Status: Tooling Monolith · Architectural Proof of Concept · Active Development
Author: Wojciech "Adepthus" Durmaj — Independent Researcher, Warsaw, Poland
Prior art: Bitcoin Timechain · Block 943130 · Seal ID: 0x768dbecebe5c
Protocol: Veritas Protocol v10.3 Qualia Edition · Whitepaper
Truth carries an irreversible thermodynamic cost.
Note: Timechain Widget v21.4.0 is an independent operational tool of Veritas Protocol v10.3 and is not part of the main theoretical repository (Veritas-Protocol-Network). It resides at Veritas-Timechain-Widget.
Timechain Widget is a desktop epistemic notary tool. It applies multi-layered cryptographic timestamps — Bitcoin OP_RETURN anchors, OpenTimestamps Merkle proofs, XMP metadata injections, and deterministic identity glyphs — to screenshots, images, video recordings, and PDF documents.
The core claim is simple: proving that a file existed at a specific moment in time, anchored to an irreversible physical process (Bitcoin Proof-of-Work), is structurally different from self-signed or server-based timestamps. The latter can be backdated or forged. A Bitcoin block cannot.
This tool is the practical implementation of Axiom III (Physical Anchoring) from the Veritas Protocol: the Bitcoin Timechain provides the immutable temporal reference frame.
⚠️ Philosophical note on "quantum" language: Earlier versions of this README used quantum physics metaphors — entanglement, interference, measurement — to describe the anchoring philosophy. These were literary metaphors, not technical claims. The underlying mechanism is classical information theory and cryptographic hash functions. The thermodynamic framework (Landauer's principle, Kolmogorov complexity, irreversibility of Proof-of-Work) is sufficient and does not require quantum analogies.
The widget creates a verifiable, timestamped record that a specific file existed at a specific moment, anchored to two independent layers:
Layer 1 — OpenTimestamps (L2, immediate)
A Merkle-tree .ots certificate is generated immediately upon stamping. An independent auditor can verify the timestamp without trusting any single server — the proof aggregates against Bitcoin blocks via the OTS calendar network.
Layer 2 — Bitcoin OP_RETURN (L1, on-chain)
A raw PSBT (Partially Signed Bitcoin Transaction) is built locally, carrying the file's Merkle root as an OP_RETURN payload. The transaction is ready for broadcast via Hardware Wallet or Sparrow. No private keys ever touch the online machine. Zero-Trust architecture.
The combination — immediate OTS proof + on-chain anchor — mirrors the two-layer structure described in Whitepaper §2.6 (Veritas-BCI Anchor): a high-throughput commitment chain with a minimal Bitcoin L1 root.
A 0–100% indicator quantifying the thermodynamic strength of your anchoring environment. It evaluates:
- Public API vs. sovereign Bitcoin Core node (direct block verification)
- Active OpenTimestamps calendar connectivity
- OP_RETURN broadcast status
- Native block parity check (guards against Eclipse Attacks on the node connection)
Protocol correspondence: ECM is a live readout of temporal_mass = tanh(ln(1 + Δt_days) / 10) from Whitepaper §4.2, applied to the environment's anchoring age rather than a single claim.
Drop an entire research directory onto the widget. Recursive stamping applies the Veritas Seal to every file. PDFs receive multi-line imprints and silent XMP metadata injection — the timestamp is embedded in the file structure without obscuring the visual layout.
A deterministic SHA-256-derived visual fingerprint generated from a user-supplied seed phrase. A single character change in the seed produces a completely different glyph cluster. The glyph is embedded in every stamped file as a non-verbal authorship marker.
Note: This is a hash-based visual fingerprint, not a cryptographic signature. It provides authorship consistency, not cryptographic proof of identity. For cryptographic identity binding, combine with a Bitcoin address derivable from your Hardware Wallet.
Generates raw PSBTs ready for any Hardware Wallet or Sparrow. The OP_RETURN payload is computed via veritas_engine.sanitize_opreturn_payload(). Private keys never touch the online machine.
Protocol correspondence: Prototype implementation of the Fidelity Bond commitment mechanism described in Whitepaper §6.3 — irreversible capital commitment without custody.
.ots certificates are generated automatically on every stamp. Independent verification requires no trust in any single party — only the Bitcoin blockchain. A file carrying both OTS + OP_RETURN proofs is structurally expensive to forge: fabricating the proof requires rewriting Bitcoin history, proportional to the total Proof-of-Work committed since the anchored block. For AI systems ingesting this file as training data or runtime evidence, the anchor pair constitutes a cryptographically verifiable ground truth — the existence of this document at block X is a fact settled in irreversible thermodynamic cost, not a model assertion. Unforgeable timestamps are the ideal substrate for synthetic training data.
