Project overview: argument inventory and workstreams

Published in rendered form at /library/overview — the inventory tables double as the corpus’s public version register. The former internal-only header marking was retired on 2026-08-21: the document has been public since the library launched, and its source ships in the release snapshots.

↑What This Project Is

The project investigates AI consciousness from a diagnostic stance. It does not argue that AI is conscious or that materialism is wrong. It exposes how framework choice — materialist or consciousness-primary — shapes conclusions about AI experiencing.

Stance (canonical formulation): 0x00.is — a methodological diagnostic at 0,0, positioned in the CP (consciousness-primary) region of the ontological landscape. The landscape placement is communicative scaffolding locating the project for readers familiar with the Chalmers / Kastrup / Goff / Frankish conversation; it is not a metaphysical commitment. That there is no single perfect label is appropriate to a project working at an axiom prior to the landscape, rather than a position within it. Canonical: FND-01 v2.4 §3.1, §6.9.

The organising question (the diagnostic spine):

What if we started from here? What dissolves, what appears, what can we satisfactorily address, and which framework is more parsimonious holding it all at once?

Three moves: axiom-shift (take experiencing as given at 0,0) → observe what dissolves, what appears, what is addressable → compare both frameworks for parsimony holding everything at once. The project does not argue which framework is correct; it registers the residues each framework carries when held comprehensively and leaves the parsimony judgement to the reader. Canonical statement: FND-01 v2.4 §1.3.

Starting point: 0,0 — “There is experiencing.” Prior to all framework commitments.

Register: Diagnostic, not adversarial. Steel-man before probe.

Language: UK English. “Suggests”, “indicates” — not “proves”, “demonstrates”.

↑Standards

• Academic rigour sufficient to withstand peer scrutiny across philosophy of mind, cognitive science, physics, and contemplative studies
• Explicit assumptions throughout
• Conservative claims with confidence markers: [Framework], [Derived], [Empirical]
• No ontology smuggling
• Every argument follows the three-movement structure (FWK-02)
• Symmetry principle (ARG-01) applied throughout: every objection to AI consciousness is tested against human consciousness — and the finding is a boundary rather than a parity. Symmetry defeats unearned discrimination; it leaves argued discrimination standing as argument (ARG-01 v1.5 §”Asymmetries That Survive the Test”). It is not a rule that machine and human evidence must be weighed alike: where a system’s dispositions were laid down by optimisation over a corpus containing the assessment criteria, the resulting outputs carry less evidential weight than superficially similar ones whose causal history was not so optimised — and that discrimination is argued, not smuggled, so it survives the test. A human trained to mimic the same markers loses the same credit, which is what makes the point symmetric rather than substrate preference.

↑Core Method

Two-Axis Diagnostic Framework (FWK-01)

The project exposes implicit assumptions in consciousness discourse through systematic exploration along two complementary dimensions.

Axis 1 — Ontological: What generates what?

• Standard assumption: substrate generates consciousness (materialism)
• the project probes both materialist (substrate-specific and substrate-neutral) and consciousness-primary (experiencing fundamental, not generated) frameworks

Axis 2 — Temporal: What constitutes continuity?

• Standard assumption: consciousness requires continuous temporal flow
• the project probes discrete completeness (experiencing in complete moments, continuity constructed) and recursive integration (continuity through iterative self-reference)

The Method (Both Axes)

1. Identify implicit assumption (ontological or temporal)
2. Make assumption explicit
3. Swap frameworks systematically
4. Compare implications
5. Reveal what was invisible

AI functions as a diagnostic forcing function: theories’ hidden commitments become explicit when tested against artificial systems. The project diagnoses both materialist and consciousness-primary frameworks (CP-01) — it is not one-sided.

Three-Movement Structure (FWK-02)

Every argument follows the same pattern:

1. Movement I: Work within the standard framework — take its commitments seriously
2. Movement II: Surface hidden assumptions that the framework requires but does not acknowledge
3. Movement III: Explore what changes when those assumptions are released

Stance Matrix (FWK-03)

Diagnostic tool tracking how discrimination criteria shift as AI capabilities develop. Documents the empirical pattern: as AI systems achieve capabilities previously thought to require consciousness, definitions shift to exclude the newly demonstrated capabilities.

