Provenance graph + plural verdicts + cascading falsification + AI-read surface is defensible, unbuilt infrastructure. That bet holds.
§ 02Blockchain + POW + POS tokenomics is not the right mechanism. Zero successful blockchain fact-check products in 8 years of attempts.
§ 03Federation on mature standards — W3C VC/DID + IETF SCITT + Sigstore/CT + libp2p — delivers every property at roughly 1/10× the cost.
§ 04Recommendation: ship Phase 1 now (upgrade existing `/factcheck` spec); Phase 2 conditional; Phase 3 (blockchain) almost certainly no.
Add critique, landscape, solutions.
→02 · deck-critique.mdThe deck bundles two things: what the web needs (the bet) and how to build it (the mechanism). Most critiques attack the mechanism while the bet goes unexamined. The three artifacts in Part I do the opposite: the bet gets the hardest look.
Sham consensus domain · oracle-work sybil + collusion · adversarial AI-generated provenance · retraction-attack cascade · certificate-spoofing.
Validator labour cost exceeds reward · AI labs do not integrate · user tools are shelfware · foundation funding is capped.
Provenance ≠ truth · claim atomisation ill-defined · domain-relative collapses into relativism · validator disagreement · model collapse of signals.
Consortium capture · jurisdictional fragmentation · dispute resolution becomes a parallel legal system · state-narrative domains.
Grounding latency · provenance graph stitching · client subscription explosion · key management · spec/protocol forks.
The deck commits to one mechanism (blockchain). The bet does not require it. Part II examines four implementations on shared axes — cost, trust model, standards reuse, adoption path, regulatory exposure — and recommends one.
Self-declared `/factcheck.json`. ClaimReview-extended. Versioned in git. No central infrastructure. Covers provenance; partial on AI-read; nothing on plural verdicts or cascades.
A v0.2 closes five gaps without breaking simplicity: claim-hash, verdictsByDomain, thirdPartyAttestations[], event feed, reference aggregator.
→11-impl-A-minimalist-factcheck.mdEvery primitive Veritas needs is a shipping standard in 2026:
The deck's raw C is incoherent. "Verification as PoW" fails all three PoW properties. PoS for signing duplicates institutional reputation. On-chain PII conflicts with GDPR Art 17. Institutional validator ROI on token rewards is -99.5%.
A corrected C uses optimistic-oracle + non-transferable governance token + fiat-bonded PoS + hashes-only-on-chain. 80% of its components overlap B.
Aggregate through end of Phase 2: ~$600–900K and 18 months. Exits are cheap because the protocol is a standard, not an operating platform.
→14-impl-D-staged-evolution.md · the full phased planFour layers of existing work (provenance, claim-review, decentralized identity, crypto-incentivised verification) each solve a slice. None solve provenance-DAG + plural verdicts + cascade together. Two catalogs — the working systems and the graveyard — feed the design.
Two catalogs, each with honest ratings. The hard, load-bearing parts of Veritas are classical CS — truth-maintenance systems, argumentation theory, epistemic logic, CRDTs, reputation math. Web3 contributes IPLD, libp2p, BLS — not epistemology.
Recommendation: drop PoW framing entirely. Prefer no-token federation (CT + Sigstore pattern). If political constraints force a crypto-native path: fiat-bonded PoS + governance-only non-transferable token. Explicitly rule out staking rewards.
→53-tokenomics-hard-analysis.md · 19 pp, quantitative analysisThe strengths concentrate at the composition level. The weaknesses concentrate in governance and economics — which is typical of infrastructure that touches institutions. Every weakness has a concrete mitigation; none require new inventions.
The answer, defended across 29 artifacts and five independent multi-agent analyses, is yes to Phase 1, conditional on Phase 2, no to Phase 3. What follows is the honest accounting.
Or, if deferring: publish this dossier as read-only reference on homototus/veritas wiki. Re-read in 6 months. Either is defensible. Both beat "keep thinking about it."
You asked for hybrid critique — radical + constructive. The radical part: the deck's blockchain framing is wrong. The constructive part: the bet underneath the deck is right, and option D delivers it.