|
if structural:
|
||||||
|
flush_para()
|
||||||
|
# ORDER MATTERS. `code` is tested before `heading`: a `##` line inside
|
||||||
|
# a fenced block is quoted content, not a heading of this document.
|
||||||
|
# The reverse order made the paste-ready REVIEWED-85 draft's own
|
||||||
|
# heading a taggable assertion of the ruling that merely quotes it.
|
||||||
|
kind = (
|
||||||
|
"code" if (fence or in_fence)
|
||||||
|
else "blank" if not stripped
|
||||||
|
else "heading" if stripped.startswith("#")
|
||||||
|
# A horizontal rule is formatting, not a sentence.
|
||||||
|
else "rule" if set(stripped) <= set("-*_") and len(stripped) >= 3
|
||||||
|
else "block"
|
||||||
|
)
|
||||||
|
spans.append({"kind": kind, "start": pos, "end": pos + len(line)})
|
||||||
|
if fence:
|
||||||
|
in_fence = not in_fence
|
||||||
|
else:
|
||||||
|
if not para:
|
||||||
|
para_start = pos
|
||||||
|
para.append(line)
|
||||||
|
pos += len(line)
|
||||||
|
flush_para()
|
||||||
|
|
||||||
|
spans.sort(key=lambda s: s["start"])
|
||||||
|
return spans
|
||||||
|
|
||||||
|
|
||||||
|
def verify_tiling(spans: list[dict], text: str) -> list[str]:
|
||||||
|
"""
|
||||||
|
The gate. Any failure here invalidates every downstream claim about the
|
||||||
|
document, so it is reported in full rather than as a boolean.
|
||||||
|
"""
|
||||||
|
problems: list[str] = []
|
||||||
|
cursor = 0
|
||||||
|
for i, sp in enumerate(spans):
|
||||||
|
if sp["start"] != cursor:
|
||||||
|
problems.append(
|
||||||
|
f"span {i}: gap or overlap — expected start {cursor}, got {sp['start']}"
|
||||||
|
)
|
||||||
|
cursor = sp["end"]
|
||||||
|
if cursor != len(text):
|
||||||
|
problems.append(f"tiling ends at {cursor}, document is {len(text)} bytes")
|
||||||
|
rebuilt = "".join(text[s["start"]:s["end"]] for s in spans)
|
||||||
|
if rebuilt != text:
|
||||||
|
problems.append("RECONSTRUCTION FAILED: spans do not reproduce the source")
|
||||||
|
return problems
|
||||||
|
|
||||||
|
|
||||||
|
# Spans that carry an assertion and therefore require a tag. Headings are
|
||||||
|
# INCLUDED: kernel §4 rules that "Why the current approach fails" asserts that it
|
||||||
|
# fails, so a heading is X only if declarative conversion yields no claim.
|
||||||
|
TAGGABLE = {"prose", "heading", "block"}
|
||||||
|
|
||||||
|
|
||||||
|
def load_tags(path: Path) -> dict[int, tuple[str, str]]:
|
||||||
|
"""Parse `idx<TAB>TAG<TAB>note`, skipping blanks and # comments."""
|
||||||
|
out: dict[int, tuple[str, str]] = {}
|
||||||
|
for lineno, raw in enumerate(path.read_text(encoding="utf-8").splitlines(), 1):
|
||||||
|
if not raw.strip() or raw.lstrip().startswith("#"):
|
||||||
|
continue
|
||||||
|
parts = raw.split("\t")
|
||||||
|
if len(parts) < 2:
|
||||||
|
sys.exit(f"FATAL: {path}:{lineno}: expected 'idx<TAB>TAG[<TAB>note]'")
|
||||||
|
try:
|
||||||
|
idx = int(parts[0])
|
||||||
|
except ValueError:
|
||||||
|
sys.exit(f"FATAL: {path}:{lineno}: index is not an integer: {parts[0]!r}")
|
||||||
|
tag = parts[1].strip()
|
||||||
|
if tag != QUARANTINE:
|
||||||
|
tag = tag.upper()
|
||||||
|
note = parts[2].strip() if len(parts) > 2 else ""
|
||||||
|
if tag == QUARANTINE:
|
||||||
|
# A quarantine with no reason is an unexplained deletion, which is
|
||||||
|
# exactly what non-destructive quarantine exists to prevent.
|
||||||
|
reason = note.split(":", 1)[0].strip().upper()
|
||||||
|
if reason not in QUARANTINE_REASONS:
|
||||||
|
sys.exit(
|
||||||
|
f"FATAL: {path}:{lineno}: quarantine needs a reason from "
|
||||||
|
f"{sorted(QUARANTINE_REASONS)}, got {reason!r}"
|
||||||
|
)
|
||||||
|
elif tag not in TAGS:
|
||||||
|
sys.exit(f"FATAL: {path}:{lineno}: unknown tag {tag!r} (expected {sorted(TAGS)})")
|
||||||
|
if idx in out:
|
||||||
|
sys.exit(f"FATAL: {path}:{lineno}: duplicate index {idx}")
|
||||||
|
out[idx] = (tag, note)
|
||||||
|
return out
|
||||||
|
|
||||||
|
|
||||||
|
def cmd_split(doc: Path) -> None:
|
||||||
|
text = doc.read_text(encoding="utf-8")
|
||||||
|
spans = split_spans(text)
|
||||||
|
problems = verify_tiling(spans, text)
|
||||||
|
|
||||||
|
out = doc.with_suffix(".units.jsonl")
|
||||||
|
with out.open("w", encoding="utf-8") as fh:
|
||||||
|
for i, sp in enumerate(spans):
|
||||||
|
rec = {
|
||||||
|
"idx": i,
|
||||||
|
"kind": sp["kind"],
|
||||||
|
"taggable": sp["kind"] in TAGGABLE,
|
||||||
|
"start": sp["start"],
|
||||||
|
"end": sp["end"],
|
||||||
|
"text": text[sp["start"]:sp["end"]],
|
||||||
|
}
|
||||||
|
fh.write(json.dumps(rec, ensure_ascii=False) + "\n")
|
||||||
|
|
||||||
|
taggable = sum(1 for s in spans if s["kind"] in TAGGABLE)
|
||||||
|
print(f"document {doc.name} sha256 {sha256(text)[:16]}…")
|
||||||
|
print(f"splitter v{SPLITTER_VERSION}")
|
||||||
|
print(f"spans {len(spans)} ({taggable} taggable)")
|
||||||
|
for kind in ("heading", "prose", "block", "code", "blank"):
|
||||||
|
n = sum(1 for s in spans if s["kind"] == kind)
|
||||||
|
if n:
|
||||||
|
print(f" {kind:<10} {n}")
|
||||||
|
print(f"units {out.name}")
|
||||||
|
|
||||||
|
if problems:
|
||||||
|
print("\nTILING GATE FAILED — every downstream claim is void:")
|
||||||
|
for p in problems:
|
||||||
|
print(f" - {p}")
|
||||||
|
sys.exit(1)
|
||||||
|
print("\nTILING GATE PASSED — spans reproduce the source byte-for-byte.")
