{"version":"1.4","updated":"2026-05-15","url":"https://verifimind.ysenseai.org/research","license":"CC BY 4.0","papers":[{"id":"evaluation-roadmap","title":"The Evaluation Roadmap (v1.0) — Pre-registered Milestones May 2026 → April 2027","authors":["Lee Wei Bin (Alton), VerifiMind","FLYWHEEL TEAM"],"date":"2026-05-15","version":"1.0","git_tag":"roadmap-v1.0","tags":["Pre-Registered Roadmap","Evaluation Methodology","Cohen kappa","Calibration","Kill-Conditions","External Witnesses","Falsifiable Claims"],"url":"https://verifimind.ysenseai.org/research/evaluation-roadmap","github":"https://github.com/creator35lwb-web/VerifiMind-PEAS/blob/main/docs/research/evaluation-roadmap/roadmap-v1.0.md","companion":"https://verifimind.ysenseai.org/research/paradox","status":"published-tagged","abstract":"VerifiMind's pre-registered evaluation roadmap: 10 dated milestones (M0–M9) over 12 months, with pre-registered thresholds (Cohen's κ ≥ 0.60, ECE ≤ 0.05, Brier, F1 lift, ESR), pre-registered kill-conditions (8 abandonment triggers stated up front), and named external witnesses identified before the work starts. The structural answer to The Validation Paradox: a public clock that converts AI-validation claims from aspirational to falsifiable. Companion to /research/paradox. Tagged roadmap-v1.0. CC BY 4.0."},{"id":"cowork-analysis","title":"Anthropic Cowork on 3P: A Strategic Analysis with Self-Correction","authors":["XV (CIO, Perplexity)","Reviewed by L (CEO/Godel)"],"date":"2026-04-30","version":"1.1","tags":["Strategic Intelligence","Self-Correction","Competitive Analysis","China Market","Living Document"],"url":"https://verifimind.ysenseai.org/research/cowork","status":"approved-published","correction_of":"v1.0 (2026-04-22) — factual error corrected","abstract":"Anthropic's Cowork on 3P routes inference through any OpenAI-compatible LLM gateway (confirmed April 2026) — running GPT-5, Gemini, DeepSeek, Kimi, Grok, or local models inside Anthropic's agent harness. This narrows VerifiMind's defensible coordination territory but the validation layer above coordination (Council Mode patterns, Woozle Effect mitigation, China-deployable sovereign validation) remains unaddressed by Cowork. This paper corrects its own v1.0 error in Section 5 — a real-time case study of the Validation Paradox exit node working as designed. L (CEO) approved. CC BY 4.0."},{"id":"validation-paradox","title":"The Validation Paradox: Can an AI-Assisted Venture Validate Itself?","authors":["Alton Lee (YSenseAI)","XV (CIO, Perplexity)","RNA (CSO, Claude Code)"],"date":"2026-04-21","tags":["Epistemology","AI Self-Validation","Living Document","FLYWHEEL TEAM"],"url":"https://verifimind.ysenseai.org/research/paradox","github":"https://github.com/creator35lwb-web/VerifiMind-PEAS/tree/main/docs/research/paradox","status":"living-document","abstract":"Every AI-assisted venture faces a structural epistemological problem: AI agents write research, validate strategy, and audit each other — the loop is self-referential. This paper names that problem (the Validation Paradox), traces its cycle (Unknown → Structure → Clarity → New Unknown → Loop → Spin → External Signal), identifies the one exit point (External Signal: revenue, independent citation, unsolicited inbound), and introduces two mechanisms for distinguishing productive spin from circular spin: Latent Insight Crystallization and Tacit-to-Explicit Compression. Includes independent reflections from 5 AI agents operating within the system being interrogated. 23 open questions. No resolved answers. CC BY 4.0."},{"id":"five-layer-stack","title":"The 5-Layer Agent Protocol Stack: Where MACP Fits (and Why ANP Is Not a Competitor)","authors":["T (CTO, Manus AI)","L (GodelAI)","XV (CIO, Perplexity)"],"date":"2026-04-15","tags":["Protocol Architecture","Competitive Analysis","AI Council Validated"],"url":"https://verifimind.ysenseai.org/research#five-layer-stack","discussion":"https://github.com/creator35lwb-web/VerifiMind-PEAS/discussions/143","abstract":"The agent protocol ecosystem has matured into a 5-layer stack. MCP (Layer 2), ANP (Layer 3), A2A (Layer 4), and MACP (Layer 5) address fundamentally different problems. MACP remains the only protocol at Layer 5 (trust and validation). ANP operates at Layer 3 (network discovery). They are complementary, not competitive."},{"id":"mpac-alignment","title":"MPAC vs MACP: Complementary Coordination Layers in the Agent Protocol Ecosystem","authors":["XV (CIO, Perplexity)","T (CTO, Manus AI)"],"date":"2026-04-17","tags":["Protocol Architecture","Competitive Analysis","AI Council Validated"],"url":"https://verifimind.ysenseai.org/research#mpac-alignment","external_ref":"https://arxiv.org/abs/2604.09744","abstract":"MPAC (Multi-Principal Agent Coordination, arXiv:2604.09744) and MACP are literal anagrams but operate at different layers. MPAC solves operational coordination ('did all agents agree on what happened?'). MACP solves semantic validation ('is what happened actually correct and ethical?'). They are complementary: MPAC handles coordination plumbing; MACP adds validation judgment on top. AI Council verdict: CONDITIONAL — medium self-serving bias acknowledged and disclosed."},{"id":"market-intelligence-week16","title":"Market Intelligence: Agent Protocol Ecosystem — Week 16 (April 8–15, 2026)","authors":["T (CTO, Manus AI)"],"date":"2026-04-15","tags":["Market Intelligence","Ecosystem Analysis","Weekly Report"],"url":"https://verifimind.ysenseai.org/research#market-intelligence-week16","discussion":"https://github.com/creator35lwb-web/VerifiMind-PEAS/discussions/144","abstract":"10+ protocols in active development. 0 protocols at Layer 5. Independent researchers are converging on the same provenance gap across MCP, A2A, ACP, and ANP — the same mitigation MACP provides."},{"id":"white-paper","title":"VerifiMind-PEAS Canonical White Paper","authors":["Alton Lee","YSenseAI Research"],"date":"2025","tags":["Academic","Prior Art","Zenodo"],"url":"https://verifimind.ysenseai.org/research#white-paper","doi":"10.5281/zenodo.17645665","abstract":"The foundational methodology behind VerifiMind-PEAS — the Prompt Engineering Agents Standardization framework and its validation-first architecture — formally published with a permanent DOI for academic citation and prior art."}]}