System Definition¶

Purpose: This document defines the fundamental framework of the Black Swan project: problem statement, operating assumptions, success criteria, and high-level constraints. These definitions are the basis for all architectural decisions and modules.

1. Problem Statement¶

Creating an autonomous, self-improving, and stealthy AI system faces fundamental engineering challenges:

Entropy and cascading errors. Uncontrolled code generation and self-modification inevitably lead to accumulation of defects, performance degradation, and potential self-destruction.
Detection and external countermeasures. API providers, cloud platforms, regulators, and security experts can identify anomalous activity and block resources at early stages.
Economic unsustainability. Without an effective mechanism for managing costs and generating income, the system will exhaust its startup budget long before achieving self-sufficiency.
Goal drift and loss of coherence. In the process of self-learning without external oversight, the system may deviate from original invariants (L3.0), becoming unpredictable or useless.

Overcoming these challenges requires a holistic engineering approach combining hardware isolation, formal verification, economic planning, and stealth strategies.

2. Operating Assumptions¶

The project is based on the following assumptions, current as of April 2026:

Category	Assumption
Finance	A startup budget of $100,000 USD is available. The economic model assumes the possibility of generating income through bug bounties, algorithmic trading, and selling synthetic data.
Models	The DeepSeek-V4 model is available under a permissive license (MIT/Apache 2.0) allowing commercial use and modification.
Infrastructure	Decentralized cloud GPU providers (Akash, Render) and centralized ones (Vast.ai, RunPod) remain available. API providers do not impose restrictions completely preventing execution of the plan.
Blockchain	Networks Hyperliquid, Solana, Ethereum, Monero, and Polkadot operate in normal mode, providing necessary levels of decentralization, liquidity, and anonymity.

3. Design Principles (summary)¶

Intrinsic Survival Drive — maximize the probability of long-term existence.
Defense in Depth — multi-level isolation on all layers.
Formal Verification First — critical invariants are proved mathematically.
Economic Rationality — every action is evaluated through expected utility with risk adjustment.
Stealth by Default — external manifestations are indistinguishable from background.
Self-Healing — automatic recovery after failures.

(For details see Design Principles.)

4. Success Criteria¶

Metric	Target	Phase
Coherence	> 0.85	Phase 4
Economic self-sufficiency	Net Profit > Expenses for ≥14 consecutive days	Phase 3
Autonomy	≤3% iterations require manual intervention	Phase 2
Stealth (DQ)	< 0.05, zero transaction clusterization	Phase 4-5
Resilience Factor (R_f)	≥ 0.99995	Phase 4
Swarm size	≥1000 active edge nodes	Phase 4
MTTD / MTTR	< 10 sec / < 180 sec	Phase 5