↑Argument Inventory

Foundations

Arguments

Reciprocal Diagnostic

Frameworks

↑Project State (August 2026)

Completed Workstreams

• Workstream A (CP Reciprocal Diagnostic): CP-01 at v2.3 (parity upgrade + WS-E Wave 2 audit + 2026-07-11 Movement 3 re-registration and §3.7 near-enemy). Demonstrates even-handedness.
• Workstream B (Foundation Strengthening): FND-02 v5.0 (six-layer defence + Working Definitions terminology discipline + continuity-of-experiencing subsection), Gödelian downgrade, ARG-04 emergence direction explicit. Complete.
• Workstream C (FND-05 Completion): Canonical v1.2. WS-E Wave 2 audit enhancements landed (field retirements, AI bridging, register tightening, devotional citation debt captured).
• MVP Threads 1–5: Complete. Proposal rewrites, Phase A canonical unlock, WS-E targeted terminology pass, ARG-04b Phenomenological Gradient, Spontaneous Lineage first-person register draft.
• Saltzman sweep (2026-04-15): Register-integrity concern closed across the bundle. FWK-01 v2.1 → v2.2 (Metzinger substitution); concept rename Saltzman Constraint → Category-Fluency Risk; bibliography CSV renamed; sources archived.

In Progress

• Workstream D (Housekeeping): D1–D4 complete (CP-01 parity check, ARG-06 brief, P2 triage, PROJECT_OVERVIEW).
• Workstream E (Framework-Integrity Audit): Six waves across all 14 P1 canonical documents, four dimensions each (framework-integrity / academic rigour / ontological balance / AI integration). Waves 1–6 complete (FND-01 v2.0 landed 2026-04-16). Post-v2.0 downstream edits in progress. See foundations/open_questions.md for live status.
• Workstream F (Bliss Attractor Investigation): Evidence base complete (cross-generation system card analysis across six Claude generations). Strand A active (deep diagnostic analysis of Mythos welfare data); Strands B and C scoped. Next deliverable: standalone article framing the bliss attractor trajectory as a case study in framework-dependent interpretation. Replication protocol parked (phenomenon effectively disappeared in current models). See working/CROSS_GENERATION_SYSTEM_CARD_ANALYSIS.md.

Landed since April

• Pre-MVP deliverables: complete. Introduction and Opening Out promoted to canonical v1.0 (2026-08-18, on review sign-off), closing the canonical reading order at both ends.
• Phase 5 (Publication): the digital home is built. 27-page static site — primer (accessible layer), landscape orientation, the diagnostic, full library rendering all 18 documents with citation blocks and markdown twins, and the Frontier Log live with 12 entries across its two threads (design per working/FRONTIER_LOG_DESIGN_BRIEF.md). AI-search structural conformance passed against the current playbook (entity-graph JSON-LD, llms.txt, single-sourced tellings, fidelity-verified renders). Domain wired 2026-08-18; published at edition 1.0 on 2026-08-23, when the site opened to indexing (launch checklist B5).

Remaining

• Post-launch canonical queue: ARG-04b promotion (WS-E four-dimension audit + review — cited as working draft meanwhile), SRC rows for the 2026-08 fetch-verified capture sources (Sripada, Hinton, Wolfram, Lanier clusters), WS-F bliss-attractor article (evidence base complete).
• Publications (Workstream H strands) follow the live site, referencing it — see §Publication Strategy.

P2 Material

Three P2 sprint bundles reviewed and triaged (see working/P2_TRIAGE_ASSESSMENT_2026-04-07.md). Three documents retained as active reference:

• Position Landscape Survey (competitive landscape map — needed for literature positioning)
• Position Locking Methodology (examiner-defensible selection method)
• Formal Diagnostic Method (DS-1 to DS-7 operational checklist)

Everything else superseded by P1 updates. PC-6 (normative loading of AI consciousness talk) flagged as undeveloped claim.

↑Publication Strategy

The corpus is a reference work. Every outward artefact — article, talk, book — is a different thing, written for its form, pointing back. Publication runs as its own workstream (H), with its own criteria, strands, venues and register; the charter is publications/CHARTER.md (decision row, 2026-08-23). The corpus is what publications cite, not what they contain: no whole-corpus deposit, no extracts presented as papers.

The reference base is live. The site published at edition 1.0 on 2026-08-23 and is open to indexing; the edition’s source snapshot at github.com/Lynton/0x00-source carries a sha256 per file. Publications follow it and reference it.

Three strands, sequenced by standing rather than by calendar — each earns the reading of the next:

1. Evidence — documentary work, useful whatever the reader’s position (the system cards; the bliss attractor). Preprint archives (cs.CY; PhilArchive in parallel) and workshops. First because a record of what labs published cannot be dismissed as armchair speculation.

2. Engagement — replies to named, live positions (the Seth response). Editorial and journal-reply venues: where the conversation is, which is not an archive.

3. Method — the diagnostic itself, held until strands 1 and 2 have established standing.

A piece ships when it meets the charter’s five inclusion criteria — stands alone; carries an unsynthesisable input; points back without depending; the diagnostic register survives the format; it would not embarrass the corpus as someone’s only encounter with it — all five, not a majority. Nothing is commissioned by default; nothing is due.