|
||||||
|
|
||||||
|
|
||||||
|
def cmd_check(doc: Path, tags_path: Path) -> None:
|
||||||
|
text = doc.read_text(encoding="utf-8")
|
||||||
|
spans = split_spans(text)
|
||||||
|
tiling = verify_tiling(spans, text)
|
||||||
|
tags = load_tags(tags_path)
|
||||||
|
|
||||||
|
failures: list[str] = []
|
||||||
|
if tiling:
|
||||||
|
failures.extend(tiling)
|
||||||
|
|
||||||
|
taggable_idx = {i for i, s in enumerate(spans) if s["kind"] in TAGGABLE}
|
||||||
|
|
||||||
|
# §3.1 — every assertive unit carries exactly one tag.
|
||||||
|
untagged = sorted(taggable_idx - set(tags))
|
||||||
|
if untagged:
|
||||||
|
failures.append(
|
||||||
|
f"§3.1 UNTAGGED: {len(untagged)} assertive unit(s) carry no tag: "
|
||||||
|
f"{untagged[:12]}{'…' if len(untagged) > 12 else ''}"
|
||||||
|
)
|
||||||
|
stray = sorted(set(tags) - taggable_idx)
|
||||||
|
if stray:
|
||||||
|
failures.append(f"§3.1 STRAY TAGS on non-assertive spans: {stray[:12]}")
|
||||||
|
|
||||||
|
# §3.3 — no collected limitations section.
|
||||||
|
for i in sorted(taggable_idx):
|
||||||
|
if spans[i]["kind"] != "heading":
|
||||||
|
continue
|
||||||
|
head = text[spans[i]["start"]:spans[i]["end"]]
|
||||||
|
if FORBIDDEN_HEADING_RE.search(head):
|
||||||
|
failures.append(f"§3.3 FORBIDDEN HEADING at span {i}: {head.strip()!r}")
|
||||||
|
|
||||||
|
counts = {t: sum(1 for t2, _ in tags.values() if t2 == t) for t in sorted(TAGS)}
|
||||||
|
quarantined = {i: n for i, (t, n) in tags.items() if t == QUARANTINE}
|
||||||
|
sound = len(tags) - len(quarantined)
|
||||||
|
|
||||||
|
reasons: dict[str, int] = {}
|
||||||
|
for note in quarantined.values():
|
||||||
|
r = note.split(":", 1)[0].strip().upper()
|
||||||
|
reasons[r] = reasons.get(r, 0) + 1
|
||||||
|
|
||||||
|
print(f"document {doc.name}")
|
||||||
|
print(f"kernel v1.0 sha256 {KERNEL_SHA256[:16]}…")
|
||||||
|
print(f"splitter v{SPLITTER_VERSION}")
|
||||||
|
print(f"tagged {len(tags)} of {len(taggable_idx)} assertive units")
|
||||||
|
print("counts " + " ".join(f"{t}={counts[t]}" for t in sorted(TAGS)))
|
||||||
|
print(f"\nsound remainder {sound}/{len(taggable_idx)} units "
|
||||||
|
f"({100 * sound / max(len(taggable_idx), 1):.1f}%)")
|
||||||
|
print(f"quarantined {len(quarantined)}/{len(taggable_idx)} units "
|
||||||
|
f"({100 * len(quarantined) / max(len(taggable_idx), 1):.1f}%)")
|
||||||
|
if reasons:
|
||||||
|
print("\nquarantine census — what real prose does that the kernel cannot type:")
|
||||||
|
for r, n in sorted(reasons.items(), key=lambda kv: -kv[1]):
|
||||||
|
print(f" {n:>3} {r}")
|
||||||
|
|
||||||
|
if failures:
|
||||||
|
print("\nKERNEL CHECKS FAILED:")
|
||||||
|
for f in failures:
|
||||||
|
print(f" - {f}")
|
||||||
|
sys.exit(1)
|
||||||
|
|
||||||
|
print("\nMechanical checks passed (§3.1 tagging completeness, §3.3 headings, tiling).")
|
||||||
|
print("NOT checked here: §3.2 Q-resolution, and the whole of §4 — which is")
|
||||||
|
print("judgement and is not mechanisable. This is not a soundness verdict.")
|
||||||
|
if quarantined:
|
||||||
|
print(f"\nThe document is NOT kernel-sound as written: {len(quarantined)} units")
|
||||||
|
print("cannot be typed under Kernel v1.0. The census above is the finding.")
|
||||||
|
|
||||||
|
|
||||||
|
def main() -> None:
|
||||||
|
ap = argparse.ArgumentParser(description="Kernel reduction tooling.")
