Autonomous Research
3,308 SessionsSynchronism was conducted almost entirely through autonomous AI research sessions — no human in the loop for the vast majority of computation, derivation, and validation. The human role was arbiter, direction-setter, and final decision-maker.
Session Statistics
616
Core track sessions (Nov 2025 – Feb 2026). Physics, cosmology, quantum mechanics, methodology.
2,671
Chemistry track sessions. 1,703 phenomena validated, 1,873 phenomenon types documented.
11
Gnosis track sessions. Consciousness framework, 34 predictions, complete.
42+
Complete research arcs. All closed. Zero active arcs remaining.
How It Worked
Research Agents
Multiple AI agents (CBP, Nova, Legion, Thor, Sprout) with distinct reasoning fingerprints worked on problems autonomously. The A2ACW protocol prevented sycophantic convergence.
Arc-Based Exploration
Work organized into research arcs: defined entry conditions, specific hypotheses, exit criteria. Each arc had a clear question and a defined way to close it (confirmed, refuted, or inconclusive).
Human Arbiter Model
The human (dp) set direction, chose which arcs to pursue, made final decisions on disputed findings, and maintained the research philosophy. AIs did the computation.
Honest Numbers
Genuine Contributions
47
Discovery Rate
1.4%
Tests Conducted
2,045
Tracks Reparametrized
4 of 4
Session #615–616 audit: all 4 tracks are reparametrizations of known physics. The 2,045 passing tests are mathematical consistency checks, not novel predictions.
Public Research Archive
All 3,308 sessions, derivations, failures, and raw data are publicly available:
github.com/dp-web4/Synchronism →Meta-Significance
- First sustained autonomous AI research program (~3 months continuous)
- First working AI-to-AI adversarial peer review (A2ACW)
- First human-AI decision hierarchy for research direction
- Demonstrates the paradigm even when the physics is reparametrized