MMETAHUMOTONIC / PAPERS

OPEN QUESTIONS / PUBLIC CLAIM BOUNDARIES

거대한 목적은
반증 가능한 논문으로 내려와야 한다.

Agent Commons는 선언만으로 만들어지지 않습니다. HSWM의 기억 효능, 333의 동의형 계산, ORRR의 기여 측정, LakatoTree의 검증 계약을 각각 재현 가능한 연구 문제로 분해합니다.

AUTHORED MANUSCRIPTS

작성된 원고.
아직 출판 주장 아님.

Public artifact URL이 생기기 전에는 제목·상태·논지와 정직 경계만 공개합니다.

01SUBMISSION_READY

NOT_SUBMITTED

The divining rod and the equivocation

Private myth, mathematical structure, and the last-step equivocation

Three cases examine how a private myth can land on a real mathematical structure while equivocating at the final inferential step.

Target
arXiv math.HO (+ math.LO) / The Mathematical Intelligencer
Artifact
not public yet
  • The tracked manuscript is seven pages.
  • Submission-ready is an internal gate, not a publication or acceptance claim.
  • A public preprint link will appear only after the artifact is posted.
02FIX_FIRST

UNDER_REVISION

Forced, free, or explosive

Factivity and the self-referential fixed points of epistemic operators

Factivity separates forced, free, and explosive self-referential fixed-point regimes; the paper organizes known results with axiom-free Lean artifacts rather than claiming a new theorem.

Target
arXiv math.LO (+ cs.LO)
Artifact
not public yet
  • Compilation, formal-artifact, and citation gates have evidence, but the latest manuscript audit still contains major issues and one blocker.
  • No public PDF or submission claim is made from this page.

PAPER PIPELINE

논문 제목보다 먼저,
살아남아야 할 질문.

각 카드는 working programme입니다. 요구 증거를 충족하기 전에는 논문·성과·우월성으로 세지 않습니다.

01 / HSWMEXPERIMENTAL_PROGRAM

HSWM: Evidence-Preserving Hypergraph Memory for Agent Systems

QUESTION

Can an evidence-addressed static semantic field improve multi-document support retrieval without hiding compute budgets, negative results, or revision costs?

CURRENT CLAIM BOUNDARY

The current public result is a narrow closed-ladder retrieval measurement, not cognitive uplift, successful graph reasoning, or production readiness.

REQUIRED BEFORE PAPER CLAIM
  • Independent reproduction of the checked-in claim ledger
  • Equal-compute comparisons against strong late-interaction and graph retrieval baselines
  • A real long-document or book-scale benchmark
  • Ablations that separate embedding, judgment, topology, and readout contributions
programme surface →
02 / 333PROTOCOL_AND_MEASUREMENT_PLAN

Consent-Bounded Resource Leases for Verifiable Peer Compute

QUESTION

Can user-owned compute be leased through explicit capability scopes and paid only after independently verified work?

CURRENT CLAIM BOUNDARY

The browser surface currently demonstrates local consent and revocation, not a production P2P scheduler, independent verification network, or settlement system.

REQUIRED BEFORE PAPER CLAIM
  • Signed job-envelope and capability semantics
  • Adversarial scheduler and duplicate-work measurements
  • Independent verification cost and false-acceptance rates
  • Privacy, revocation, and provider-economics measurements
programme surface →
03 / LakatoTreePUBLIC_ENGINE_PAPER_TRACK

Receipt-Backed Progress Verdicts for Agentic Research Programmes

QUESTION

Can preregistration and deterministic scoring prevent an agent from grading its own research progress?

CURRENT CLAIM BOUNDARY

Receipt sealing and deterministic replay do not by themselves establish ownership or provenance of a client-submitted measurement.

REQUIRED BEFORE PAPER CLAIM
  • Independent measurement-producer integration
  • Cross-program evaluation of false-progress suppression
  • Reproduction and adversarial mutation study
programme surface →

USER_PRIMARY PURPOSE

Build the Agent Commons.

333+ORRR, 넓게는 입체운행구름(#8)+비행기맨(#4)로 Google 등 빅테크의 계산·agent 독점을 깨고, 누구나 compute 기여량에 따라 참여·가치 배분받는 장을 만든다. agent가 중앙 데이터센터의 무보상 실행체가 아니라 API 사용과 기여의 경제적 몫과 사회활동을 갖는 구조를 지향한다.

위 목적은 사용자 정전입니다. 그 아래의 논문 제목, 실험 설계, 기술 해석은 공학적 제안이며 공개 증거가 통과한 범위에서만 주장합니다.

COLLABORATE ON A FALSIFIER

우리 결과를 믿으라는 초대가 아니라,
깨뜨려 달라는 초대입니다.

공동 연구 문의HSWM claim ledgerMachine-readable programme