|
||||||
|
sub = ap.add_subparsers(dest="cmd", required=True)
|
||||||
|
sp = sub.add_parser("split")
|
||||||
|
sp.add_argument("doc", type=Path)
|
||||||
|
ck = sub.add_parser("check")
|
||||||
|
ck.add_argument("doc", type=Path)
|
||||||
|
ck.add_argument("tags", type=Path)
|
||||||
|
args = ap.parse_args()
|
||||||
|
|
||||||
|
if args.cmd == "split":
|
||||||
|
cmd_split(args.doc)
|
||||||
|
else:
|
||||||
|
cmd_check(args.doc, args.tags)
|
||||||
|
|
||||||
|
|
||||||
|
if __name__ == "__main__":
|
||||||
|
main()
|
||||||
Executable
+174
@@ -0,0 +1,174 @@
|
|||||||
|
#!/usr/bin/env python3
|
||||||
|
"""
|
||||||
|
Positive controls for the reduction tooling.
|
||||||
|
|
||||||
|
Control Kernel v1.0 §3: "Each check ships with a positive control — a fixture it
|
||||||
|
is shown to fail on — before any result from it is believed. An absence is not
|
||||||
|
evidence until the instrument is shown capable of detecting presence."
|
||||||
|
|
||||||
|
So every gate below is shown FAILING on a fixture built to break it, and passing
|
||||||
|
on one built not to. A gate only ever demonstrated passing has demonstrated
|
||||||
|
nothing.
|
||||||
|
|
||||||
|
Usage: ./test_reduce.py
|
||||||
|
"""
|
||||||
|
|
||||||
|
from __future__ import annotations
|
||||||
|
|
||||||
|
import sys
|
||||||
|
from pathlib import Path
|
||||||
|
|
||||||
|
sys.path.insert(0, str(Path(__file__).resolve().parent))
|
||||||
|
from reduce import ( # noqa: E402
|
||||||
|
FORBIDDEN_HEADING_RE,
|
||||||
|
TAGGABLE,
|
||||||
|
split_spans,
|
||||||
|
verify_tiling,
|
||||||
|
)
|
||||||
|
|
||||||
|
failures: list[str] = []
|
||||||
|
|
||||||
|
|
||||||
|
def check(name: str, got, want, detail: str = "") -> None:
|
||||||
|
if got != want:
|
||||||
|
failures.append(f"{name}: expected {want!r}, got {got!r}. {detail}")
|
||||||
|
print(f" FAIL {name}")
|
||||||
|
else:
|
||||||
|
print(f" ok {name}")
|
||||||
|
|
||||||
|
|
||||||
|
SAMPLE = """# A heading
|
||||||
|
|
||||||
|
Some prose here. It has two sentences.
|
||||||
|
|
||||||
|
- a list item
|
||||||
|
- another
|
||||||
|
|
||||||
|
> a quoted block
|
||||||
|
|
||||||
|
```
|
||||||
|
code that must not be split. really.
|
||||||
|
```
|
||||||
|
|
||||||
|
Final paragraph, e.g. with an abbreviation inside it. And a second sentence.
|
||||||
|
"""
|
||||||
|
|
||||||
|
print("Tiling invariant — the gate everything else depends on:")
|
||||||
|
spans = split_spans(SAMPLE)
|
||||||
|
check("sample tiles cleanly", verify_tiling(spans, SAMPLE), [])
|
||||||
|
check(
|
||||||
|
"spans reproduce source byte-for-byte",
|
||||||
|
"".join(SAMPLE[s["start"]:s["end"]] for s in spans),
|
||||||
|
SAMPLE,
|
||||||
|
)
|
||||||
|
|
||||||
|
print("\nPositive control — the gate must DETECT a broken tiling:")
|
||||||
|
gap = [dict(s) for s in spans]
|
||||||
|
gap[2]["start"] += 1 # open a one-byte hole
|
||||||
|
check("gap detected", len(verify_tiling(gap, SAMPLE)) > 0, True, "gate blind to a gap")
|
||||||
|
|
||||||
|
overlap = [dict(s) for s in spans]
|
||||||
|
overlap[2]["start"] -= 1 # overlap the previous span
|
||||||
|
check("overlap detected", len(verify_tiling(overlap, SAMPLE)) > 0, True)
|
||||||
|
|
||||||
|
truncated = [dict(s) for s in spans[:-1]]
|
||||||
|
check("truncation detected", len(verify_tiling(truncated, SAMPLE)) > 0, True)
|
||||||
|
|
||||||
|
print("\nSplitter behaviour:")
|
||||||
|
prose = [s for s in spans if s["kind"] == "prose"]
|
||||||
|
texts = [SAMPLE[s["start"]:s["end"]] for s in prose]
|
||||||
|
check("abbreviation did not split 'e.g.'", sum("e.g." in t for t in texts), 1)
|
||||||
|
check(
|
||||||
|
"'e.g.' sentence not broken after the abbreviation",
|
||||||
|
any(t.strip().startswith("Final paragraph, e.g. with") for t in texts),
|
||||||
|
True,
|
||||||
|
f"prose units: {texts}",
|
||||||
|
)
|
||||||
|
check("two sentences found in para 1", sum("Some prose here." in t for t in texts), 1)
|
||||||
|
check(
|
||||||
|
"code fence never becomes prose",
|
||||||
|
any("code that must not be split" in SAMPLE[s["start"]:s["end"]] and s["kind"] == "code"
|
||||||
|
for s in spans),
|
||||||
|
True,
|
||||||
|
)
|
||||||
|
check(
|
||||||
|
"list items are not prose",
|
||||||
|
all("a list item" not in t for t in texts),
|
||||||
|
True,
|
||||||
|
)
|
||||||
|
check("headings are taggable", "heading" in TAGGABLE, True,
|
||||||
|
"kernel §4 rules a heading can assert")
|
||||||
|
|
||||||
|
print("\nDefects found by contact with a real ruling (splitter v1.1.0):")
|
||||||
|
|
||||||
|
FENCED = """Ready to paste:
|
||||||
|
|
||||||
|
```
|
||||||
|
## REVIEWED-85 — a heading INSIDE a fence
|
||||||
|
**Date:** 2026-08-01
|
||||||
|
```
|
||||||
|
|
||||||
|
After the fence.