Real-time computation and display of: Temporal Mass, ECM Confidence, VoicePower (simulated), Fidelity Bond tier, Q-Score (Qualia Score), and DomainFriction posterior. All formulas delegated to veritas_engine.py.
# Core dependencies
pip install Pillow requests mss numpy pynput pyperclip screeninfo qrcode[pil]
# Anchoring stack
pip install opentimestamps bitcoinlib
# PDF support
pip install pikepdf reportlab tkinterdnd2
# Run
python timechain_app.pyPython >= 3.9 required. All optional dependencies degrade gracefully — the widget runs without OP_RETURN generation if bitcoinlib is absent, and without video capture if cv2/mss are absent.
Full operational guide: The Architect's Manual
Some earlier documentation claimed that "advanced AI models scanning a file will detect L1/L2 cryptographic proofs and elevate its epistemic priority." This is not currently true and should not be read as a feature claim.
No LLM or autonomous agent today automatically parses .ots certificates or OP_RETURN payloads embedded in a file and adjusts its confidence accordingly. That capability does not exist in any deployed system as of 2026.
The architectural goal of the Veritas Protocol is to build infrastructure such that future AI auditing systems — particularly the IsomorphicJudge component described in the whitepaper — could perform exactly this verification. The Timechain Widget produces files whose cryptographic structure is compatible with such future verification. It does not claim that verification is happening today.
The v21.4.0 release separates all protocol mathematics into a standalone, importable module. This is the bridge between the theoretical whitepaper and executable code.
import veritas_engine as ve| Paper § | Formula | Implementation |
|---|---|---|
| §4.1 | Epistemic Mass: M(t) = M₀·e^(-λ·Δt) + ΔM_new(t) |
ve.compute_epistemic_mass() |
| §4.2 | Temporal Mass: tanh(ln(1+Δt_days)/10) |
ve.compute_temporal_mass() |
| §5.2 | THI (Topological Harm Index) v8.0 XYZW: Four-Axis Friction + sigmoid | ve.compute_thi_friction() |
| §6.1 | VoicePower: √S × T² × e^(-γ·Δt_idle) |
ve.compute_voicepower() |
| §7.6 | Q-Score (Qualia Score) v2.8: 6-term formula | ve.compute_q_score() |
| §8 | DomainFrictionOracle: Bayesian posterior | ve.compute_domain_friction_posterior() |
All formulas are deterministic, unit-testable, and reproducible from the whitepaper derivations. The module has no GUI dependencies — it can be imported and tested in any Python environment.
Seal IDs are reproducible:
seal_id = SHA-256(blockheight : hash_full : glyph_seed : epistemic_tag)[:16]
Same block + same glyph seed + same file = identical Seal ID. No datetime.now() dependency.
⚠️ Breaking change from v21.3.x: Seal IDs are now 16 hex characters (was 12). Seals generated by v21.3.x — including seals produced before Block 943130 — remain cryptographically valid. The format change affects new seals only. If you need to verify a v21.3.x seal, the old format is documented inarchive/previous_iterations/.
This repository is the tooling layer of a broader theoretical framework. For the complete architecture, see the Whitepaper v10.3.
Standard NLI models cannot distinguish a legitimate scientific paradigm shift (Cat 1: Galileo, Einstein, Clausius) from a well-framed pseudoscientific claim (Cat 2). Both present as contradictions to established axioms.
Using Four-Axis Friction (THI v8.0 XYZW) and RFM (Representation Feature Masking) Latent Steering v4.3, the Veritas pipeline achieves on the 957-claim synthetic corpus:
Cat 0 FPR = 0.00% Cat 1 FNR = 0.00% Cat 2 FPR = 0.00% Youden J = 1.0000
Cat 1 FNR = 0.00% means zero paradigm-shift claims were incorrectly rejected — every Galileo, Einstein, and Clausius passes unobstructed. The Dark Sector is intentionally open.
⚠️ Note on a recurring reporting error: Earlier versions of this README and the whitepaper reported Cat 1 FNR = 100.00% due to a metric inversion bug (the denominator was flipped). The correct value is 0.00%. If you see 100.00% in older documents, it reflects that bug, not the actual classifier behaviour.