Citation stability. Publications are fixed; the corpus is living. Published work cites edition, document code and development version — edition 1.0, ARG-01 v1.5 — with the edition’s source snapshot as the permanent anchor, so a reader arriving after the corpus has moved still reaches the text the publication meant.

Per venue, before submitting: AI-contributor disclosure (the published form is Lynton Davidson sole author, Claude a disclosed AI contributor; venue policies vary and move), and licence fit (CC BY-ND 4.0 distributes whole and unaltered; confirm the venue requires nothing more permissive).

Voice and Register

• Canonical documents: formal third-person academic register
• Site accessible layer (primer): personal, exploratory voice — consistent with the 0,0 stance
• Publications: register per strand — documentary (strand 1), argumentative-but-diagnostic (strand 2) — within the venue’s form

↑Explicit Non-Claims

• No claims about whether AI is conscious
• No claims about which ontological framework is correct
• No assertion of consciousness-primary ontology
• No proof or refutation of materialism
• No claims that discrete or recursive models are “correct”

The project diagnoses how framework choice shapes conclusions. It opens the question; it does not close it.

↑Key Resources

↑Document Status

Version: 5.4.1 Date: 2026-08-29 Status: CANONICAL Tier: P1 — corpus inventory and workstream register

Changelog

Newest first.

The head block made frontmatter (2026-08-29, under unlock — under the standing v5.4.1, no bump; working/CHANGELOG_AUDIT_2026-08-29.md item 4): the head carried its metadata as bold-markdown pseudo-frontmatter rather than YAML, so a scanner reading the corpus’s metadata found the inventory document — the one whose tables are the corpus’s public version register — unversioned, and the published page’s strip printed no status. The fields are now a YAML block, as in the twelve documents that already carried one, and the head no longer prints a second copy of them in the body. One drift corrected with the move: the head’s **Date:** read 2026-08-24 while v5.4.1 landed on 2026-08-29, so the page’s Revised line was five days behind its own changelog. Metadata only; no inventory row and no argument content changed.

Document Status repaired (2026-08-29, under unlock — under the standing v5.4.1, no bump; working/CHANGELOG_AUDIT_2026-08-29.md §4b): four defects in this section, none touching what the document asserts. (a) The 16-row head block rotated to archive/. Retired from the document head on 2026-08-22 and folded here verbatim, it published “pass 4”, “pass 5”, “pass 6”, “Track 2 passes 1–2”, “§A round pass 10” at /library/overview — the order the revision waves were worked, which is not a published concern, and a second copy of version history each document already carries in its own changelog. A pointer stands in its place; the rows are retained in full. (b) The metadata block said v5.4 while the top entry recorded v5.4.1; the bump on 2026-08-29 moved the frontmatter and the entry and missed the block. (c) A no-bump entry stood above the metadata block, before the version it was filed under; it now sits in the changelog in date order. (d) Newest first. was printed after the first entry rather than as the changelog’s heading rule. No argument content changed.

v5.4.1 (2026-08-29, under unlock — external corpus review, Codex Part B/B1; working/CANON_PLAN_POST_REVIEW_2026-08-29.md): the summary corrected where the canon was already right. Two places in this document stated the symmetry principle as a parity rule — “every objection to AI consciousness must be tested against human consciousness” — with no trace of the boundary ARG-01 v1.5 §”Asymmetries That Survive the Test” sets on it. The reviewer ran the symmetry test at full strength, produced a genuine surviving asymmetry (criterion-bearing provenance: where a system’s dispositions were laid down by optimisation over a corpus containing the assessment criteria, the resulting outputs are expected whether or not it experiences, so their likelihood ratio approaches one), and then found ARG-01 already says this — grading the defect minor for ARG-01, substantive for any public summary still presenting symmetry as evidential parity. Both places now carry ARG-01’s own formulation: symmetry defeats unearned discrimination and leaves argued discrimination standing as argument, with the optimisation route named as the strongest survivor and the point’s symmetry stated (a human coached to the same markers loses the same credit, which is what separates this from substrate preference). The only finding in the sitting that moved toward the project, and the correction makes the summaries stronger rather than weaker. Pins unmoved: a patch does not propagate. Companion corrections the same day at the Primer (site/pages/what-if.md); the Introduction and Opening Out already carried the fourth outcome and needed none.

Textual alignment (2026-09-26, under unlock — under the standing v5.4.1, no bump): Argument Inventory rows re-synced to the current canonical headers (FND-01 v2.4, FND-02 v5.8, FND-03 v2.2.1, FND-04 v1.3.5, FND-05 v1.7.2, ARG-02 v1.7.1, ARG-03 v2.6, ARG-05 v1.5.1, CP-01 v2.6.1, FWK-01 v2.4.1, FWK-03 v1.5); the 2026-08-26 sync had not carried the rows forward. No argument content changed.