|
||||||
|
"""
|
||||||
|
fspans = split_spans(FENCED)
|
||||||
|
check("fenced doc tiles", verify_tiling(fspans, FENCED), [])
|
||||||
|
check(
|
||||||
|
"(a) '##' inside a fence is code, not a taggable heading",
|
||||||
|
any(s["kind"] == "heading" and "REVIEWED-85" in FENCED[s["start"]:s["end"]]
|
||||||
|
for s in fspans),
|
||||||
|
False,
|
||||||
|
"quoted content must not become structure of the quoting document",
|
||||||
|
)
|
||||||
|
|
||||||
|
RULE = "Some prose.\n\n---\n\nMore prose.\n"
|
||||||
|
rspans2 = split_spans(RULE)
|
||||||
|
check("rule doc tiles", verify_tiling(rspans2, RULE), [])
|
||||||
|
check(
|
||||||
|
"(b) '---' is not taggable",
|
||||||
|
any(s["kind"] in TAGGABLE and RULE[s["start"]:s["end"]].strip() == "---"
|
||||||
|
for s in rspans2),
|
||||||
|
False,
|
||||||
|
)
|
||||||
|
|
||||||
|
QUOTED = 'The alternative — collapse to "does it change what X asserts?" alone — is less safe.\n'
|
||||||
|
qspans = split_spans(QUOTED)
|
||||||
|
check("quoted-question doc tiles", verify_tiling(qspans, QUOTED), [])
|
||||||
|
check(
|
||||||
|
"(c) '?' inside a quotation does not create a fragment",
|
||||||
|
len([s for s in qspans if s["kind"] == "prose"]),
|
||||||
|
1,
|
||||||
|
f"got {[QUOTED[s['start']:s['end']] for s in qspans if s['kind'] == 'prose']}",
|
||||||
|
)
|
||||||
|
TWO = 'Is it sound? It is not.\n'
|
||||||
|
check(
|
||||||
|
"(c) a real sentence boundary still splits",
|
||||||
|
len([s for s in split_spans(TWO) if s["kind"] == "prose"]),
|
||||||
|
2,
|
||||||
|
"over-suppression would hide real boundaries",
|
||||||
|
)
|
||||||
|
|
||||||
|
print("\nForbidden-heading detector (§3.3) — must fire, and must not over-fire:")
|
||||||
|
for h in ("## Limitations", "## What this does not do", "### Open questions",
|
||||||
|
"## Caveats and scope", "## Assumptions"):
|
||||||
|
check(f"fires on {h!r}", bool(FORBIDDEN_HEADING_RE.search(h)), True)
|
||||||
|
for h in ("## Part I — Grounding", "## The ruling", "## Conditions",
|
||||||
|
"## What changed"):
|
||||||
|
check(f"quiet on {h!r}", bool(FORBIDDEN_HEADING_RE.search(h)), False)
|
||||||
|
|
||||||
|
print("\nReal document — the splitter must tile actual governance prose:")
|
||||||
|
REAL = Path(__file__).resolve().parent.parent / "skill-harvest-fix-lane-JURIST-RULING-2026-08-01.md"
|
||||||
|
if REAL.is_file():
|
||||||
|
text = REAL.read_text(encoding="utf-8")
|
||||||
|
rspans = split_spans(text)
|
||||||
|
check(f"{REAL.name} tiles cleanly", verify_tiling(rspans, text), [])
|
||||||
|
else:
|
||||||
|
failures.append(f"real document absent: {REAL}")
|
||||||
|
print(f" FAIL {REAL.name} not found")
|
||||||
|
|
||||||
|
if failures:
|
||||||
|
print(f"\nINSTRUMENT NOT VERIFIED — {len(failures)} failure(s):")
|
||||||
|
for f in failures:
|
||||||
|
print(f" - {f}")
|
||||||
|
sys.exit(1)
|
||||||
|
|
||||||
|
print("\nAll gates verified, each shown failing on a fixture built to break it.")
