🧠 Systems & Intelligence¶
The Core Questions¶
TL;DR
This repository now begins with a foundational audit and then studies two bounded questions:
- Model identification: How can an observer move from traces to useful candidate process models — through construction, world-coupling, intervention, and revision?
- Viability arc: How can an optimizing system remain viable when growing capability loads several constraints at once?
The Foundations Reconstruction establishes the common process language and shows where extra assumptions enter. The first research question is folded into From Trace to World-Binding. The second is mapped in Canonical Path v2.
Epistemic status — read this first
This is a research notebook, not a theory of everything. The process foundation is established mathematics rather than a novel empirical theory. Identity remains test-relative; learning and intelligence require declared tasks; phenomenal consciousness is not derived. The inverse-reconstruction results are measured in small, controlled systems. The Viable Corridor necessity result is conditional on its model assumptions; sufficiency is conjectured, not proved; the civilizational mapping is heuristic, not measured; and the AI-specific predictions have not been tested on real agent ecologies. The explicit negative space is maintained at What This Project Does NOT Claim.
The measured core is the inverse-reconstruction benchmark: known-family inversion, equivalence classes, intervention, family search, model exploitation, marked uncertainty, closed-loop revision, ensembles, and process composition. The Viability Arc is currently supported by formal results and two synthetic models; external calibration and real-agent tests remain open.
Navigate the Research¶
Recommended Reading Path¶
New to this project? Start here for the most direct path through the core theory and proofs.
- Foundations Reconstruction — Minimal primitives, axioms, derivations, counterexamples, and comparison with neighboring theories.
- From Trace to World-Binding — The bounded model-identification loop.
- Inverse-Reconstruction Benchmark — The measured equivalence classes and intervention results.
- Emergence Manifesto — The earlier emergence claim set, read under the reconstructed foundation.
- Optimization and Its Blindness — The hinge into the viability arc.
- The Viable Corridor — One conditional formal model.
- Canonical Path v2 — The full map and migration status.
Course Path¶
A compact linear route through the current material. Earlier book chapters remain available as evolving synthesis, while this spine keeps the route aligned with the present repository.
Core Theory¶
The reconstructed foundation first; then the formal and exploratory essays it now constrains.
Simulations¶
Runnable simulations: Boids, Kuramoto, SOC, Lenia, IFS, L-systems, TEO Civilization, Identity Morphospace, and more.
The Measured Core¶
The inverse-reconstruction benchmark: the spine's asymmetry as curves. Equivalence classes counted and collapsed, the search wall, Occam's world-dependent payoff, the optimizer's curse, its cure — and the closed loop. Every claim runs in seconds.
The Papers¶
The Viable Corridor — the flagship formal synthesis (constraint architecture + capability loading; a working draft). Its empirical companion is Quantifying Emergent Utility & Stability in Multi-Agent LLM Ecosystems.
Thinking Space¶
Questions before concepts: encounters, contradictions, and unfinished probes may remain open without promising to become canonical claims.
Sci-Fi Synthesis¶
Narrative stress tests that make abstract theoretical constraints visible in lived, emotional scenarios.
What's Next¶
The framework is drafted, not closed. The next step is external contact, not deployment.
The strongest current artifacts are controlled toy experiments and synthetic models. The next phase should test their boundaries against learned systems, external baselines, and critical review.
Near-term — the real frontier:
- Learned searchers vs. the family-search floor — give LLMs or program synthesizers the same partial traces and query budgets as the exact CA baseline; pre-register consistency, truth recovery, description size, support violations, and cost.
- External review of the Viable Corridor — freeze the paper's conceptual scope while its dynamical assumptions, sufficiency gap, and empirical mapping receive a critical read.
- Rigorous sufficiency — replace single-trajectory evidence with open-set tests and, if possible, construct or bound \(\gamma_c\).
- Separate the real-model questions correctly — live Agentic Identity Suite runs test Mirror/Chord/binding claims; P7/P8 require a distinct real-agent ecology with hard/soft budgets and independently varied constraint architecture.
- Separate selection from drift after cost relocation — v1.13's local group funding nearly removes selection against support and suppresses seeded cheaters, but it does not meet the retention criterion and makes within-group variation nearly cost-neutral. The next discriminating arm should introduce a measurable within-group cost gradient without changing the matched group budget.
Speculative / long-horizon [SPECULATIVE]:
- Hardware prototyping (analog / memristor circuits for physical \(\gamma\)-pin vetoes).
- Protocol-level "Substrate Veto" specifications for decentralized governance systems.
If you want to contribute, the project is open for critical review, preregistered experiments, external baselines, and corrections to any claim that exceeds its evidence.
Quick Start¶
# Clone the repository
git clone https://github.com/frnkptrln/systems-and-intelligence.git
cd systems-and-intelligence
# Run the TEO Civilization Simulation
python simulation-models/alignment-and-veto/teo-civilization/teo_simulation.py
# Run the Black Swan Resilience Simulation
python simulation-models/alignment-and-veto/black-swan-resilience/black_swan_simulation.py
# Serve this book locally
pip install mkdocs-material python-markdown-math
mkdocs serve
Living Document
This repository is a thought experiment developed by Frank Peterlein in collaboration with AI. It is a space to capture, explore, and formalize ideas about emergent intelligence — kept going by nothing grander than curiosity. Feedback, corrections, and discussions are always welcome.