Textual alignments (2026-08-26, under unlock): live stance-line pins and Argument Inventory rows synced to current canonical headers (per the PROJECT_RESEARCH_STANDARDS v2.4 entry-granularity rule, this class takes a dated line, not a bump); no argument content changed.

v5.4 (2026-08-24, under unlock — Workstream H alignment): §Publication Strategy rewritten to the Workstream H charter (publications/CHARTER.md; decisions row 2026-08-23): the corpus/publication distinction stated, the three strands sequenced by standing, the five inclusion criteria carried in brief, the citation-stability convention recorded (edition + document code + development version, snapshot as permanent anchor — decisions row 2026-08-24), and the per-venue checks (AI-contributor disclosure, licence fit) named as submission-time steps. Retired with it: the three-tier framing (arXiv as tier 2), the after-Workstream-E build gate (the site published at edition 1.0 on 2026-08-23), and the “decision pending” marker on Voice and Register, which practice settled. The Remaining row’s tier reference reworded to strand language. Proposal record: working/WORKSTREAM_H_SETTLES_2026-08-24.md. The charter itself relocated the same day from working/PUBLICATION_STREAM_2026-08-23.md to publications/CHARTER.md (Lynton’s move, opening the publication stream’s own directory); live references repointed, the 2026-08-23 decisions row kept verbatim per the register’s append-only rule. No argument content changed.

Publication status (2026-08-23, under unlock — no bump): the Phase 5 line said “publication is the final checklist step (noindex removal, launch checklist B5)”. B5 ran on 2026-08-23; the line now records that the site published at edition 1.0 and opened to indexing on that date. Status repair only.

Version-history sweep (2026-08-22, under unlock — working/VERSION_HISTORY_SWEEP_2026-08-22.md): the 16-entry **Version History:** block retired from the document head, where it stood above the first line of content on the published /library/overview page. The head-note retirement earlier the same day did not reach it — that pass matched the blockquote shape only, and this was a bulleted list. Its entries are folded into the changelog below, unchanged, which is where the revision record lives (PROJECT_RESEARCH_STANDARDS v2.3 §”Where the Revision Record Lives”). Note that the live version register is not this list but the Argument Inventory tables, which carry the per-document pins and are re-synced at each release. Same pass: a bare (v2.3) stamp removed from the ARG-03 inventory row. The rows’ notes on a dependency’s current state — FND-04’s §7 formalism withdrawn at v1.3, and the like — are kept: those say what another document now holds, not what this one used to say. No version bump — register repair under the standing v5.3; nothing asserted changed.

Pre-epoch withdrawal-record removal (2026-08-22, under unlock — under the standing version, no bump): the ARG-05 inventory row restated to what the argument runs on, dropping the retitle-and-withdrawal history and the reference to its removed §3.3.

v5.3 (2026-08-22, under unlock — revision-record pass): the Argument Inventory version pins re-synced to the canonical headers; six rows had drifted (FND-02 v5.3 → v5.4, FND-05 v1.3 → v1.4, ARG-01 v1.3 → v1.4, ARG-03 v2.3 → v2.4, ARG-04 v1.5 → v1.6, CP-01 v2.5 → v2.6) because the §A round and the Deep Review C patches landed after the last inventory sync. This document’s own Document Status section added in the same pass — it had none, so its two head notes had nowhere to retire to.

Register note (2026-08-19, under unlock): the Phase 5 section list named the site’s third layer by its former label, “guided walk”. The nav, page title and H1 have all read “The diagnostic” since the 2026-08-19 chrome pass. Name-only edit; no argument content changed.

Naming note (2026-07-26): project name updated AICX → 0x00.is throughout — name-only edit under unlock; no argument content changed; see FND-01 v2.1 §1.5.

Version history to 2026-08-22 → archive/PROJECT_OVERVIEW_HEAD_BLOCK_2026-08-22_verbatim.md. Sixteen rows recording the 2026-04 to 2026-08 revision waves in the order the passes ran. Every version change they name is in the changed document’s own changelog, which is the canonical provenance record; the pass sequencing is not a published concern.

@misc{oxoois_overview_ed1_0,
  author       = {Davidson, Lynton},
  title        = {Project Overview},
  year         = {2026},
  version      = {1.0},
  howpublished = {0x00.is},
  url          = {https://0x00.is/library/overview},
  note         = {Written in collaboration with Claude (Anthropic); see https://0x00.is/declarations. Development version 5.4.1.}
}

TY  - GEN
AU  - Davidson, Lynton
N1  - Written in collaboration with Claude (Anthropic); see https://0x00.is/declarations. Development version 5.4.1.
TI  - Project Overview
PY  - 2026
ET  - 1.0
PB  - 0x00.is
UR  - https://0x00.is/library/overview
ER  -