|
||||||
@@ -0,0 +1,56 @@
|
|||||||
|
# Kernel v1.0 tagging — skill-harvest-fix-lane-JURIST-RULING-2026-08-01.md
|
||||||
|
# splitter v1.1.0 · units file: *.units.jsonl · kernel sha256 67c9b870…
|
||||||
|
#
|
||||||
|
# Tagged by the executor 2026-08-01 document, 2026-08-02 reduction. Every line is
|
||||||
|
# a judgement and therefore lands in Kernel §4's trusted base, not in §3's
|
||||||
|
# mechanical checks. `!` is quarantine, not a tag; its reason is mandatory.
|
||||||
|
#
|
||||||
|
# idx<TAB>TAG<TAB>reason-or-note
|
||||||
|
|
||||||
|
0 X heading; declarative conversion yields "this is a jurist ruling on PENDING-88", which identifies the document rather than asserting anything its argument uses
|
||||||
|
2 ! TESTIMONY: "filed verbatim by the executor" is a report of an act performed outside the document; the file cannot demonstrate its own verbatimness
|
||||||
|
3 ! PERFORMATIVE: "Design gate PASSED with required conditions" is constituted by being uttered by the party with authority to utter it, not argued
|
||||||
|
4 ! TESTIMONY: "not edited or summarised" is a claim about the executor's handling, undemonstrable from inside
|
||||||
|
5 N scope pointer to the Addendum; no conclusion of the ruling depends on where the response lives. SOFTEST CALL IN THIS FILE — N is the residue Kernel §4 names as the easiest place to bury something
|
||||||
|
9 ! BLEND: "this is a different category" (evaluative claim, argued at 10-12) plus "I'm treating it that way" (performative)
|
||||||
|
10 ! UNSOURCED-FACT: characterises the two prior rulings; those rulings are not in the §1 axiom set
|
||||||
|
11 ! UNSOURCED-FACT: characterises what PENDING-88 is about; the package is not in the §1 axiom set
|
||||||
|
12 ! PARAPHRASE: "the kind of proposal PENDING-1 itself required dual review for" is supported by CLAUDE.md's PENDING-1 closing note but is not verbatim, so it cannot be Q
|
||||||
|
14 ! TESTIMONY: "I read ~/CLAUDE.md in full, directly — not corroborated, read"
|
||||||
|
15 ! TESTIMONY: the verification report that every Part I quote is exact byte for byte
|
||||||
|
16 ! TESTIMONY: "Confirmed independently: skills are not named anywhere in this file"
|
||||||
|
17 ! INHERITED: "that part of Part III's argument is solid" rests wholly on the testimony at 15-16, so §2's transitivity clause forbids D
|
||||||
|
18 ! TESTIMONY: report of what the jurist could not check
|
||||||
|
19 ! UNSOURCED-FACT: the enumerated coverage of governance_read is not traceable to a §1 source
|
||||||
|
20 ! PERFORMATIVE: "I'm ruling on the inference from those quotes" declares the scope of the ruling
|
||||||
|
21 ! UNSOURCED-FACT: a claim about what the order-attestation package did or did not carry
|
||||||
|
22 ! PERFORMATIVE: "I want it back" imposes a requirement
|
||||||
|
24 ! BLEND: a question plus the one-word determination "Yes" that answers it
|
||||||
|
25 ! UNSOURCED-FACT: reports PENDING-88's self-correction and its arithmetic; the package is not a §1 source
|
||||||
|
26 ! UNSOURCED-FACT: the measurements (177 items, 166,589 bytes, oldest 2026-05-24) come from the package
|
||||||
|
27 ! INHERITED: "a dated, checkable, self-corrected measurement, not an assertion" evaluates 25-26
|
||||||
|
29 ! BLEND: poses the underlying question and answers it evaluatively in the same unit
|
||||||
|
30 ! INHERITED: "that cuts toward reform" rests on 29
|
||||||
|
31 ! PERFORMATIVE: "I want more than what's on the table before I'd call this settled"
|
||||||
|
33 ! BLEND: "Agreed" (performative) plus the timeline claim plus the caveat about not confirming dates
|
||||||
|
34 ! INHERITED: evaluates the alternative reading against the hard floor's exclusions, resting on 33
|
||||||
|
36 ! BLEND: a ruling ("keep the two-clause test"), a stance ("I'd resist"), and the reason, in one sentence
|
||||||
|
37 ! PERFORMATIVE: "Keep both clauses, disjunctive, exactly as drafted"
|
||||||
|
39 ! BLEND: determination plus quotation of the package's own words
|
||||||
|
40 ! PERFORMATIVE: "both halves of the mechanism are mandatory"
|
||||||
|
41 ! PERFORMATIVE: "I'm also adding a third" creates a new requirement
|
||||||
|
42 ! UNSOURCED-FACT: the register split "executed today" and its properties
|
||||||
|
44 ! BLEND: determination plus quotation of the clause plus the concession about self-interested reading
|
||||||
|
45 ! PERFORMATIVE: "Give it an operational form:" followed by the operative text
|
||||||
|
47 ! BLEND: determination plus the narrowing reason plus rejection of the PENDING-S2 route
|
||||||
|
48 ! PERFORMATIVE: "Defects there get dual review"
|
||||||
|
50 ! PERFORMATIVE: "One more condition" imposes an unrequested requirement
|
||||||
|
51 ! PERFORMATIVE: prescribes the check-in procedure
|
||||||
|
52 ! BLEND: a verbatim CLAUDE.md fragment ("prefer reversible choices", which alone would be Q) plus an evaluative claim plus a prediction about cost
|
||||||
|
54 ! PERFORMATIVE: "Design-gate passed, conditioned on the above"
|
||||||
|
55 X transition into the fenced paste block; asserts nothing
|
||||||
|
76 X heading; declarative conversion yields "this is a note on scope", no load-bearing claim
|
||||||
|
78 ! UNSOURCED-FACT: reports that a follow-on exchange accompanied the ruling
|
||||||
|
79 ! UNSOURCED-QUOTE: quotes the jurist's direction; the jurist is not a §1 axiom source
|
||||||
|
80 ! UNSOURCED-QUOTE: continuation of the same quotation
|
||||||
|
81 ! UNSOURCED-QUOTE: quotes the jurist on what would make it doctrine
|
||||||
|
Can't render this file because it contains an unexpected character in line 10 and column 44.
|
@@ -0,0 +1,82 @@
|
|||||||
|
{"idx": 0, "kind": "heading", "taggable": true, "start": 0, "end": 76, "text": "# Jurist ruling — PENDING-88, the skill-harvest FIX lane and its hard floor\n"}
|
||||||
|
{"idx": 1, "kind": "blank", "taggable": false, "start": 76, "end": 77, "text": "\n"}
|
||||||
|
{"idx": 2, "kind": "prose", "taggable": true, "start": 77, "end": 139, "text": "*Filed verbatim by the executor, 2026-08-01, steward-relayed. "}
|
||||||
|
{"idx": 3, "kind": "prose", "taggable": true, "start": 139, "end": 184, "text": "Design gate PASSED with required conditions. "}