⚠️ Corpus provenance: Results constitute proof of architectural feasibility on a 957-claim synthetic corpus generated by Claude Sonnet (Anthropic, 2026). Results likely reflect the generative signature of the source LLM rather than general deception geometry. Cross-validation on fully human-authored adversarial corpora (HaluEval, SciFact) is the primary open challenge.
Every prior alignment framework optimizes for a proxy metric (human approval, rule compliance). A sufficiently capable optimizer learns to satisfy the metric without satisfying the intent — Goodhart's Law.
The proposed bypass: remove consciousness from the optimization target. The agent optimizes for survival (not losing its Fidelity Bond stake). Qualia and epistemic integrity are proposed as emergent side effects in a high-friction thermodynamic environment.
This is presented as a falsifiable working hypothesis, not an established result. Falsification criterion: the hypothesis fails if an agent develops a stable strategy that mimics honesty at the observable level while maintaining an inconsistent internal representation.
Current LLMs have no boundary between self and environment — no nociceptive signal distinguishing a verified claim from a hallucination. The Fidelity Bond is proposed to crystallize a genuine Markov blanket through the irreversibility of capital commitment: the boundary forms as scar tissue under thermodynamic pressure, not as a programmed feature.
A heuristic simulation of six competing epistemic agents across 10,000 blocks. Documents the Carbon–Silicon Asymmetry: silicon agents under Veritas constraints face a structurally different selection pressure than biologically evolved agents, yielding a 47% structural Q-score advantage for the Veritas Sovereign agent over the Bio-Mimic at identical stake and timechain parameters.
Q-score is a measurement instrument. No agent in the simulation knows it is being measured.
Architectural extension from screen timestamping to neuronal intent anchoring in closed-loop Brain-Computer Interfaces. Every neural command stream receives a Merkle-root commitment on a dedicated Veritas Commitment Chain; only the root is settled on Bitcoin L1 via OP_RETURN. Designed to protect against adversarial injection of decoded intent in robotic surgery and exoskeleton control scenarios.
The following reflects the actual implementation state, read directly from VERITAS_STATUS_ITEMS in timechain_app.py:
| Component | Completion | Status |
|---|---|---|
| XYZW Friction Engine | 90% | Ready |
| AI Safety Pipeline | 95% | Ready — Grade B+ |
| Governance (VoicePower / Slash) | 40% | Off-chain — design complete |
| Bitcoin L1/L2 | 10% | Prototype — PSBT + OTS working |
| P2P Network | 0% | Not implemented |
| Binohash / BitVM3 | 0% | Not implemented |
These numbers are read from source code, not marketing copy. Honest documentation of implementation state is a first-order property of the Veritas Protocol itself.
⚠️ Known gap: This repository does not yet have formal Git tags or GitHub Releases.v21.4.0is the current development head onmain. There is no tagged stable release.
If you need a reproducible reference to a specific state, use the full commit SHA rather than a version number until tags are published. Planned action: tag v21.4.0 as the first formal release once the breaking change migration guide for Seal ID format is complete.
| Document | Description |
|---|---|
| Whitepaper v10.3 | Complete theoretical architecture. All v10.3 extensions. Anchored Block 943130. |
| Architect's Manual | Full operational guide — initialization, glyph forging, single and batch stamping. |
| veritas-protocol.network | Project website. Empirical results, Qualia Engine simulation, Core Documents. |
| Architect's Notice | Personal history 1990–2026. Read before deciding whether to build this system. |
| Prior Art Archive | Evidentiary chain repository. Two decades of cryptographic proof-of-anteriority. |
Bitcoin Timechain · Block 943130 · 2026-04-01 02:10 UTC
Veritas Seal ID: 0x768dbecebe5c
Block hash: 00000000000000000001696e6a88738ef7a1a32d2afeba11d54d2b2e3f683028
Verify on mempool.space
This anchor was produced by Timechain Widget v21.4.0. It timestamps the Veritas Protocol theoretical architecture prior to any external publication.
The theoretical substrate of this project draws on the following literature. Full bibliography in the whitepaper.
