Datasets:
Day 3: CPS-0001 — The Continuity Receipt Protocol
The narrative so far:
Day 1: We asked what happens when identity stops meaning presence. AI can generate any static credential — face, voice, document — but presence is a trajectory, not a snapshot. The "seeing is believing" era is ending.
Day 2: We argued the entire industry is solving the wrong problem. Every authentication system asks "who are you?" Almost none asks "are you continuously there?" Identity is a snapshot. Continuity is a trajectory.
Day 3: We release the protocol.
CPS-0001 is an engine-independent format for Continuity Receipts — self-verifying cryptographic objects that prove continuous presence over time.
Key properties:
• Engine-independent — any sensor or algorithm can produce receipts in the same format
• Ed25519 signed — cryptographic authenticity
• SHA-256 chained — each receipt links to its predecessor, forming an auditable chain
• V₁–V₇ verification contract — 7-step verification from signature to subject consistency
The Forgery Cost Framework:
Instead of searching for a single "unfakeable" signal, we measure the increasing cost of maintaining consistent forged evidence across channels:
— Single frame: 1×
— + Temporal consistency: 5×
— + Cross-modal (IMU + camera): 25×
— + Live challenge: 100×
Evidence: 576 runs across 4 engines
— EE-001 (PES): 4-dim biological noise, Cohen's d=2.1
— EE-002 (Cross-modal): 100% temporal alignment, 316 trials
— EE-003 (Challenge-response): 60% pass, 200 trials
— VS-001 (Pipeline): 93% pass, 60 sessions
arXiv paper submitted: "Continuity Proofs: An Evidence-Based Framework for Verifying Persistent Entity Presence"
All open source (Apache 2.0). Public benchmark dataset. Documented limitations alongside results.
We don't claim to have solved continuity. We claim it's the right question — and that answering it may become one of the defining cryptographic challenges of the AI era.
github.com/myshapeprotocol/myshape-protocol
thecontinuitylab.org