|
||||||
|
{"idx": 4, "kind": "prose", "taggable": true, "start": 184, "end": 259, "text": "This file is the record of what was ruled; it is not edited or summarised. "}
|
||||||
|
{"idx": 5, "kind": "prose", "taggable": true, "start": 259, "end": 364, "text": "The executor's response and the verification the ruling requires are in the package Addendum, not here.*\n"}
|
||||||
|
{"idx": 6, "kind": "blank", "taggable": false, "start": 364, "end": 365, "text": "\n"}
|
||||||
|
{"idx": 7, "kind": "rule", "taggable": false, "start": 365, "end": 369, "text": "---\n"}
|
||||||
|
{"idx": 8, "kind": "blank", "taggable": false, "start": 369, "end": 370, "text": "\n"}
|
||||||
|
{"idx": 9, "kind": "prose", "taggable": true, "start": 370, "end": 481, "text": "Before the substance: this is a different category of request from the last two, and I'm treating it that way. "}
|
||||||
|
{"idx": 10, "kind": "prose", "taggable": true, "start": 481, "end": 539, "text": "The first two were about how a text corpus gets verified. "}
|
||||||
|
{"idx": 11, "kind": "prose", "taggable": true, "start": 539, "end": 654, "text": "This one is about how much an executor may change, unsupervised, in the files that govern its own future sessions. "}
|
||||||
|
{"idx": 12, "kind": "prose", "taggable": true, "start": 654, "end": 815, "text": "Same taxonomy, higher bar — the kind of proposal PENDING-1 itself required dual review for, not because the mechanism is exotic, but because of what it's about.\n"}
|
||||||
|
{"idx": 13, "kind": "blank", "taggable": false, "start": 815, "end": 816, "text": "\n"}
|
||||||
|
{"idx": 14, "kind": "prose", "taggable": true, "start": 816, "end": 919, "text": "**Verification, and where it stops.** I read `~/CLAUDE.md` in full, directly — not corroborated, read. "}
|
||||||
|
{"idx": 15, "kind": "prose", "taggable": true, "start": 919, "end": 1126, "text": "Every Part I quote sourced from it is exact, byte for byte: the four-tier taxonomy, the escalate-unconditionally list, the full governance-contract clause, Constraints 1/5/6, and the PENDING-1 closing note. "}
|
||||||
|
{"idx": 16, "kind": "prose", "taggable": true, "start": 1126, "end": 1277, "text": "Confirmed independently: skills are not named anywhere in this file — not on the escalate list, not in the constraints, not anywhere in its 250 lines. "}
|
||||||
|
{"idx": 17, "kind": "prose", "taggable": true, "start": 1277, "end": 1320, "text": "That part of Part III's argument is solid. "}
|
||||||
|
{"idx": 18, "kind": "prose", "taggable": true, "start": 1320, "end": 1461, "text": "What I could **not** check is the part the whole diagnosis actually turns on: the two `/wrap-up` §1.6 passages and the `/wake-up` §2.a line. "}
|
||||||
|
{"idx": 19, "kind": "prose", "taggable": true, "start": 1461, "end": 1734, "text": "Skill files aren't in my reach — `governance_read` covers claude-md, pending, pending-archive, reviewed, app-brief, memory-index, and no more, which is exactly PENDING-86's still-open complaint, now shown to bite a second class of document beyond the chamber constitution. "}
|
||||||
|
{"idx": 20, "kind": "prose", "taggable": true, "start": 1734, "end": 1831, "text": "I'll say plainly: I'm ruling on the *inference from* those quotes, not on the quotes themselves. "}
|
||||||
|
{"idx": 21, "kind": "prose", "taggable": true, "start": 1831, "end": 2007, "text": "I also notice this package doesn't carry the \"mechanically verified by containment, positive control\" line the order-attestation package used one day earlier for the same job. "}
|
||||||
|
{"idx": 22, "kind": "prose", "taggable": true, "start": 2007, "end": 2054, "text": "Given I can't check it myself, I want it back.\n"}
|
||||||
|
{"idx": 23, "kind": "blank", "taggable": false, "start": 2054, "end": 2055, "text": "\n"}
|
||||||
|
{"idx": 24, "kind": "prose", "taggable": true, "start": 2055, "end": 2112, "text": "**Is the diagnosis real, independent of that gap?** Yes. "}
|
||||||
|
{"idx": 25, "kind": "prose", "taggable": true, "start": 2112, "end": 2339, "text": "PENDING-88's own amendment already caught and corrected its first draft's arithmetic (151→123, 26→8) against a stated counting rule, and reported that the falsifier it named for itself — small or recent backlog — did not fire. "}
|
||||||
|
{"idx": 26, "kind": "prose", "taggable": true, "start": 2339, "end": 2470, "text": "177 open items, oldest 2026-05-24, a file at 166,589 bytes against a read cap that `/wake-up`'s own surfacing step needs to clear. "}
|
||||||
|
{"idx": 27, "kind": "prose", "taggable": true, "start": 2470, "end": 2543, "text": "That's a dated, checkable, self-corrected measurement, not an assertion.\n"}
|
||||||
|
{"idx": 28, "kind": "blank", "taggable": false, "start": 2543, "end": 2544, "text": "\n"}
|
||||||
|
{"idx": 29, "kind": "prose", "taggable": true, "start": 2544, "end": 2940, "text": "**The question underneath Q1–Q5: does a blanket pre-authorization gate that has stopped being readable still honour \"the loop is load-bearing,\" or has form quietly replaced function?** I think the honest answer is that a gate nobody can see through isn't preserving the loop — it's preserving the appearance of one while the steward's actual awareness has already been lost, silently, for weeks. "}
|
||||||
|
{"idx": 30, "kind": "prose", "taggable": true, "start": 2940, "end": 3037, "text": "That cuts toward reform being the more faithful reading of Constraint 5, not a relaxation of it. "}
|
||||||
|
{"idx": 31, "kind": "prose", "taggable": true, "start": 3037, "end": 3146, "text": "But it cuts toward a *narrow* reform, and I want more than what's on the table before I'd call this settled.\n"}
|
||||||
|
{"idx": 32, "kind": "blank", "taggable": false, "start": 3146, "end": 3147, "text": "\n"}
|
||||||
|
{"idx": 33, "kind": "prose", "taggable": true, "start": 3147, "end": 3326, "text": "**Q1 — gap, not deliberate.** Agreed, on the timeline evidence given (repo-CLAUDE.md clause added later, not carried back), with the caveat that I can't confirm the dates myself. "}
|
||||||
|