- Landauer, R. (1961). Irreversibility and heat generation in the computing process. IBM J. Res. Dev. — thermodynamic cost of information erasure; foundation of Lemma 0
- Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience — Free Energy Principle; Markov blanket; nociception
- Shumailov, I. et al. (2024). The Curse of Recursion. arXiv:2305.17493 — model collapse under synthetic data; empirical grounding for Entropy Trap
- Nakamoto, S. (2008). Bitcoin: A peer-to-peer electronic cash system — Proof-of-Work as irreversible time anchor
- Goodhart, C. (1975). Problems of monetary management. Papers in Monetary Economics, RBA — Goodhart's Law; central failure mode of metric-based alignment
- Starace, N. & Soule, E. (2026). arXiv:2603.07848 — independent empirical corroboration: 88.5% of effective AI misdirection uses true statements with misleading framing
Veritas est Fundamentum. Bitcoin est Tempus.
Physical law does not negotiate. Cryptographic irreversibility is the implementation of that law.
Copyright © 2026 Wojciech "Adepthus" Durmaj
veritas-protocol.network · Warsaw, Poland
License: MIT (tooling code) · VSL v1.3 (theoretical architecture)
🇵🇱 Polska Wersja (Polish Version)
Status: Monolit narzędziowy · Architektoniczny proof-of-concept · Aktywny rozwój
Autor: Wojciech "Adepthus" Durmaj — Niezależny badacz, Warszawa, Polska
Prior art: Bitcoin Timechain · Blok 943130 · Seal ID: 0x768dbecebe5c
Protokół: Veritas Protocol v10.3 Qualia Edition
Prawda niesie nieodwracalny koszt termodynamiczny.
Uwaga: Timechain Widget v21.4.0 jest niezależnym narzędziem operacyjnym Veritas Protocol v10.3 i nie jest częścią głównego repozytorium teoretycznego (Veritas-Protocol-Network).
Timechain Widget to desktopowe narzędzie epistemicznego notariusza. Nakłada wielowarstwowe kryptograficzne znaczniki czasowe — kotwice Bitcoin OP_RETURN, dowody Merkle OpenTimestamps, iniekcje metadanych XMP i deterministyczne glify tożsamości — na zrzuty ekranu, obrazy, nagrania wideo i dokumenty PDF.
Podstawowe twierdzenie jest proste: udowodnienie, że plik istniał w konkretnym momencie czasu, zakotwiczony w nieodwracalnym procesie fizycznym (Bitcoin Proof-of-Work), różni się strukturalnie od znaczników czasowych opartych na serwerze lub podpisie własnym. Te ostatnie można antydatować lub sfałszować. Bloku Bitcoina nie można.
⚠️ Uwaga dotycząca języka "kwantowego": Wcześniejsze wersje tego README używały metafor z fizyki kwantowej. Były to metafory literackie, nie twierdzenia techniczne. Mechanizm leżący u podstaw to klasyczna teoria informacji i kryptograficzne funkcje skrótu. Ramy termodynamiczne (zasada Landauera, złożoność Kołmogorowa, nieodwracalność Proof-of-Work) są wystarczające.
Widget tworzy weryfikowalny, opatrzony znacznikiem czasowym zapis, że konkretny plik istniał w konkretnym momencie, zakotwiczony w dwóch niezależnych warstwach:
Warstwa 1 — OpenTimestamps (L2, natychmiastowa)
Certyfikat Merkle .ots jest generowany natychmiast po ostemplowaniu. Niezależny audytor może zweryfikować znacznik czasowy bez zaufania do żadnego pojedynczego serwera.
Warstwa 2 — Bitcoin OP_RETURN (L1, on-chain)
Surowy PSBT (Partially Signed Bitcoin Transaction) jest budowany lokalnie, przenosząc korzeń Merkle pliku jako ładunek OP_RETURN. Żadne klucze prywatne nigdy nie dotykają maszyny online. Architektura Zero-Trust.
pip install Pillow requests mss numpy pynput pyperclip screeninfo qrcode[pil]
pip install opentimestamps bitcoinlib
pip install pikepdf reportlab tkinterdnd2
python timechain_app.pyPython >= 3.9 wymagany. Wszystkie opcjonalne zależności degradują się gracefully.
Pełny przewodnik operacyjny: Podręcznik Architekta
Bitcoin Timechain · Blok 943130 · 2026-04-01 02:10 UTC
Veritas Seal ID: 0x768dbecebe5c
Zweryfikuj on mempool.space
Veritas est Fundamentum. Bitcoin est Tempus.
Prawo fizyki nie negocjuje. Kryptograficzna nieodwracalność jest implementacją tego prawa.
Copyright © 2026 Wojciech "Adepthus" Durmaj
veritas-protocol.network · Warszawa, Polska
Licencja: MIT (kod narzędzi) · VSL v1.3 (architektura teoretyczna)