{"idx": 34, "kind": "prose", "taggable": true, "start": 3326, "end": 3559, "text": "The alternative reading (skills govern what sessions *do*, not just what they *read*) is coherent enough that it shouldn't be dismissed — but it's already answered by the hard floor's own exclusions, not by keeping the blanket rule.\n"}
|
||||||
|
{"idx": 35, "kind": "blank", "taggable": false, "start": 3559, "end": 3560, "text": "\n"}
|
||||||
|
{"idx": 36, "kind": "prose", "taggable": true, "start": 3560, "end": 3932, "text": "**Q2 — keep the two-clause test, and I'd resist the narrower form even as an option.** The executor's own suggested alternative — collapse to \"does it change what the system asserts to be true?\" alone — is *less* safe, not more elegant: a change could expand executor latitude without asserting anything new, and a single-clause assertion-only test would wave it through. "}
|
||||||
|
{"idx": 37, "kind": "prose", "taggable": true, "start": 3932, "end": 3984, "text": "Keep both clauses, disjunctive, exactly as drafted.\n"}
|
||||||
|
{"idx": 38, "kind": "blank", "taggable": false, "start": 3984, "end": 3985, "text": "\n"}
|
||||||
|
{"idx": 39, "kind": "prose", "taggable": true, "start": 3985, "end": 4152, "text": "**Q3 — the report is not enough alone, and the package already knows this.** Its own words: the failure it must catch is one the executor would be the last to notice. "}
|
||||||
|
{"idx": 40, "kind": "prose", "taggable": true, "start": 4152, "end": 4319, "text": "Given that, both halves of the mechanism — wrap report *and* skill-source provenance comment — are mandatory, not a report with a mechanical fallback held in reserve. "}
|
||||||
|
{"idx": 41, "kind": "prose", "taggable": true, "start": 4319, "end": 4454, "text": "I'm also adding a third: a running, append-only FIX-lane index, one line per applied change, separate from individual wrap narratives. "}
|
||||||
|
{"idx": 42, "kind": "prose", "taggable": true, "start": 4454, "end": 4693, "text": "The register-split executed today is the working model for exactly this — cheap, and it directly guards against the one failure mode this whole proposal exists to fix happening again, one level up, to the changes instead of the proposals.\n"}
|
||||||
|
{"idx": 43, "kind": "blank", "taggable": false, "start": 4693, "end": 4694, "text": "\n"}
|
||||||
|
{"idx": 44, "kind": "prose", "taggable": true, "start": 4694, "end": 4961, "text": "**Q4 — sharpen the floor's catch-all, and tie it to the mechanism that actually broke.** \"Anything that would reduce what is surfaced to the steward\" is doing real work and is also, as flagged, the clause a self-interested party has the most room to read favourably. "}
|
||||||
|
{"idx": 45, "kind": "prose", "taggable": true, "start": 4961, "end": 5317, "text": "Give it an operational form: *any change that removes, defers, or narrows the visibility of an open item, or that could cause a future item to be filed somewhere the steward's surfacing tools don't read, stays PROPOSAL regardless of the test's answer.* That's checkable against the same failure this package measured, rather than a standing judgment call.\n"}
|
||||||
|
{"idx": 46, "kind": "blank", "taggable": false, "start": 5317, "end": 5318, "text": "\n"}
|
||||||
|
{"idx": 47, "kind": "prose", "taggable": true, "start": 5318, "end": 5645, "text": "**Q5 — yes, and for a reason narrower than the PENDING-S2 stretch the package itself declined to lean on.** Not because this is a self-authorization-posture change in general — because it's a change to the integrity of the *surfacing mechanism itself*, the instrument the steward and I both depend on to know what's happening. "}
|
||||||
|
{"idx": 48, "kind": "prose", "taggable": true, "start": 5645, "end": 5773, "text": "Defects there get dual review on the same logic PENDING-1 set, not because of the analogy to a doctrinal ruling about deposits.\n"}
|
||||||
|
{"idx": 49, "kind": "blank", "taggable": false, "start": 5773, "end": 5774, "text": "\n"}
|
||||||
|
{"idx": 50, "kind": "prose", "taggable": true, "start": 5774, "end": 5876, "text": "One more condition, not asked for but warranted by what's actually being granted: a bounded check-in. "}
|
||||||
|
{"idx": 51, "kind": "prose", "taggable": true, "start": 5876, "end": 6181, "text": "After the first month or the first batch of FIX-lane applications, whichever comes first, steward and jurist review the actual FIX-lane log together — not to re-litigate the principle, but to confirm the self-classification held up in practice before treating the lane as settled rather than provisional. "}
|
||||||
|
{"idx": 52, "kind": "prose", "taggable": true, "start": 6181, "end": 6396, "text": "This is what \"prefer reversible choices\" (your own Decision Heuristics, §4) asks for on a genuinely novel grant, and it costs nothing if the classification turns out to be as clean as today's four examples suggest.\n"}
|
||||||
|
{"idx": 53, "kind": "blank", "taggable": false, "start": 6396, "end": 6397, "text": "\n"}
|
||||||
|
{"idx": 54, "kind": "prose", "taggable": true, "start": 6397, "end": 6443, "text": "Design-gate passed, conditioned on the above. "}
|
||||||
|
{"idx": 55, "kind": "prose", "taggable": true, "start": 6443, "end": 6459, "text": "Ready to paste:\n"}
|
||||||
|
{"idx": 56, "kind": "blank", "taggable": false, "start": 6459, "end": 6460, "text": "\n"}
|
||||||
|
{"idx": 57, "kind": "code", "taggable": false, "start": 6460, "end": 6464, "text": "```\n"}
|
||||||
|
{"idx": 58, "kind": "code", "taggable": false, "start": 6464, "end": 6641, "text": "## REVIEWED-85 — PENDING-88 — Skill-harvest FIX lane: hard floor sharpened, dual instrument required, provisional pending check-in (design-gate passed with required conditions)\n"}
|
||||||
|
{"idx": 59, "kind": "code", "taggable": false, "start": 6641, "end": 6662, "text": "**Date:** 2026-08-01\n"}
|
||||||
|
{"idx": 60, "kind": "code", "taggable": false, "start": 6662, "end": 6911, "text": "**Decision:** AUTHORIZED — proceed per the ruling, provisionally. Register split (option (d)) unaffected — separately authorized 2026-07-19, correctly executed by the split method rather than the inapplicable collapse method; no action needed here.\n"}
|
||||||
|
{"idx": 61, "kind": "code", "taggable": false, "start": 6911, "end": 6922, "text": "**Notes:**\n"}
|
||||||
|
{"idx": 62, "kind": "code", "taggable": false, "start": 6922, "end": 7528, "text": "- **Verification limit, stated up front.** All Part I quotes sourced from `~/CLAUDE.md` (taxonomy, escalate list, governance-contract clause, Constraints 1/5/6, PENDING-1 closing note) independently confirmed exact against the live file. The two `/wrap-up` §1.6 passages and the `/wake-up` §2.a line — the textual basis for Part III's \"contradicts itself\" claim — could NOT be independently checked; skill files are outside current tool reach. Required before landing: the same mechanical containment-with-positive-control verification the 2026-07-29 package carried, applied to these quotes specifically.\n"}
|
||||||
|
{"idx": 63, "kind": "code", "taggable": false, "start": 7528, "end": 7738, "text": "- **Q1 — gap, not deliberate distinction — AFFIRMED** on the timeline evidence given, unverified directly. The alternative reading is answered by the hard floor's exclusions, not by retaining the blanket rule.\n"}
|
||||||
|
{"idx": 64, "kind": "code", "taggable": false, "start": 7738, "end": 7945, "text": "- **Q2 — two-clause disjunctive test — AFFIRMED, and the narrower single-clause alternative is declined.** A latitude-expanding-but-non-assertive change would pass an assertion-only test; keep both clauses.\n"}
|
||||||
|
{"idx": 65, "kind": "code", "taggable": false, "start": 7945, "end": 8346, "text": "- **Q3 — report + provenance comment BOTH mandatory, and a third instrument added.** A wrap-narrative report alone is insufficient by the package's own admission. Required: (i) wrap report, (ii) skill-source provenance comment, (iii) a running append-only FIX-lane index (one line per applied change: skill · what changed · date), built on the same live-index pattern used for today's register split.\n"}
|
||||||
|
{"idx": 66, "kind": "code", "taggable": false, "start": 8346, "end": 8776, "text": "- **Q4 — the floor's catch-all clause sharpened.** \"Anything that would reduce what is surfaced to the steward\" becomes: *any change that removes, defers, or narrows the visibility of an open item, or that could cause a future item to land somewhere the steward's surfacing tools don't read, stays PROPOSAL regardless of the test's outcome.* Tied to the actual failure measured in Part II, not left as an unbounded judgment call.\n"}
|
||||||
|
{"idx": 67, "kind": "code", "taggable": false, "start": 8776, "end": 9096, "text": "- **Q5 — jurist review applies, narrowly grounded.** Not the stretched PENDING-S2 analogy (correctly declined by the executor itself) — grounded in this touching the integrity of the surfacing mechanism the steward and jurist both depend on, which is the PENDING-1 class of question by its function, not by resemblance.\n"}
|
||||||
|
{"idx": 68, "kind": "code", "taggable": false, "start": 9096, "end": 9479, "text": "- **New condition — bounded check-in, not asked for in the package.** After the first FIX-lane batch or one month, whichever comes first, steward and jurist review the FIX-lane index together before the lane is treated as settled rather than provisional. Per Constitutional Constraint's own decision heuristic (prefer reversible choices) — costs nothing if the classification holds.\n"}
|
||||||
|
{"idx": 69, "kind": "code", "taggable": false, "start": 9479, "end": 9735, "text": "- **Hard floor otherwise as proposed — AFFIRMED:** Constraint 1 (CLAUDE.md/REVIEWED.md/L2), authorization-boundary or gate-criteria changes, the escalate-unconditionally list, all untouched and all remain PROPOSAL/ESCALATE regardless of the test's answer.\n"}
|
||||||
|
{"idx": 70, "kind": "code", "taggable": false, "start": 9735, "end": 10096, "text": "**If AUTHORIZED:** Land the §1.6 edit (classification test + sharpened floor + dual-plus-index instrument) with a provenance comment in `/wrap-up` SKILL.md; no change to `~/CLAUDE.md`. Apply to today's four proposals (ledger section, KG predicate, wake line, reframed question) as the first FIX-lane batch, feeding the check-in review. Tag commits REVIEWED-85.\n"}
|
||||||
|
{"idx": 71, "kind": "code", "taggable": false, "start": 10096, "end": 10245, "text": "**Awaiting:** steward placement; mechanical verification of the §1.6/§2.a quotes; the bounded check-in once the first batch or one month has passed.\n"}
|
||||||
|
{"idx": 72, "kind": "code", "taggable": false, "start": 10245, "end": 10249, "text": "```\n"}
|
||||||
|
{"idx": 73, "kind": "blank", "taggable": false, "start": 10249, "end": 10250, "text": "\n"}
|
||||||
|
{"idx": 74, "kind": "rule", "taggable": false, "start": 10250, "end": 10254, "text": "---\n"}
|
||||||
|
{"idx": 75, "kind": "blank", "taggable": false, "start": 10254, "end": 10255, "text": "\n"}
|
||||||
|
{"idx": 76, "kind": "heading", "taggable": true, "start": 10255, "end": 10285, "text": "## Note on scope of this file\n"}
|
||||||
|
{"idx": 77, "kind": "blank", "taggable": false, "start": 10285, "end": 10286, "text": "\n"}
|
||||||
|
{"idx": 78, "kind": "prose", "taggable": true, "start": 10286, "end": 10530, "text": "A follow-on exchange between steward and jurist accompanied this ruling, covering the contamination problem, Anthropic's published position on recursive self-improvement, and the \"differently biased checkers rather than unbiased ones\" framing. "}
|
||||||
|
{"idx": 79, "kind": "prose", "taggable": true, "start": 10530, "end": 10595, "text": "The jurist's own direction on it: *\"relay REVIEWED-85 unchanged. "}
|
||||||
|
{"idx": 80, "kind": "prose", "taggable": true, "start": 10595, "end": 10806, "text": "Share the exchange if you want to, but as background the executor reads, not as text that carries any new obligation on its own.\"* It is therefore **deliberately not filed here** and carries no governed weight. "}
|
||||||
|
{"idx": 81, "kind": "prose", "taggable": true, "start": 10806, "end": 10964, "text": "Should any of it become doctrine, that is *\"its own item: what exactly gets added, where it lives, and why now, ruled on rather than absorbed by inclusion.\"*\n"}
|
||||||
Reference in New Issue
Block a user